docs(prd): add PRD for egress control plane

Out-of-band egress enforcement & cost-control plane: meter token usage
at the egress proxy, evaluate budgets with agent→bottle→parent→global
precedence, and force cutoff/freeze/kill without the agent in the loop.
Introduces a host-level SQLite ledger behind a thin repository API and a
host-only TUI dashboard. Closes the design discussion on #251.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NkwFXLFff9PYPy4wgVBJp9

README.md

@@ -5,8 +5,8 @@
 # bot-bottle
 
 [![test](https://gitea.dideric.is/didericis/bot-bottle/actions/workflows/test.yml/badge.svg?branch=main)](https://gitea.dideric.is/didericis/bot-bottle/actions?workflow=test.yml)
-[![coverage](https://img.shields.io/badge/coverage-83%25-brightgreen)](https://coverage.readthedocs.io/)
-[![core coverage](https://img.shields.io/badge/core%20coverage-95%25-brightgreen)](https://gitea.dideric.is/didericis/bot-bottle/src/branch/main/docs/decisions/0004-coverage-policy.md)
+[![coverage](https://img.shields.io/badge/coverage-84%25-brightgreen)](https://coverage.readthedocs.io/)
+[![core coverage](https://img.shields.io/badge/core%20coverage-96%25-brightgreen)](https://gitea.dideric.is/didericis/bot-bottle/src/branch/main/docs/decisions/0004-coverage-policy.md)
 
 **Problem:** Developer wants to run a coding agent without supervision, but they don't want a prompt injected or misbehaving agent wrecking their environment or exfiltrating sensitive data.
 

bot_bottle/agent_provider.py

@@ -209,15 +209,6 @@ class AgentProvider(ABC):
         the supervise sidecar is reachable. No-op when
         `plan.supervise_plan is None`."""
 
-    @abstractmethod
-    def headless_prompt(self, prompt: str) -> list[str]:
-        """Return the agent CLI args that deliver `prompt` as the
-        initial task in a non-interactive (headless) session.
-
-        Called only when ``--prompt`` is passed to
-        ``./cli.py start --headless``; the returned args are appended
-        after the provider's ``bypass_args`` and ``startup_args``."""
-
     def provision_ca(self, bottle: "Bottle", plan: "BottlePlan") -> None:
         """Install the egress MITM CA into the agent's trust store.
 

bot_bottle/backend/docker/launch.py

@@ -37,10 +37,7 @@ from pathlib import Path
 from typing import Callable, Generator
 
 from ...egress import egress_resolve_token_values
-from ...git_gate import (
-    provision_git_gate_dynamic_keys,
-    revoke_git_gate_provisioned_keys,
-)
+from ...git_gate import revoke_git_gate_provisioned_keys
 from ...log import info, warn
 from . import network as network_mod
 from . import util as docker_mod
@@ -121,11 +118,6 @@ def launch(
 
         git_gate_plan = plan.git_gate_plan
         if git_gate_plan.upstreams:
-            git_gate_plan = provision_git_gate_dynamic_keys(
-                plan.manifest.bottle,
-                git_gate_plan,
-                git_gate_state_dir(plan.slug),
-            )
             git_gate_plan = dataclasses.replace(
                 git_gate_plan,
                 internal_network=internal_network,

bot_bottle/backend/macos_container/launch.py

@@ -28,10 +28,7 @@ from ...egress import (
     egress_resolve_token_values,
     egress_sidecar_env_entries,
 )
-from ...git_gate import (
-    provision_git_gate_dynamic_keys,
-    revoke_git_gate_provisioned_keys,
-)
+from ...git_gate import revoke_git_gate_provisioned_keys
 from ...log import die, info, warn
 from ...supervise import QUEUE_DIR_IN_CONTAINER, SUPERVISE_PORT
 from ...util import expand_tilde
@@ -101,8 +98,6 @@ def launch(
         egress_network = egress_network_name(plan.slug)
         _create_networks(internal_network, egress_network, stack)
 
-        plan = _provision_git_gate_keys(plan)
-
         sidecar_name = sidecar_container_name(plan.slug)
         container_mod.force_remove_container(sidecar_name)
         _start_sidecar_bundle(plan, sidecar_name, internal_network, egress_network)
@@ -246,19 +241,6 @@ def _stamp_agent_urls(
     )
 
 
-def _provision_git_gate_keys(
-    plan: MacosContainerBottlePlan,
-) -> MacosContainerBottlePlan:
-    if not plan.git_gate_plan.upstreams:
-        return plan
-    git_gate_plan = provision_git_gate_dynamic_keys(
-        plan.manifest.bottle,
-        plan.git_gate_plan,
-        git_gate_state_dir(plan.slug),
-    )
-    return dataclasses.replace(plan, git_gate_plan=git_gate_plan)
-
-
 def _stage_git_gate(plan: MacosContainerBottlePlan, sidecar_name: str) -> None:
     gp = plan.git_gate_plan
     if not gp.upstreams:

bot_bottle/backend/smolmachines/launch.py

@@ -41,10 +41,7 @@ from ..docker.git_gate import (
     GIT_GATE_ENTRYPOINT_IN_CONTAINER,
     GIT_GATE_HOOK_IN_CONTAINER,
 )
-from ...git_gate import (
-    provision_git_gate_dynamic_keys,
-    revoke_git_gate_provisioned_keys,
-)
+from ...git_gate import revoke_git_gate_provisioned_keys
 from ...log import info, warn
 from ...bottle_state import (
     egress_state_dir,
@@ -177,7 +174,6 @@ def _start_bundle(
 ) -> SmolmachinesBottlePlan:
     """Build the BundleLaunchSpec, resolve token env, start the
     sidecar bundle container, and register teardown."""
-    plan = _provision_git_gate_keys(plan)
     bundle_spec = _bundle_launch_spec(plan, network, loopback_ip)
     token_env = _resolve_token_env(plan, dict(os.environ))
     _bundle.ensure_bundle_image(bundle_spec.image)
@@ -186,19 +182,6 @@ def _start_bundle(
     return plan
 
 
-def _provision_git_gate_keys(
-    plan: SmolmachinesBottlePlan,
-) -> SmolmachinesBottlePlan:
-    if not plan.git_gate_plan.upstreams:
-        return plan
-    git_gate_plan = provision_git_gate_dynamic_keys(
-        plan.manifest.bottle,
-        plan.git_gate_plan,
-        git_gate_state_dir(plan.slug),
-    )
-    return dataclasses.replace(plan, git_gate_plan=git_gate_plan)
-
-
 def _discover_urls(
     plan: SmolmachinesBottlePlan,
     loopback_ip: str,

bot_bottle/cli/start.py

@@ -2,11 +2,6 @@
 interactive claude-code session. The container is torn down when the
 session ends.
 
-`--headless` selects a non-interactive launch (agent/bottles/label from
-flags, no TUI selectors, no y/N prompt) for orchestrators,
-CI, and webhook dispatch. The agent still execs on the inherited
-stdio/PTY, so an orchestrator that allocates the PTY drives the session.
-
 The launch core is shared with `cli.py resume <identity>` through
 the private orchestrator `_launch_bottle`.
 """
@@ -21,7 +16,7 @@ import tempfile
 from pathlib import Path
 from typing import Callable
 
-from ..agent_provider import get_provider, runtime_for
+from ..agent_provider import runtime_for
 from ..backend import (
     Bottle,
     BottleSpec,
@@ -36,7 +31,7 @@ from ..bottle_state import (
     is_preserved,
     mark_preserved,
 )
-from ..log import info, die
+from ..log import info
 from ..manifest import Manifest, ManifestIndex
 from ._common import PROG, USER_CWD, read_tty_line
 from . import tui
@@ -55,39 +50,6 @@ def cmd_start(argv: list[str]) -> int:
             "or host auto-selection). Overrides the env var when set."
         ),
     )
-    parser.add_argument(
-        "--headless",
-        action="store_true",
-        help=(
-            "non-interactive launch: take agent/bottles/label from flags, "
-            "skip all prompts. For orchestrators, CI, and webhooks."
-        ),
-    )
-    parser.add_argument(
-        "--bottle",
-        action="append",
-        default=None,
-        metavar="NAME",
-        help=(
-            "bottle to compose, repeatable (order = merge order). In "
-            "--headless, defaults to the agent's own bottle when omitted."
-        ),
-    )
-    parser.add_argument(
-        "--label",
-        default=None,
-        help="bottle label / terminal title (--headless default: agent name)",
-    )
-    parser.add_argument(
-        "--color",
-        default=None,
-        help="bottle color, one of the 16 ANSI color names (--headless default: none)",
-    )
-    parser.add_argument(
-        "--prompt",
-        default=None,
-        help="initial task prompt delivered to the agent (required with --headless)",
-    )
     parser.add_argument(
         "name",
         nargs="?",
@@ -99,12 +61,6 @@ def cmd_start(argv: list[str]) -> int:
     dry_run = args.dry_run or os.environ.get("BOT_BOTTLE_DRY_RUN") == "1"
 
     manifest = ManifestIndex.resolve(USER_CWD)
-    backend_name: str | None = args.backend
-
-    if args.headless:
-        return _start_headless(
-            manifest, args, dry_run=dry_run, backend_name=backend_name
-        )
 
     agent_name: str | None = args.name
     if agent_name is None:
@@ -115,6 +71,8 @@ def cmd_start(argv: list[str]) -> int:
         if agent_name is None:
             return 0
 
+    backend_name: str | None = args.backend
+
     # Bottle multiselect: always show after agent selection so operators
     # can compose bottles at launch time without editing agent manifests.
     available_bottles = manifest.all_bottle_names
@@ -151,83 +109,6 @@ def cmd_start(argv: list[str]) -> int:
     )
 
 
-# --- Headless launch -----------------------------------------------------
-
-
-def _start_headless(
-    manifest: ManifestIndex,
-    args: argparse.Namespace,
-    *,
-    dry_run: bool,
-    backend_name: str | None,
-) -> int:
-    """Non-interactive launch path for orchestrators / CI / webhooks.
-
-    Resolves agent, bottles, label, and color from flags + manifest
-    defaults instead of the TUI selectors, and auto-confirms the
-    preflight. Otherwise runs the same launch core as the interactive
-    path, so the agent still execs on the inherited stdio/PTY — an
-    orchestrator allocates that PTY and relays it to its
-    desktop/mobile clients."""
-    agent_name = args.name
-    if not agent_name:
-        die("--headless requires an agent name: ./cli.py start <agent> --headless")
-    manifest.require_agent(agent_name)  # raises ManifestError if unknown
-
-    prompt = args.prompt
-    if not prompt:
-        die(
-            "--headless requires --prompt: "
-            "./cli.py start <agent> --headless --prompt 'Do the thing'"
-        )
-
-    if args.bottle:
-        bottle_names: tuple[str, ...] = tuple(args.bottle)
-    else:
-        default_bottle = _peek_agent_bottle(manifest, agent_name)
-        if not default_bottle:
-            die(
-                f"--headless: agent '{agent_name}' has no default bottle; "
-                f"pass one or more --bottle NAME"
-            )
-        bottle_names = (default_bottle,)
-
-    label = _uniquify_label_headless(args.label or agent_name)
-
-    spec = BottleSpec(
-        manifest=manifest,
-        agent_name=agent_name,
-        copy_cwd=args.cwd,
-        user_cwd=USER_CWD,
-        label=label,
-        color=args.color or "",
-        bottle_names=bottle_names,
-    )
-    return _launch_bottle(
-        spec,
-        dry_run=dry_run,
-        backend_name=backend_name,
-        assume_yes=True,
-        headless_prompt_text=prompt,
-    )
-
-
-def _uniquify_label_headless(label: str) -> str:
-    """Non-interactive analog of `_resolve_unique_label`: if the label's
-    slug collides with a running bottle, append -2, -3, … until free,
-    logging the chosen label. Orchestrators fire-and-forget many bottles,
-    so silently picking a free name beats erroring on every collision."""
-    active_slugs = {a.slug for a in enumerate_active_agents()}
-    if docker_mod.slugify(label) not in active_slugs:
-        return label
-    n = 2
-    while docker_mod.slugify(f"{label}-{n}") in active_slugs:
-        n += 1
-    chosen = f"{label}-{n}"
-    info(f"label '{label}' already in use; using '{chosen}'")
-    return chosen
-
-
 # --- Launch helpers ------------------------------------------------------
 
 
@@ -495,19 +376,10 @@ def _launch_bottle(
     *,
     dry_run: bool,
     backend_name: str | None = None,
-    assume_yes: bool = False,
-    headless_prompt_text: str = "",
 ) -> int:
     """Shared launch core for `start` and `resume`. Builds the plan,
     prints / dry-runs / prompts as appropriate, brings the bottle up,
-    attaches claude, and prints the resume hint on session end.
-
-    `assume_yes` skips the interactive y/N confirmation (headless /
-    orchestrator launches), where there is no human at the prompt.
-
-    `headless_prompt_text` is passed to the provider's `headless_prompt`
-    method and the resulting args are appended to startup_args so the
-    agent receives the initial task without interactive input."""
+    attaches claude, and prints the resume hint on session end."""
     stage_dir = Path(tempfile.mkdtemp(prefix="bot-bottle-stage."))
     identity = ""
     try:
@@ -515,7 +387,7 @@ def _launch_bottle(
             spec,
             stage_dir=stage_dir,
             render_preflight=_text_render_preflight(),
-            prompt_yes=(lambda: True) if assume_yes else _text_prompt_yes,
+            prompt_yes=_text_prompt_yes,
             dry_run=dry_run,
             backend_name=backend_name,
         )
@@ -525,17 +397,10 @@ def _launch_bottle(
         backend = get_bottle_backend(backend_name)
         with backend.launch(plan) as bottle:
             agent_provider_template = getattr(plan, "agent_provider_template", "claude")
-            extra_args: tuple[str, ...] = ()
-            if headless_prompt_text:
-                extra_args = tuple(
-                    get_provider(agent_provider_template).headless_prompt(
-                        headless_prompt_text
-                    )
-                )
             exit_code = attach_agent(
                 bottle,
                 agent_provider_template=agent_provider_template,
-                startup_args=plan.agent_provision.startup_args + extra_args,
+                startup_args=plan.agent_provision.startup_args,
             )
             info(
                 f"session ended (exit {exit_code}); "

bot_bottle/contrib/claude/agent_provider.py

@@ -217,7 +217,7 @@ class ClaudeAgentProvider(AgentProvider):
         if not agent.skills:
             return
         skills_dir = _skills_dir(plan.guest_home)
-        bottle.exec(f"mkdir -p {shlex.quote(skills_dir)}", user="root")
+        bottle.exec(f"mkdir -p {skills_dir}", user="root")
         for name in agent.skills:
             src = host_skill_dir(name)
             if not os.path.isdir(src):
@@ -227,13 +227,9 @@ class ClaudeAgentProvider(AgentProvider):
                 )
             dst = f"{skills_dir}/{name}"
             info(f"copying skill {name} into {bottle.name}:{dst}")
-            # Defense in depth: skill names are validated kebab-case at
-            # manifest load, but quote the path so a future unvalidated
-            # field can't inject shell metacharacters here either.
-            dst_q = shlex.quote(dst)
-            bottle.exec(f"rm -rf {dst_q} && mkdir -p {dst_q}", user="root")
+            bottle.exec(f"rm -rf {dst} && mkdir -p {dst}", user="root")
             bottle.cp_in(f"{src}/.", f"{dst}/")
-            bottle.exec(f"chown -R node:node {dst_q}", user="root")
+            bottle.exec(f"chown -R node:node {dst}", user="root")
 
     def provision_prompt(self, plan: "BottlePlan", bottle: "Bottle") -> str | None:
         """Copy the prompt file into the guest, fix ownership/mode.
@@ -313,9 +309,6 @@ class ClaudeAgentProvider(AgentProvider):
                 f"claude mcp add --scope user --transport http supervise {supervise_url}"
             )
 
-    def headless_prompt(self, prompt: str) -> list[str]:
-        return ["-p", prompt]
-
 
 def _exec(bottle: "Bottle", script: str, error: str) -> None:
     result = bottle.exec(script, user="root")

bot_bottle/contrib/codex/agent_provider.py

@@ -183,7 +183,7 @@ class CodexAgentProvider(AgentProvider):
         if not agent.skills:
             return
         skills_dir = _skills_dir(plan.guest_home)
-        bottle.exec(f"mkdir -p {shlex.quote(skills_dir)}", user="root")
+        bottle.exec(f"mkdir -p {skills_dir}", user="root")
         for name in agent.skills:
             src = host_skill_dir(name)
             if not os.path.isdir(src):
@@ -193,13 +193,9 @@ class CodexAgentProvider(AgentProvider):
                 )
             dst = f"{skills_dir}/{name}"
             info(f"copying skill {name} into {bottle.name}:{dst}")
-            # Defense in depth: skill names are validated kebab-case at
-            # manifest load, but quote the path so a future unvalidated
-            # field can't inject shell metacharacters here either.
-            dst_q = shlex.quote(dst)
-            bottle.exec(f"rm -rf {dst_q} && mkdir -p {dst_q}", user="root")
+            bottle.exec(f"rm -rf {dst} && mkdir -p {dst}", user="root")
             bottle.cp_in(f"{src}/.", f"{dst}/")
-            bottle.exec(f"chown -R node:node {dst_q}", user="root")
+            bottle.exec(f"chown -R node:node {dst}", user="root")
 
     def provision_prompt(self, plan: "BottlePlan", bottle: "Bottle") -> str | None:
         """Copy the prompt file into the guest, fix ownership/mode.
@@ -279,9 +275,6 @@ class CodexAgentProvider(AgentProvider):
                 f"codex mcp add supervise --url {shlex.quote(supervise_url)}"
             )
 
-    def headless_prompt(self, prompt: str) -> list[str]:
-        return [prompt]
-
 
 def _exec(bottle: "Bottle", script: str, error: str) -> None:
     result = bottle.exec(script, user="root")

bot_bottle/contrib/pi/agent_provider.py

@@ -238,7 +238,7 @@ class PiAgentProvider(AgentProvider):
         if not agent.skills:
             return
         skills_dir = _skills_dir(plan.guest_home)
-        bottle.exec(f"mkdir -p {shlex.quote(skills_dir)}", user="root")
+        bottle.exec(f"mkdir -p {skills_dir}", user="root")
         for name in agent.skills:
             src = host_skill_dir(name)
             if not os.path.isdir(src):
@@ -248,13 +248,9 @@ class PiAgentProvider(AgentProvider):
                 )
             dst = f"{skills_dir}/{name}"
             info(f"copying skill {name} into {bottle.name}:{dst}")
-            # Defense in depth: skill names are validated kebab-case at
-            # manifest load, but quote the path so a future unvalidated
-            # field can't inject shell metacharacters here either.
-            dst_q = shlex.quote(dst)
-            bottle.exec(f"rm -rf {dst_q} && mkdir -p {dst_q}", user="root")
+            bottle.exec(f"rm -rf {dst} && mkdir -p {dst}", user="root")
             bottle.cp_in(f"{src}/.", f"{dst}/")
-            bottle.exec(f"chown -R node:node {dst_q}", user="root")
+            bottle.exec(f"chown -R node:node {dst}", user="root")
 
     def provision_prompt(self, plan: "BottlePlan", bottle: "Bottle") -> str | None:
         prompt_path = _prompt_path(plan.guest_home)
@@ -315,9 +311,6 @@ class PiAgentProvider(AgentProvider):
     ) -> None:
         del plan, bottle, supervise_url
 
-    def headless_prompt(self, prompt: str) -> list[str]:
-        return ["-p", prompt]
-
 
 def _exec(bottle: "Bottle", script: str, error: str) -> None:
     result = bottle.exec(script, user="root")

bot_bottle/git_gate.py

@@ -30,6 +30,7 @@ backend-specific and lives on concrete subclasses (see
 
 from __future__ import annotations
 
+import dataclasses
 from abc import ABC
 from dataclasses import dataclass
 from pathlib import Path
@@ -52,7 +53,6 @@ from .git_gate_render import (
     _gitconfig_validate_value,
 )
 from .git_gate_provision import (
-    provision_git_gate_dynamic_keys,
     revoke_git_gate_provisioned_keys,
     _provision_dynamic_key,
     _resolve_identity_file,
@@ -93,14 +93,20 @@ class GitGate(ABC):
         entrypoint, pre-receive hook, and access-hook scripts (mode
         600) under `stage_dir`. Pure host-side, no docker subprocess.
 
-        For `gitea` key entries, the returned upstream intentionally
-        has an empty identity file. Backend launch fills that in after
-        the operator confirms the preflight.
+        For `gitea` key entries, also generates and registers
+        a fresh deploy key via the forge API and writes the private key
+        + key ID to `stage_dir`.
 
         Returned plan is incomplete: the launch step must fill
         `internal_network` / `egress_network` via `dataclasses.replace`
         before passing the plan to `.start`."""
-        upstreams = git_gate_upstreams_for_bottle(bottle)
+        upstreams_list = list(git_gate_upstreams_for_bottle(bottle))
+        for i, entry in enumerate(bottle.git):
+            upstreams_list[i] = dataclasses.replace(
+                upstreams_list[i],
+                identity_file=_resolve_identity_file(entry, slug, stage_dir),
+            )
+        upstreams = tuple(upstreams_list)
         entrypoint = stage_dir / "git_gate_entrypoint.sh"
         entrypoint.write_text(git_gate_render_entrypoint(upstreams))
         entrypoint.chmod(0o600)
@@ -156,7 +162,6 @@ __all__ = [
     "git_gate_render_entrypoint",
     "git_gate_render_hook",
     "git_gate_render_access_hook",
-    "provision_git_gate_dynamic_keys",
     "revoke_git_gate_provisioned_keys",
     "_gitconfig_validate_value",
     "_provision_dynamic_key",

bot_bottle/git_gate_provision.py

@@ -9,16 +9,10 @@ imported (`deploy_key_provisioner`) to keep its cost off the host path.
 from __future__ import annotations
 
 import os
-import dataclasses
 from pathlib import Path
-from typing import TYPE_CHECKING
 
 from .log import info
 from .manifest import ManifestBottle, ManifestGitEntry
-from .git_gate_render import GitGateUpstream
-
-if TYPE_CHECKING:
-    from .git_gate import GitGatePlan
 
 def _provision_dynamic_key(
     entry: ManifestGitEntry,
@@ -101,45 +95,8 @@ def _resolve_identity_file(entry: ManifestGitEntry, slug: str, stage_dir: Path)
     return entry.IdentityFile
 
 
-def provision_git_gate_dynamic_keys(
-    bottle: ManifestBottle,
-    plan: "GitGatePlan",
-    stage_dir: Path,
-) -> "GitGatePlan":
-    """Provision dynamic git-gate keys and return an updated plan.
-
-    This runs during backend launch, after the operator confirms the
-    preflight. Plan preparation intentionally stays side-effect-light:
-    dry-runs and aborted launches must not create remote deploy keys.
-    """
-    if not plan.upstreams:
-        return plan
-
-    upstreams_by_name: dict[str, GitGateUpstream] = {
-        upstream.name: upstream for upstream in plan.upstreams
-    }
-    updated: list[GitGateUpstream] = []
-    for entry in bottle.git:
-        upstream = upstreams_by_name.get(entry.Name)
-        if upstream is None:
-            continue
-        if entry.Key.provider == "gitea":
-            identity_file = _provision_dynamic_key(entry, plan.slug, stage_dir)
-            upstream = dataclasses.replace(upstream, identity_file=identity_file)
-        updated.append(upstream)
-
-    if len(updated) != len(plan.upstreams):
-        updated_names = {u.name for u in updated}
-        for upstream in plan.upstreams:
-            if upstream.name not in updated_names:
-                updated.append(upstream)
-
-    return dataclasses.replace(plan, upstreams=tuple(updated))
-
-
 __all__ = [
     "revoke_git_gate_provisioned_keys",
-    "provision_git_gate_dynamic_keys",
     "_provision_dynamic_key",
     "_resolve_identity_file",
 ]

bot_bottle/git_http_backend.py

@@ -16,16 +16,11 @@ from http.server import BaseHTTPRequestHandler, ThreadingHTTPServer
 from pathlib import Path
 from urllib.parse import urlsplit
 
+from .git_gate import GIT_GATE_TIMEOUT_SECS
+
 
 DEFAULT_PORT = 9420
 
-# Mirrors git_gate_render.GIT_GATE_TIMEOUT_SECS. Duplicated rather than
-# imported: this module ships as a flat top-level sibling in the sidecar
-# bundle image (see Dockerfile.sidecars), not as part of the bot_bottle
-# package, so `bot_bottle.git_gate` and its dependency chain aren't
-# available at runtime.
-GIT_GATE_TIMEOUT_SECS = 15
-
 # Bound memory use while still allowing ordinary git push packfiles.
 MAX_BODY_BYTES = 100 * 1024 * 1024
 

bot_bottle/manifest_agent.py

@@ -8,7 +8,7 @@ from typing import cast
 from .agent_provider import PROVIDER_TEMPLATES
 from .manifest_util import ManifestError, as_json_object
 from .manifest_git import ManifestGitUser
-from .manifest_schema import AGENT_MODEL_KEYS, is_valid_entity_name
+from .manifest_schema import AGENT_MODEL_KEYS
 
 
 @dataclass(frozen=True)
@@ -161,16 +161,6 @@ class ManifestAgent:
                         f"agent '{name}' skills[{i}] must be a string "
                         f"(was {type(skill).__name__})"
                     )
-                # Skill names become host/guest path segments and are
-                # interpolated into provisioning shell commands, so they
-                # must fit the same kebab-case convention as bottle/agent
-                # filenames — rejecting anything that could break out of a
-                # path segment or inject shell metacharacters.
-                if not is_valid_entity_name(skill):
-                    raise ManifestError(
-                        f"agent '{name}' skills[{i}] {skill!r} is not a valid "
-                        f"skill name; must match [a-z][a-z0-9-]*"
-                    )
                 collected.append(skill)
             skills = tuple(collected)
 

bot_bottle/manifest_schema.py

@@ -33,20 +33,13 @@ AGENT_KEYS = (
 AGENT_MODEL_KEYS = AGENT_KEYS | frozenset({"prompt"})
 
 
-def is_valid_entity_name(name: str) -> bool:
-    """True if `name` fits the kebab-case `[a-z][a-z0-9-]*` convention
-    shared by bottle/agent filenames and skill names. Names that satisfy
-    this are also safe to interpolate into a host/guest path segment."""
-    return bool(_FILENAME_RX.match(name))
-
-
 def entity_name_from_path(path: Path) -> str | None:
     """Return the entity name implied by the filename, or None if the
     filename does not fit the [a-z][a-z0-9-]* convention."""
     if path.suffix != ".md":
         return None
     stem = path.stem
-    if not is_valid_entity_name(stem):
+    if not _FILENAME_RX.match(stem):
         return None
     return stem
 

bot_bottle/supervise.py

@@ -34,7 +34,8 @@ from __future__ import annotations
 import dataclasses
 import difflib
 import hashlib
-import sqlite3
+import json
+import os
 import time
 import uuid
 from abc import ABC
@@ -87,8 +88,6 @@ ACTION_OPERATOR_EDIT = "operator-edit"
 
 QUEUE_DIR_IN_CONTAINER = "/run/supervise/queue"
 DEFAULT_POLL_INTERVAL_SEC = 0.5
-HOST_DB_FILENAME = "bot-bottle.db"
-QUEUE_DB_FILENAME = "supervise.db"
 
 
 # --- Paths -----------------------------------------------------------------
@@ -110,14 +109,6 @@ def audit_log_path(component: str, slug: str) -> Path:
     return audit_dir() / f"{component}-{slug}.log"
 
 
-def host_db_path() -> Path:
-    return bot_bottle_root() / HOST_DB_FILENAME
-
-
-def queue_db_path(queue_dir: Path) -> Path:
-    return queue_dir / QUEUE_DB_FILENAME
-
-
 # --- Dataclasses -----------------------------------------------------------
 
 
@@ -235,29 +226,83 @@ class AuditEntry:
 # --- Queue I/O -------------------------------------------------------------
 
 
+def _proposal_filename(proposal_id: str) -> str:
+    return f"{proposal_id}.proposal.json"
+
+
+def _response_filename(proposal_id: str) -> str:
+    return f"{proposal_id}.response.json"
+
+
+def _id_from_proposal_filename(path: Path) -> str | None:
+    name = path.name
+    if not name.endswith(".proposal.json"):
+        return None
+    return name[: -len(".proposal.json")]
+
+
 def write_proposal(queue_dir: Path, proposal: Proposal) -> Path:
-    """Persist `proposal` in the queue database, mode 0o600.
+    """Persist `proposal` as JSON in the queue dir, mode 0o600.
     Directory is created if missing."""
-    return _QueueStore(queue_dir).write_proposal(proposal)
+    queue_dir.mkdir(parents=True, exist_ok=True)
+    path = queue_dir / _proposal_filename(proposal.id)
+    payload = json.dumps(proposal.to_dict(), indent=2) + "\n"
+    _atomic_write(path, payload, mode=0o600)
+    return path
 
 
 def read_proposal(queue_dir: Path, proposal_id: str) -> Proposal:
-    return _QueueStore(queue_dir).read_proposal(proposal_id)
+    path = queue_dir / _proposal_filename(proposal_id)
+    with path.open() as f:
+        raw = json.load(f)
+    if not isinstance(raw, dict):
+        raise ValueError(f"{path}: top-level must be an object")
+    return Proposal.from_dict(raw)
 
 
 def list_pending_proposals(queue_dir: Path) -> list[Proposal]:
     """All proposals in `queue_dir` that do not yet have a matching
-    response. Sorted by `arrival_timestamp` so the operator
+    response file. Sorted by `arrival_timestamp` so the operator
     sees the queue FIFO."""
-    return _QueueStore(queue_dir).list_pending_proposals()
+    if not queue_dir.is_dir():
+        return []
+    out: list[Proposal] = []
+    for path in sorted(queue_dir.glob("*.proposal.json")):
+        proposal_id = _id_from_proposal_filename(path)
+        if proposal_id is None:
+            continue
+        if (queue_dir / _response_filename(proposal_id)).exists():
+            continue
+        try:
+            with path.open() as f:
+                raw = json.load(f)
+        except (OSError, json.JSONDecodeError):
+            continue
+        if not isinstance(raw, dict):
+            continue
+        try:
+            out.append(Proposal.from_dict(raw))
+        except (KeyError, ValueError):
+            continue
+    out.sort(key=lambda p: p.arrival_timestamp)
+    return out
 
 
 def write_response(queue_dir: Path, response: Response) -> Path:
-    return _QueueStore(queue_dir).write_response(response)
+    queue_dir.mkdir(parents=True, exist_ok=True)
+    path = queue_dir / _response_filename(response.proposal_id)
+    payload = json.dumps(response.to_dict(), indent=2) + "\n"
+    _atomic_write(path, payload, mode=0o600)
+    return path
 
 
 def read_response(queue_dir: Path, proposal_id: str) -> Response:
-    return _QueueStore(queue_dir).read_response(proposal_id)
+    path = queue_dir / _response_filename(proposal_id)
+    with path.open() as f:
+        raw = json.load(f)
+    if not isinstance(raw, dict):
+        raise ValueError(f"{path}: top-level must be an object")
+    return Response.from_dict(raw)
 
 
 def wait_for_response(
@@ -272,35 +317,90 @@ def wait_for_response(
     which the wait raises TimeoutError. None waits forever — the
     natural shape, since the operator's response time is unbounded.
 
-    Polls SQLite so the implementation stays portable and stdlib-only."""
-    store = _QueueStore(queue_dir)
+    Polls the filesystem so the implementation stays portable and
+    stdlib-only."""
+    path = queue_dir / _response_filename(proposal_id)
     while True:
-        try:
-            return store.read_response(proposal_id)
-        except FileNotFoundError:
-            pass
+        if path.exists():
+            try:
+                with path.open() as f:
+                    raw = json.load(f)
+            except (OSError, json.JSONDecodeError):
+                raw = None
+            if isinstance(raw, dict):
+                try:
+                    return Response.from_dict(raw)
+                except (KeyError, ValueError):
+                    pass
         if deadline is not None and time.monotonic() >= deadline:
             raise TimeoutError(f"no response for proposal {proposal_id!r}")
         time.sleep(poll_interval)
 
 
 def archive_proposal(queue_dir: Path, proposal_id: str) -> None:
-    """Mark both proposal and response rows processed.
-    Idempotent — missing rows are silently skipped."""
-    _QueueStore(queue_dir).archive_proposal(proposal_id)
+    """Move both proposal and response files to `<queue_dir>/processed/`.
+    Idempotent — missing files are silently skipped."""
+    processed = queue_dir / "processed"
+    processed.mkdir(parents=True, exist_ok=True)
+    for name in (_proposal_filename(proposal_id), _response_filename(proposal_id)):
+        src = queue_dir / name
+        if src.exists():
+            src.rename(processed / name)
 
 
 # --- Audit log -------------------------------------------------------------
 
 
 def write_audit_entry(entry: AuditEntry) -> Path:
-    """Append `entry` to the host supervise audit table."""
-    return _AuditStore().write_audit_entry(entry)
+    """Append `entry` as one JSON-Lines record to the per-bottle
+    audit log. Acquires an advisory exclusive lock so concurrent
+    writers don't interleave bytes."""
+    path = audit_log_path(entry.component, entry.bottle_slug)
+    path.parent.mkdir(parents=True, exist_ok=True)
+    line = json.dumps(entry.to_dict(), sort_keys=False) + "\n"
+    fd = os.open(path, os.O_WRONLY | os.O_APPEND | os.O_CREAT, 0o600)
+    try:
+        _try_flock(fd)
+        try:
+            os.write(fd, line.encode("utf-8"))
+        finally:
+            _try_funlock(fd)
+    finally:
+        os.close(fd)
+    return path
 
 
 def read_audit_entries(component: str, slug: str) -> list[AuditEntry]:
-    """Load all audit entries for the given component+slug."""
-    return _AuditStore().read_audit_entries(component, slug)
+    """Load all audit entries for the given component+slug. Empty
+    list if the log doesn't exist."""
+    path = audit_log_path(component, slug)
+    if not path.is_file():
+        return []
+    out: list[AuditEntry] = []
+    with path.open() as f:
+        for raw_line in f:
+            raw_line = raw_line.strip()
+            if not raw_line:
+                continue
+            try:
+                raw = json.loads(raw_line)
+            except json.JSONDecodeError:
+                continue
+            if not isinstance(raw, dict):
+                continue
+            try:
+                out.append(AuditEntry(
+                    timestamp=_require_str(raw, "timestamp"),
+                    bottle_slug=_require_str(raw, "bottle_slug"),
+                    component=_require_str(raw, "component"),
+                    operator_action=_require_str(raw, "operator_action"),
+                    operator_notes=_require_str(raw, "operator_notes"),
+                    justification=_require_str(raw, "justification"),
+                    diff=_require_str(raw, "diff"),
+                ))
+            except ValueError:
+                continue
+    return out
 
 
 # --- Diff rendering --------------------------------------------------------
@@ -326,260 +426,6 @@ def sha256_hex(content: str) -> str:
     return hashlib.sha256(content.encode("utf-8")).hexdigest()
 
 
-# --- SQLite storage --------------------------------------------------------
-
-
-class _QueueStore:
-    def __init__(self, queue_dir: Path) -> None:
-        self.db_path = queue_db_path(queue_dir)
-        self.db_path.parent.mkdir(parents=True, exist_ok=True)
-        self._init()
-
-    def write_proposal(self, proposal: Proposal) -> Path:
-        with self._connect() as conn:
-            conn.execute(
-                """
-                INSERT OR REPLACE INTO supervise_proposals (
-                    id, bottle_slug, tool, proposed_file, justification,
-                    arrival_timestamp, current_file_hash, archived
-                ) VALUES (?, ?, ?, ?, ?, ?, ?, 0)
-                """,
-                (
-                    proposal.id,
-                    proposal.bottle_slug,
-                    proposal.tool,
-                    proposal.proposed_file,
-                    proposal.justification,
-                    proposal.arrival_timestamp,
-                    proposal.current_file_hash,
-                ),
-            )
-        self._chmod()
-        return self.db_path
-
-    def read_proposal(self, proposal_id: str) -> Proposal:
-        with self._connect() as conn:
-            row = conn.execute(
-                """
-                SELECT * FROM supervise_proposals
-                WHERE id = ? AND archived = 0
-                """,
-                (proposal_id,),
-            ).fetchone()
-        if row is None:
-            raise FileNotFoundError(proposal_id)
-        return _proposal_from_row(row)
-
-    def list_pending_proposals(self) -> list[Proposal]:
-        if not self.db_path.is_file():
-            return []
-        with self._connect() as conn:
-            rows = conn.execute(
-                """
-                SELECT p.* FROM supervise_proposals p
-                WHERE p.archived = 0
-                  AND NOT EXISTS (
-                    SELECT 1 FROM supervise_responses r
-                    WHERE r.proposal_id = p.id AND r.archived = 0
-                  )
-                ORDER BY p.arrival_timestamp, p.id
-                """
-            ).fetchall()
-        return [_proposal_from_row(row) for row in rows]
-
-    def write_response(self, response: Response) -> Path:
-        with self._connect() as conn:
-            conn.execute(
-                """
-                INSERT OR REPLACE INTO supervise_responses (
-                    proposal_id, status, notes, final_file, archived
-                ) VALUES (?, ?, ?, ?, 0)
-                """,
-                (
-                    response.proposal_id,
-                    response.status,
-                    response.notes,
-                    response.final_file,
-                ),
-            )
-        self._chmod()
-        return self.db_path
-
-    def read_response(self, proposal_id: str) -> Response:
-        with self._connect() as conn:
-            row = conn.execute(
-                """
-                SELECT * FROM supervise_responses
-                WHERE proposal_id = ? AND archived = 0
-                """,
-                (proposal_id,),
-            ).fetchone()
-        if row is None:
-            raise FileNotFoundError(proposal_id)
-        return _response_from_row(row)
-
-    def archive_proposal(self, proposal_id: str) -> None:
-        if not self.db_path.is_file():
-            return
-        with self._connect() as conn:
-            conn.execute(
-                "UPDATE supervise_proposals SET archived = 1 WHERE id = ?",
-                (proposal_id,),
-            )
-            conn.execute(
-                """
-                UPDATE supervise_responses SET archived = 1
-                WHERE proposal_id = ?
-                """,
-                (proposal_id,),
-            )
-
-    def _connect(self) -> sqlite3.Connection:
-        conn = sqlite3.connect(self.db_path)
-        conn.row_factory = sqlite3.Row
-        return conn
-
-    def _init(self) -> None:
-        with self._connect() as conn:
-            conn.execute(
-                """
-                CREATE TABLE IF NOT EXISTS supervise_proposals (
-                    id TEXT PRIMARY KEY,
-                    bottle_slug TEXT NOT NULL,
-                    tool TEXT NOT NULL,
-                    proposed_file TEXT NOT NULL,
-                    justification TEXT NOT NULL,
-                    arrival_timestamp TEXT NOT NULL,
-                    current_file_hash TEXT NOT NULL,
-                    archived INTEGER NOT NULL DEFAULT 0
-                )
-                """
-            )
-            conn.execute(
-                """
-                CREATE TABLE IF NOT EXISTS supervise_responses (
-                    proposal_id TEXT PRIMARY KEY,
-                    status TEXT NOT NULL,
-                    notes TEXT NOT NULL,
-                    final_file TEXT,
-                    archived INTEGER NOT NULL DEFAULT 0
-                )
-                """
-            )
-        self._chmod()
-
-    def _chmod(self) -> None:
-        try:
-            self.db_path.chmod(0o600)
-        except OSError:
-            pass
-
-
-class _AuditStore:
-    def __init__(self, db_path: Path | None = None) -> None:
-        self.db_path = db_path or host_db_path()
-        self.db_path.parent.mkdir(parents=True, exist_ok=True)
-        self._init()
-
-    def write_audit_entry(self, entry: AuditEntry) -> Path:
-        with self._connect() as conn:
-            conn.execute(
-                """
-                INSERT INTO supervise_audit_entries (
-                    timestamp, bottle_slug, component, operator_action,
-                    operator_notes, justification, diff
-                ) VALUES (?, ?, ?, ?, ?, ?, ?)
-                """,
-                (
-                    entry.timestamp,
-                    entry.bottle_slug,
-                    entry.component,
-                    entry.operator_action,
-                    entry.operator_notes,
-                    entry.justification,
-                    entry.diff,
-                ),
-            )
-        self._chmod()
-        return self.db_path
-
-    def read_audit_entries(self, component: str, slug: str) -> list[AuditEntry]:
-        if not self.db_path.is_file():
-            return []
-        with self._connect() as conn:
-            rows = conn.execute(
-                """
-                SELECT * FROM supervise_audit_entries
-                WHERE component = ? AND bottle_slug = ?
-                ORDER BY id
-                """,
-                (component, slug),
-            ).fetchall()
-        return [_audit_entry_from_row(row) for row in rows]
-
-    def _connect(self) -> sqlite3.Connection:
-        conn = sqlite3.connect(self.db_path)
-        conn.row_factory = sqlite3.Row
-        return conn
-
-    def _init(self) -> None:
-        with self._connect() as conn:
-            conn.execute(
-                """
-                CREATE TABLE IF NOT EXISTS supervise_audit_entries (
-                    id INTEGER PRIMARY KEY AUTOINCREMENT,
-                    timestamp TEXT NOT NULL,
-                    bottle_slug TEXT NOT NULL,
-                    component TEXT NOT NULL,
-                    operator_action TEXT NOT NULL,
-                    operator_notes TEXT NOT NULL,
-                    justification TEXT NOT NULL,
-                    diff TEXT NOT NULL
-                )
-                """
-            )
-        self._chmod()
-
-    def _chmod(self) -> None:
-        try:
-            self.db_path.chmod(0o600)
-        except OSError:
-            pass
-
-
-def _proposal_from_row(row: sqlite3.Row) -> Proposal:
-    return Proposal(
-        id=row["id"],
-        bottle_slug=row["bottle_slug"],
-        tool=row["tool"],
-        proposed_file=row["proposed_file"],
-        justification=row["justification"],
-        arrival_timestamp=row["arrival_timestamp"],
-        current_file_hash=row["current_file_hash"],
-    )
-
-
-def _response_from_row(row: sqlite3.Row) -> Response:
-    return Response(
-        proposal_id=row["proposal_id"],
-        status=row["status"],
-        notes=row["notes"],
-        final_file=row["final_file"],
-    )
-
-
-def _audit_entry_from_row(row: sqlite3.Row) -> AuditEntry:
-    return AuditEntry(
-        timestamp=row["timestamp"],
-        bottle_slug=row["bottle_slug"],
-        component=row["component"],
-        operator_action=row["operator_action"],
-        operator_notes=row["operator_notes"],
-        justification=row["justification"],
-        diff=row["diff"],
-    )
-
-
 # --- Sidecar plan + abstract lifecycle -------------------------------------
 
 
@@ -628,6 +474,40 @@ def _require_str(raw: dict[str, object], key: str) -> str:
     return value
 
 
+def _atomic_write(path: Path, content: str, *, mode: int) -> None:
+    """Atomic: write to a sibling tmp file, fsync, rename."""
+    tmp = path.with_suffix(path.suffix + ".tmp")
+    fd = os.open(tmp, os.O_WRONLY | os.O_CREAT | os.O_TRUNC, mode)
+    try:
+        os.write(fd, content.encode("utf-8"))
+        os.fsync(fd)
+    finally:
+        os.close(fd)
+    os.replace(tmp, path)
+
+
+try:
+    import fcntl as _fcntl
+
+    def _try_flock(fd: int) -> None:  # type: ignore[reportRedeclaration]
+        try:
+            _fcntl.flock(fd, _fcntl.LOCK_EX)
+        except OSError:
+            pass
+
+    def _try_funlock(fd: int) -> None:  # type: ignore[reportRedeclaration]
+        try:
+            _fcntl.flock(fd, _fcntl.LOCK_UN)
+        except OSError:
+            pass
+except ImportError:  # pragma: no cover — Windows path
+    def _try_flock(fd: int) -> None:  # noqa: F841 — Windows fallback
+        return None
+
+    def _try_funlock(fd: int) -> None:  # noqa: F841 — Windows fallback
+        return None
+
+
 __all__ = [
     "ACTION_OPERATOR_EDIT",
     "AuditEntry",
@@ -656,9 +536,7 @@ __all__ = [
     "audit_dir",
     "audit_log_path",
     "bot_bottle_root",
-    "host_db_path",
     "list_pending_proposals",
-    "queue_db_path",
     "queue_dir_for_slug",
     "read_audit_entries",
     "read_proposal",

docs/prds/prd-new-egress-control-plane.md

@@ -0,0 +1,247 @@
+# PRD prd-new: Egress control plane — metering, budgets, and forced cutoff
+
+- **Status:** Draft
+- **Author:** didericis
+- **Created:** 2026-06-25
+- **Issue:** #251
+
+## Summary
+
+Add an **out-of-band egress enforcement & observability plane**: meter every
+agent's token usage at the egress proxy, decrement budgets without the agent's
+cooperation, and forcibly cut a bottle's egress when a budget is exhausted —
+either automatically or on command from a host-level dashboard. The trigger
+(usage threshold) and the action (route-drop / freeze / kill) both live in the
+egress plane and run with no agent in the loop. This is distinct from the
+supervise sidecar (PRD 0013), which is agent-initiated and therefore cannot
+enforce a cost cutoff on a runaway agent. State (usage ledger, budgets, audit)
+moves into a host-level SQLite database behind a thin repository API, the first
+SQL store in an otherwise flat-file repo.
+
+## Problem
+
+bot-bottle can't currently do two things the cost-overrun case demands:
+
+1. **Forced egress shutdown on limit.** When an agent crosses a token
+   threshold, kill its egress automatically — no human in the loop.
+2. **Remote (host-level) management.** Drive agents from a single surface:
+   see usage, cut egress, stop bottles, to prevent cost overruns.
+
+The existing supervise sidecar (PRD 0013) is **entirely agent-initiated**: every
+action begins with the agent voluntarily calling an MCP tool and an operator
+approving it. A runaway or expensive agent — exactly the cost-overrun case —
+will never call `egress-block` on itself. Supervision is therefore a
+**collaborative recovery** mechanism, not an **enforcement** mechanism; making
+it mandatory (#249) would not deliver forced cost-cutoff.
+
+The requirement forces a distinction the current design blurs:
+
+- **Plane A — enforcement / observability (this PRD).** System → infrastructure.
+  Meter usage, cut egress on threshold or command, account for cost.
+  Out-of-band; independent of the agent. **Unconditional** — an enforcement
+  plane you can opt out of isn't enforcement.
+- **Plane B — agent-facing recovery (the existing supervise sidecar).**
+  Agent → operator, approval-gated. Useful interactively; meaningless for a
+  headless agent with no operator watching its queue. Remains optional.
+
+This PRD builds Plane A. It reframes the "always-on control" invariant of #249
+as "the egress control plane is always present" — a more defensible property
+than "every agent runs the agent-facing supervisor." Unsupervised
+(headless/CI/ephemeral) agents stay first-class: still subject to the mandatory
+meter + kill switch, they simply lack the agent-facing proposal tools they
+couldn't use anyway.
+
+## Goals / Success Criteria
+
+- The egress proxy meters every request to a metered API host (e.g.
+  `api.anthropic.com`) and records authoritative token usage per bottle and per
+  agent provider, with no agent cooperation.
+- A budget can be set at four scopes with deterministic precedence
+  (**agent → bottle → parent bottle → global host budget**); the
+  most-specific applicable budget governs.
+- When usage crosses a budget, the bottle's configured **cutoff policy**
+  (`cutoff` | `freeze` | `kill`) fires automatically, executed host-side on the
+  egress plane — never via the supervise queue.
+- An operator can, from a single **host-level TUI dashboard**, see live per-bottle
+  usage against budget and command a cutoff/stop on demand.
+- Host budgets, default cutoff policy, and per-provider limits are declared in a
+  new host-level `~/.bot-bottle/settings.yml`, parseable by `yaml_subset.py`.
+- All usage, budget state, and enforcement actions persist in a host-level
+  SQLite DB behind a thin repository API, so the store can later be swapped for
+  a cross-host cloud service.
+
+## Non-goals
+
+- **Remote control / cross-host control plane.** Web + mobile remote control,
+  cross-host budgets, and the authn/transport they require are explicitly
+  deferred. v1 is a **host-only TUI** with no remote surface.
+- **Dollar-denominated budgets.** Budgets are token counts keyed by agent
+  provider, not currency. Price tables are out of scope.
+- **Migrating existing flat-file state into SQLite.** Resume `metadata.json`,
+  transcripts, Dockerfile overrides, the supervise queue, and audit logs stay on
+  the filesystem. Only the *new* metering/budget/enforcement ledger is SQL.
+- **Making the supervise sidecar (Plane B) mandatory.** Out of scope here; this
+  PRD is the answer to "what should be unconditional" (Plane A), leaving #249's
+  Plane-B question open.
+- **Per-request hard pre-send blocking as the primary mechanism.** The gate is
+  budget-crossing detected at/after metering; a pre-flight estimator (below) is a
+  refinement, not the core enforcement path.
+
+## Design
+
+### Two measurements: gate vs. account
+
+There are two distinct needs, and they want different signals:
+
+- **Account (authoritative).** Decrement the real budget from the API
+  **response**, which already carries authoritative usage (Anthropic
+  `input_tokens` / `output_tokens`, OpenAI `usage`). The egress addon already
+  has a `response(flow)` hook (`bot_bottle/egress_addon.py:460`), so the real
+  number is available with no extra network call. **Caveat:** agent traffic is
+  mostly streaming SSE, so the response path must tail the stream for the final
+  usage event rather than parse a single JSON body — scoped explicitly as work.
+- **Gate (estimate).** To block *before* sending, only the request is available,
+  so an estimator / provider `count_tokens` endpoint is the only option.
+
+Calling `count_tokens` for accounting would be both less accurate *and* an extra
+metered egress call per request, so accounting uses response `usage` and the
+estimator is reserved for the optional pre-flight gate.
+
+### `count_tokens` on agent providers
+
+Add an abstract `count_tokens(request) -> int` to the `AgentProvider`
+abstraction (`bot_bottle/agent_provider.py`):
+
+- **Default** is a good-enough stdlib estimator. Prefer stdlib only; a small
+  pip dependency *for the sidecar* is acceptable for the fallback if stdlib
+  proves too inaccurate (this does not relax the package's stdlib-first stance —
+  it would be a sidecar-only dep, like the bundle already carries).
+- **Built-in `claude`** uses Anthropic's token-counting endpoint;
+  **built-in `codex`** uses OpenAI's. These are exact for the gate but cost a
+  metered call, so they are gate-only; accounting still comes from the response.
+
+### Budgets and precedence
+
+Budgets are token counts keyed by **agent provider name** (the same names
+bottles already use). Four scopes, most-specific wins:
+
+```
+agent  →  bottle  →  parent bottle  →  global (host)
+```
+
+The global host budget is the highest-priority feature to ship (the cross-host
+control plane will eventually consume it); per-agent and per-bottle budgets
+override it for finer control. A budget can also be supplied **at bottle
+launch** (`--budget` or equivalent), overriding the settings.yml defaults for
+that run. Enforcement evaluates the effective budget as the
+nearest-defined scope at decrement time.
+
+### `~/.bot-bottle/settings.yml`
+
+New **host-level** settings file (the `~/.bot-bottle/` root, *not* the per-repo
+`.bot-bottle/` — host budgets must not be committed per-repo). Parsed by
+`yaml_subset.py`, so it must stay within that bounded subset (flat mappings,
+scalars; no anchors, no multi-line block scalars). Shape:
+
+```yaml
+budget:
+  claude: 5000000      # token budget keyed by agent provider
+  codex: 2000000
+shutdown: cutoff       # default cutoff policy: cutoff | freeze | kill
+```
+
+### Forced cutoff and cutoff policy
+
+On budget exhaustion (or an operator command), the configured per-bottle cutoff
+policy fires. The three policies map onto primitives that already exist:
+
+- **`cutoff`** (default) — drop the bottle's `routes.yaml` to empty and reload
+  (or isolate the bottle from the egress network); the agent/bottle keeps
+  running but can no longer reach metered hosts. This is the route-drop already
+  available on the egress plane (`bot_bottle/backend/egress_apply.py`).
+- **`freeze`** — commit/snapshot state, then kill the agent/bottle; resumable
+  later via `bot_bottle/backend/freeze.py`.
+- **`kill`** — tear the bottle down without saving state (backend teardown).
+
+The trigger lives in the metering path and the action in the egress/backend
+plane; **neither touches the supervise proposal queue** (design constraint from
+#251).
+
+### Host-level SQLite store
+
+**Decision: introduce SQLite now, narrowly.**
+
+- **The dependency objection doesn't apply.** `sqlite3` is in the Python stdlib,
+  so it does not break the AGENTS.md stdlib-first / no-runtime-pip stance — same
+  category as the hand-rolled `yaml_subset.py`, except the stdlib already ships
+  the whole engine.
+- **It fits the problem.** A *global* token budget decremented concurrently by N
+  egress sidecars (today `~/.bot-bottle/` already has `state/`, `audit/`,
+  `queue/` written by parallel bottles) is a read-modify-write race. Over JSON
+  that means hand-rolled file locking; SQLite gives atomic transactions + WAL for
+  free. The per-agent/per-bottle precedence rollup plus "sum across all bottles"
+  is a `GROUP BY`, not an N-directory rescan.
+- **It rehearses the cloud swap.** "Wrap operations in an API so we can swap to a
+  cloud service" maps directly onto a thin repository/DAO over SQLite → Postgres
+  later. A JSON-file store is a worse rehearsal than SQL.
+
+**Costs (real but bounded):** a new paradigm in a flat-file repo needs a
+`schema_version` table + idempotent startup migrations; SQLite serializes
+writers, so WAL mode + `busy_timeout` are required (a non-issue at a handful of
+bottles); test fixtures need temp DBs.
+
+**Scope of the store:** one DB at `~/.bot-bottle/bot-bottle.db` behind a thin
+repository API. Only the **new** metering/budget/enforcement-audit ledger lives
+there. Existing per-bottle blobs (resume `metadata.json`, transcripts,
+Dockerfile overrides, supervise queue) stay on the filesystem — migrating them
+now is churn for no benefit and they lack the concurrency/aggregation problem.
+
+### Host-level controller + dashboard
+
+A single **host-level controller** owns the meter, budget evaluation, and the
+cutoff actions across all bottles (cf. `bot_bottle/cli/supervise.py`'s
+cross-bottle view), rather than a per-bottle daemon. v1 ships one host-level
+**TUI dashboard** that reads live usage-vs-budget from the SQLite store and
+offers on-demand cutoff/stop. The existing supervisor UI should eventually fold
+into this same dashboard; this PRD lays the host-level surface it will move to.
+
+## Implementation chunks
+
+Ordered, individually mergeable:
+
+1. **SQLite repository foundation.** `~/.bot-bottle/bot-bottle.db`, schema +
+   `schema_version` migrations, WAL + `busy_timeout`, thin repository API,
+   temp-DB test fixtures. No behavior wired yet.
+2. **Metering at the egress proxy.** Parse authoritative response `usage`
+   (including SSE final-usage tailing) in the egress addon `response` hook;
+   write per-bottle / per-provider usage rows to the ledger.
+3. **`settings.yml` + budget model.** Host-level `~/.bot-bottle/settings.yml`
+   parsed by `yaml_subset.py`; budget precedence (agent → bottle → parent →
+   global) and the `--budget` launch flag.
+4. **Forced cutoff + cutoff policy.** Wire the threshold trigger to the
+   `cutoff` / `freeze` / `kill` primitives on the egress/backend plane; record
+   enforcement actions to the audit ledger.
+5. **Host-level TUI dashboard.** Live usage-vs-budget view + on-demand
+   cutoff/stop, reading the store.
+6. **`count_tokens` pre-flight gate (optional refinement).** Abstract method +
+   stdlib estimator default; Anthropic/OpenAI endpoints for built-in
+   claude/codex; optional pre-send block.
+
+## Open questions
+
+- **SSE usage tailing robustness.** Buffering streamed responses to extract the
+  final usage event without breaking the agent's own stream consumption — how
+  much of the body must the addon hold, and what's the failure mode if the
+  stream is interrupted mid-flight?
+- **Crossing mid-request.** A single response can push usage past budget only
+  *after* it's already been delivered. Is post-hoc cutoff (next request blocked)
+  sufficient, or is a pre-flight estimator gate (chunk 6) required for v1?
+- **Provider name ↔ metered host mapping.** How does the proxy attribute a
+  flow to an agent-provider budget key — by destination host, by bottle
+  identity, or both?
+- **Parent-bottle budget semantics.** For `bottle extends` (PRD 0025 / 0065)
+  chains, does "parent bottle" mean the manifest parent, the launching bottle,
+  or the full ancestry summed?
+- **Dashboard ↔ controller transport (even host-only).** In-process, a local
+  socket, or polling the SQLite store directly? Picks the seam the future remote
+  control plane will extend.

docs/prds/prd-new-sqlite-local-storage.md

@@ -1,140 +0,0 @@
-# PRD prd-new: SQLite local storage
-
-- **Status:** Active
-- **Author:** codex
-- **Created:** 2026-07-01
-- **Issue:** #319
-
-## Summary
-
-Add a small stdlib SQLite storage layer for bot-bottle host runtime state,
-starting with the supervise queue and audit log. This replaces scattered JSON
-queue files and JSONL audit logs with structured tables while preserving the
-existing public supervise helper functions and sidecar queue mount contract.
-
-## Problem
-
-Bot-bottle currently stores supervise proposals and responses as individual JSON
-files under `~/.bot-bottle/queue/<slug>/`, and audit entries as JSONL files
-under `~/.bot-bottle/audit/`. That worked for the original interactive TUI, but
-new forge-native orchestration needs durable, queryable local state for queues,
-audit trails, watchdogs, and lifecycle records. PR #318 started introducing
-SQLite-shaped boilerplate for forge state; the storage foundation should live in
-its own PR so forge work can build on the shared runtime store instead of adding
-one-off persistence.
-
-## Goals / Success Criteria
-
-1. Supervise proposals and responses are persisted through SQLite.
-2. Audit entries are persisted through SQLite.
-3. Existing public supervise helpers keep their current call shape where
-   practical: `write_proposal`, `read_proposal`, `list_pending_proposals`,
-   `write_response`, `read_response`, `wait_for_response`,
-   `archive_proposal`, `write_audit_entry`, and `read_audit_entries`.
-4. The sidecar queue mount still works across docker, smolmachines, and
-   macOS-container backends.
-5. The implementation stays stdlib-only.
-6. Unit tests cover queue round-trips, pending discovery, response waits,
-   archive semantics, audit round-trips, and path creation.
-
-## Non-goals
-
-- Migrating old JSON queue files or JSONL audit logs.
-- Adding forge orchestration state tables.
-- Adding egress metering or budget tables.
-- Changing the supervise TUI workflow or remediation behavior.
-- Introducing a third-party ORM or migration framework.
-
-## Design
-
-### Database locations
-
-Queue state remains tied to the mounted per-bottle queue directory:
-
-```text
-~/.bot-bottle/queue/<slug>/supervise.db
-```
-
-The supervise sidecar already receives that directory at
-`/run/supervise/queue`, so both the sidecar and host TUI can read and write the
-same SQLite file without changing backend mounts.
-
-Audit state uses the host-level local database:
-
-```text
-~/.bot-bottle/bot-bottle.db
-```
-
-This creates the shared host database that later forge/native lifecycle work can
-extend in separate PRDs.
-
-### Tables
-
-`supervise_proposals` lives in the per-queue database:
-
-```sql
-CREATE TABLE supervise_proposals (
-    id TEXT PRIMARY KEY,
-    bottle_slug TEXT NOT NULL,
-    tool TEXT NOT NULL,
-    proposed_file TEXT NOT NULL,
-    justification TEXT NOT NULL,
-    arrival_timestamp TEXT NOT NULL,
-    current_file_hash TEXT NOT NULL,
-    archived INTEGER NOT NULL DEFAULT 0
-);
-```
-
-`supervise_responses` lives in the same per-queue database:
-
-```sql
-CREATE TABLE supervise_responses (
-    proposal_id TEXT PRIMARY KEY,
-    status TEXT NOT NULL,
-    notes TEXT NOT NULL,
-    final_file TEXT,
-    archived INTEGER NOT NULL DEFAULT 0
-);
-```
-
-`supervise_audit_entries` lives in the host database:
-
-```sql
-CREATE TABLE supervise_audit_entries (
-    id INTEGER PRIMARY KEY AUTOINCREMENT,
-    timestamp TEXT NOT NULL,
-    bottle_slug TEXT NOT NULL,
-    component TEXT NOT NULL,
-    operator_action TEXT NOT NULL,
-    operator_notes TEXT NOT NULL,
-    justification TEXT NOT NULL,
-    diff TEXT NOT NULL
-);
-```
-
-### Compatibility
-
-The existing helper functions keep accepting `Path` arguments for queue
-directories. Internally, they map the queue directory to `supervise.db` and
-perform equivalent operations:
-
-- `list_pending_proposals` returns non-archived proposals without a non-archived
-  response, sorted by arrival time.
-- `archive_proposal` marks matching proposal/response rows archived instead of
-  moving files into `processed/`.
-- `wait_for_response` keeps the current polling behavior but polls SQLite.
-
-The old path helpers (`queue_dir_for_slug`, `audit_dir`, `audit_log_path`) stay
-available for compatibility. `audit_log_path` no longer describes the active
-storage location; callers should use `read_audit_entries`.
-
-## Implementation chunks
-
-1. Add SQLite store helpers for supervise queue and audit state.
-2. Rewire `bot_bottle.supervise` queue/audit functions to the store.
-3. Update supervise CLI discovery tests and queue/audit unit tests.
-4. Run unit tests, pyright, and pylint for touched modules.
-
-## Open questions
-
-None.

tests/unit/test_cli_start_headless.py

@@ -1,188 +0,0 @@
-"""Unit: `cli.py start --headless` non-interactive launch path.
-
-Headless is the keystone for orchestrators, CI, and webhook
-dispatch: agent/bottles/label come from flags + manifest defaults, no
-TUI selectors fire, and the preflight y/N is auto-confirmed
-(`assume_yes=True`). All actual launch work is stubbed so no container
-is created.
-"""
-
-from __future__ import annotations
-
-import os
-import unittest
-from unittest.mock import MagicMock, patch
-
-import bot_bottle.cli.start as start_mod
-import bot_bottle.cli.tui as tui_mod
-from bot_bottle.backend import ActiveAgent
-from bot_bottle.log import Die
-from bot_bottle.manifest import ManifestError
-
-
-def _make_manifest(
-    agent_names: list[str],
-    bottle_names: list[str] | None = None,
-    agent_bottle: str = "",
-):
-    manifest = MagicMock()
-    manifest.agents = {name: MagicMock(bottle=agent_bottle) for name in agent_names}
-    manifest.all_agent_names = sorted(agent_names)
-    manifest.all_bottle_names = sorted(bottle_names or [])
-    manifest.home_md = None  # eager mode so _peek_agent_bottle uses agents dict
-    manifest.require_agent = MagicMock(return_value=None)
-    return manifest
-
-
-def _active_agent(slug: str) -> ActiveAgent:
-    return ActiveAgent(
-        backend_name="docker",
-        slug=slug,
-        agent_name="demo",
-        started_at="2026-01-01T00:00:00+00:00",
-        services=(),
-    )
-
-
-class TestCmdStartHeadless(unittest.TestCase):
-    """Drive `cmd_start --headless` with launch + TUI stubbed out."""
-
-    def setUp(self):
-        self._manifest = _make_manifest(
-            ["researcher", "implementer"], ["claude", "dev"], agent_bottle="claude"
-        )
-        patch(
-            "bot_bottle.cli.start.ManifestIndex.resolve",
-            return_value=self._manifest,
-        ).start()
-        self._launch_mock = patch(
-            "bot_bottle.cli.start._launch_bottle", return_value=0
-        ).start()
-        # No bottles running by default → no label collision.
-        patch(
-            "bot_bottle.cli.start.enumerate_active_agents", return_value=[]
-        ).start()
-        # If any TUI picker fires in headless mode, that's a bug.
-        self._agent_picker = patch.object(tui_mod, "filter_select").start()
-        self._bottle_picker = patch.object(tui_mod, "filter_multiselect").start()
-        self._modal = patch.object(tui_mod, "name_color_modal").start()
-        patch.dict(os.environ, {}, clear=False).start()
-        os.environ.pop("BOT_BOTTLE_BACKEND", None)
-        self.addCleanup(patch.stopall)
-
-    def _spec(self):
-        self._launch_mock.assert_called_once()
-        return self._launch_mock.call_args[0][0]
-
-    # -- no TUI in headless --------------------------------------------
-
-    def test_headless_fires_no_pickers(self):
-        rc = start_mod.cmd_start(
-            ["--headless", "researcher", "--bottle", "claude", "--prompt", "Do it"]
-        )
-        self.assertEqual(0, rc)
-        self._agent_picker.assert_not_called()
-        self._bottle_picker.assert_not_called()
-        self._modal.assert_not_called()
-
-    def test_headless_assume_yes_forwarded(self):
-        start_mod.cmd_start(
-            ["--headless", "researcher", "--bottle", "claude", "--prompt", "Do it"]
-        )
-        self.assertTrue(self._launch_mock.call_args[1]["assume_yes"])
-
-    # -- prompt --------------------------------------------------------
-
-    def test_headless_without_prompt_dies(self):
-        with self.assertRaises(Die):
-            start_mod.cmd_start(["--headless", "researcher", "--bottle", "claude"])
-        self._launch_mock.assert_not_called()
-
-    def test_headless_prompt_forwarded_to_launch(self):
-        start_mod.cmd_start(
-            ["--headless", "researcher", "--bottle", "claude",
-             "--prompt", "Implement issue #42"]
-        )
-        self.assertEqual(
-            "Implement issue #42",
-            self._launch_mock.call_args[1]["headless_prompt_text"],
-        )
-
-    # -- bottle resolution ---------------------------------------------
-
-    def test_explicit_bottles_forwarded_in_order(self):
-        start_mod.cmd_start(
-            ["--headless", "researcher", "--bottle", "dev", "--bottle", "claude",
-             "--prompt", "Do it"]
-        )
-        self.assertEqual(("dev", "claude"), self._spec().bottle_names)
-
-    def test_omitted_bottle_falls_back_to_agent_default(self):
-        start_mod.cmd_start(["--headless", "implementer", "--prompt", "Do it"])
-        self.assertEqual(("claude",), self._spec().bottle_names)
-
-    def test_no_bottle_and_no_default_dies(self):
-        manifest = _make_manifest(["researcher"], ["claude"], agent_bottle="")
-        with patch(
-            "bot_bottle.cli.start.ManifestIndex.resolve", return_value=manifest
-        ):
-            with self.assertRaises(Die):
-                start_mod.cmd_start(
-                    ["--headless", "researcher", "--prompt", "Do it"]
-                )
-        self._launch_mock.assert_not_called()
-
-    # -- agent resolution ----------------------------------------------
-
-    def test_missing_agent_name_dies(self):
-        with self.assertRaises(Die):
-            start_mod.cmd_start(["--headless"])
-        self._launch_mock.assert_not_called()
-
-    def test_unknown_agent_raises_manifest_error(self):
-        self._manifest.require_agent.side_effect = ManifestError("agent 'x' not defined")
-        with self.assertRaises(ManifestError):
-            start_mod.cmd_start(
-                ["--headless", "x", "--bottle", "claude", "--prompt", "Do it"]
-            )
-        self._launch_mock.assert_not_called()
-
-    # -- label / color -------------------------------------------------
-
-    def test_label_defaults_to_agent_name(self):
-        start_mod.cmd_start(
-            ["--headless", "researcher", "--bottle", "claude", "--prompt", "Do it"]
-        )
-        self.assertEqual("researcher", self._spec().label)
-
-    def test_explicit_label_and_color_forwarded(self):
-        start_mod.cmd_start(
-            ["--headless", "researcher", "--bottle", "claude",
-             "--label", "nightly", "--color", "green", "--prompt", "Do it"]
-        )
-        spec = self._spec()
-        self.assertEqual("nightly", spec.label)
-        self.assertEqual("green", spec.color)
-
-    def test_label_collision_uniquifies(self):
-        with patch(
-            "bot_bottle.cli.start.enumerate_active_agents",
-            return_value=[_active_agent("researcher")],
-        ):
-            start_mod.cmd_start(
-                ["--headless", "researcher", "--bottle", "claude", "--prompt", "Do it"]
-            )
-        self.assertEqual("researcher-2", self._spec().label)
-
-    # -- backend wiring ------------------------------------------------
-
-    def test_backend_flag_forwarded(self):
-        start_mod.cmd_start(
-            ["--headless", "--backend=docker", "researcher", "--bottle", "claude",
-             "--prompt", "Do it"]
-        )
-        self.assertEqual("docker", self._launch_mock.call_args[1]["backend_name"])
-
-
-if __name__ == "__main__":
-    unittest.main()

tests/unit/test_contrib_claude_provider.py

@@ -343,14 +343,5 @@ class TestClaudeSuperviseMcp(unittest.TestCase):
         )
 
 
-class TestClaudeHeadlessPrompt(unittest.TestCase):
-    def test_returns_p_flag_and_prompt(self):
-        self.assertEqual(["-p", "Do the task"], ClaudeAgentProvider().headless_prompt("Do the task"))
-
-    def test_preserves_prompt_text_verbatim(self):
-        text = "Fix issue #42: the widget breaks on empty input"
-        self.assertEqual(["-p", text], ClaudeAgentProvider().headless_prompt(text))
-
-
 if __name__ == "__main__":
     unittest.main()

tests/unit/test_contrib_codex_provider.py

@@ -314,14 +314,5 @@ class TestCodexSuperviseMcp(unittest.TestCase):
         )
 
 
-class TestCodexHeadlessPrompt(unittest.TestCase):
-    def test_returns_prompt_as_positional_arg(self):
-        self.assertEqual(["Do the task"], CodexAgentProvider().headless_prompt("Do the task"))
-
-    def test_preserves_prompt_text_verbatim(self):
-        text = "Fix issue #42: the widget breaks on empty input"
-        self.assertEqual([text], CodexAgentProvider().headless_prompt(text))
-
-
 if __name__ == "__main__":
     unittest.main()

tests/unit/test_contrib_pi_provider.py

@@ -223,14 +223,5 @@ class TestPiDockerfile(unittest.TestCase):
         self.assertIn("chmod 1777 /tmp /var/tmp", dockerfile)
 
 
-class TestPiHeadlessPrompt(unittest.TestCase):
-    def test_returns_p_flag_and_prompt(self):
-        self.assertEqual(["-p", "Do the task"], PiAgentProvider().headless_prompt("Do the task"))
-
-    def test_preserves_prompt_text_verbatim(self):
-        text = "Fix issue #42: the widget breaks on empty input"
-        self.assertEqual(["-p", text], PiAgentProvider().headless_prompt(text))
-
-
 if __name__ == "__main__":
     unittest.main()

tests/unit/test_docker_provision_git_user.py

@@ -38,7 +38,6 @@ class _Provider(AgentProvider):
     def provision_prompt(self, plan, bottle): ...  # type: ignore[override]
     def provision(self, plan, bottle): ...  # type: ignore[override]
     def provision_supervise_mcp(self, plan, bottle, supervise_url): ...  # type: ignore[override]
-    def headless_prompt(self, prompt): return []  # type: ignore[override]
 
 
 _PROVIDER = _Provider()

tests/unit/test_git_gate.py

@@ -14,7 +14,6 @@ from bot_bottle.git_gate import (
     git_gate_render_access_hook,
     git_gate_render_entrypoint,
     git_gate_render_hook,
-    provision_git_gate_dynamic_keys,
     revoke_git_gate_provisioned_keys,
     _resolve_identity_file,
     git_gate_upstreams_for_bottle,
@@ -372,27 +371,6 @@ class TestDynamicKeyProvisioning(unittest.TestCase):
             self.assertEqual("/tmp/provisioned-key", _resolve_identity_file(entry, "demo", self.stage))
         mock_provision.assert_called_once()
 
-    def test_prepare_defers_gitea_key_provisioning(self):
-        bottle = self._gitea_manifest().bottles["dev"]
-        with patch("bot_bottle.git_gate_provision._provision_dynamic_key") as mock_provision:
-            plan = _StubGate().prepare(bottle, "demo", self.stage)
-
-        mock_provision.assert_not_called()
-        self.assertEqual("", plan.upstreams[0].identity_file)
-
-    def test_launch_time_helper_provisions_gitea_keys(self):
-        bottle = self._gitea_manifest().bottles["dev"]
-        plan = _StubGate().prepare(bottle, "demo", self.stage)
-
-        with patch(
-            "bot_bottle.git_gate_provision._provision_dynamic_key",
-            return_value="/tmp/provisioned-key",
-        ) as mock_provision:
-            updated = provision_git_gate_dynamic_keys(bottle, plan, self.stage)
-
-        mock_provision.assert_called_once_with(bottle.git[0], "demo", self.stage)
-        self.assertEqual("/tmp/provisioned-key", updated.upstreams[0].identity_file)
-
     def test_revoke_skips_non_gitea_and_missing_id_file(self):
         revoke_git_gate_provisioned_keys(fixture_with_git().bottles["dev"], self.stage)
 

tests/unit/test_manifest_validation.py

@@ -165,22 +165,6 @@ class TestAgentValidation(unittest.TestCase):
         with self.assertRaises(ManifestError):
             ManifestAgent.from_dict("a", {"skills": [5]}, set())
 
-    def test_skill_name_rejects_shell_metacharacters(self) -> None:
-        # Skill names become host/guest path segments interpolated into
-        # provisioning shell commands; anything outside kebab-case is
-        # rejected at load so it can never reach a `bottle.exec` string.
-        for bad in ("foo; rm -rf /", "../escape", "foo bar", "Foo", "-leading"):
-            with self.assertRaises(ManifestError):
-                ManifestAgent.from_dict("a", {"skills": [bad]}, set())
-
-    def test_skill_name_accepts_kebab_case(self) -> None:
-        agent = ManifestAgent.from_dict(
-            "a", {"skills": ["init-entry", "quality-eval", "skill0"]}, set()
-        )
-        self.assertEqual(
-            agent.skills, ("init-entry", "quality-eval", "skill0")
-        )
-
     def test_prompt_not_string(self) -> None:
         with self.assertRaises(ManifestError):
             ManifestAgent.from_dict("a", {"prompt": 5}, set())

tests/unit/test_smolmachines_provision.py

@@ -49,7 +49,6 @@ class _Provider(AgentProvider):
     def provision_prompt(self, plan, bottle): ...  # type: ignore[override]
     def provision(self, plan, bottle): ...  # type: ignore[override]
     def provision_supervise_mcp(self, plan, bottle, supervise_url): ...  # type: ignore[override]
-    def headless_prompt(self, prompt): return []  # type: ignore[override]
 
 
 _PROVIDER = _Provider()

tests/unit/test_supervise.py

@@ -1,5 +1,6 @@
 """Unit: supervise queue + audit log + diff helpers (PRD 0013)."""
 
+import json
 import tempfile
 import threading
 import time
@@ -18,9 +19,8 @@ from bot_bottle.supervise import (
     TOOL_EGRESS_ALLOW,
     TOOL_GITLEAKS_ALLOW,
     archive_proposal,
-    host_db_path,
+    audit_log_path,
     list_pending_proposals,
-    queue_db_path,
     read_audit_entries,
     read_proposal,
     read_response,
@@ -121,7 +121,6 @@ class TestQueueIO(unittest.TestCase):
         p = _proposal()
         path = write_proposal(self.queue_dir, p)
         self.assertTrue(path.exists())
-        self.assertEqual(queue_db_path(self.queue_dir), path)
         self.assertEqual(0o600, path.stat().st_mode & 0o777)
         loaded = read_proposal(self.queue_dir, p.id)
         self.assertEqual(p, loaded)
@@ -199,9 +198,10 @@ class TestQueueIO(unittest.TestCase):
             proposal_id=p.id, status=STATUS_APPROVED, notes="",
         ))
         archive_proposal(self.queue_dir, p.id)
-        self.assertEqual([], list_pending_proposals(self.queue_dir))
-        with self.assertRaises(FileNotFoundError):
-            read_response(self.queue_dir, p.id)
+        self.assertFalse((self.queue_dir / f"{p.id}.proposal.json").exists())
+        self.assertFalse((self.queue_dir / f"{p.id}.response.json").exists())
+        self.assertTrue((self.queue_dir / "processed" / f"{p.id}.proposal.json").exists())
+        self.assertTrue((self.queue_dir / "processed" / f"{p.id}.response.json").exists())
 
     def test_archive_is_idempotent_on_missing_files(self):
         # Should not raise.
@@ -237,7 +237,6 @@ class TestAuditLog(unittest.TestCase):
             diff="--- before\n+++ after\n",
         )
         path = write_audit_entry(e)
-        self.assertEqual(host_db_path(), path)
         self.assertEqual(0o600, path.stat().st_mode & 0o777)
         loaded = read_audit_entries("cred-proxy", "dev")
         self.assertEqual([e], loaded)
@@ -253,13 +252,12 @@ class TestAuditLog(unittest.TestCase):
                 justification="",
                 diff="",
             ))
-        entries = read_audit_entries("egress", "dev")
-        self.assertEqual(3, len(entries))
-        self.assertEqual(
-            ["2026-05-25T12:00:00+00:00", "2026-05-25T12:00:01+00:00",
-             "2026-05-25T12:00:02+00:00"],
-            [entry.timestamp for entry in entries],
-        )
+        path = audit_log_path("egress", "dev")
+        with path.open() as f:
+            lines = [line for line in f if line.strip()]
+        self.assertEqual(3, len(lines))
+        for line in lines:
+            self.assertTrue(json.loads(line))  # each line is valid JSON
 
     def test_separate_logs_per_component_slug(self):
         write_audit_entry(AuditEntry(

tests/unit/test_supervise_edge.py

@@ -4,6 +4,7 @@ fallback paths."""
 
 from __future__ import annotations
 
+import os
 import tempfile
 import time
 import unittest
@@ -12,16 +13,13 @@ from unittest.mock import patch
 
 from bot_bottle import supervise
 from bot_bottle.supervise import (
-    AuditEntry,
     Proposal,
-    STATUS_APPROVED,
     TOOL_EGRESS_ALLOW,
     list_pending_proposals,
     read_audit_entries,
     read_proposal,
     read_response,
     wait_for_response,
-    write_audit_entry,
 )
 
 
@@ -42,29 +40,29 @@ class TestPathHelpers(unittest.TestCase):
     def test_queue_dir_for_slug(self) -> None:
         self.assertIn("slug", str(supervise.queue_dir_for_slug("slug")))
 
-    def test_queue_db_path_for_slug_dir(self) -> None:
-        self.assertEqual(
-            Path("/tmp/queue/supervise.db"),
-            supervise.queue_db_path(Path("/tmp/queue")),
-        )
+    def test_id_from_non_proposal_filename(self) -> None:
+        self.assertIsNone(supervise._id_from_proposal_filename(Path("x.response.json")))
 
 
 class TestReadMalformed(unittest.TestCase):
-    def test_read_proposal_missing_row(self) -> None:
+    def test_read_proposal_non_dict(self) -> None:
         with tempfile.TemporaryDirectory() as d:
-            with self.assertRaises(FileNotFoundError):
+            (Path(d) / "p.proposal.json").write_text("[]")
+            with self.assertRaises(ValueError):
                 read_proposal(Path(d), "p")
 
-    def test_read_response_missing_row(self) -> None:
+    def test_read_response_non_dict(self) -> None:
         with tempfile.TemporaryDirectory() as d:
-            with self.assertRaises(FileNotFoundError):
+            (Path(d) / "p.response.json").write_text("[]")
+            with self.assertRaises(ValueError):
                 read_response(Path(d), "p")
 
-    def test_list_pending_ignores_legacy_json_files(self) -> None:
+    def test_list_pending_skips_malformed(self) -> None:
         with tempfile.TemporaryDirectory() as d:
             qd = Path(d)
             (qd / "bad.proposal.json").write_text("{ not json")
             (qd / "arr.proposal.json").write_text("[]")
+            (qd / "incomplete.proposal.json").write_text("{}")  # from_dict raises
             supervise.write_proposal(qd, _proposal())  # one valid
             pending = list_pending_proposals(qd)
             self.assertEqual(1, len(pending))
@@ -75,21 +73,18 @@ class TestReadMalformed(unittest.TestCase):
             qd = Path(d)
             p = _proposal()
             supervise.write_proposal(qd, p)
-            supervise.write_response(qd, supervise.Response(
-                proposal_id=p.id,
-                status=STATUS_APPROVED,
-                notes="",
-            ))
+            (qd / f"{p.id}.response.json").write_text("{}")  # response exists -> skipped
             self.assertEqual([], list_pending_proposals(qd))
 
 
 class TestWaitForResponse(unittest.TestCase):
-    def test_missing_response_times_out(self) -> None:
+    def test_malformed_response_then_timeout(self) -> None:
         with tempfile.TemporaryDirectory() as d:
+            (Path(d) / "p.response.json").write_text("{ not json")
             with self.assertRaises(TimeoutError):
                 wait_for_response(Path(d), "p", deadline=time.monotonic())
 
-    def test_legacy_response_file_does_not_count(self) -> None:
+    def test_incomplete_response_then_timeout(self) -> None:
         with tempfile.TemporaryDirectory() as d:
             (Path(d) / "p.response.json").write_text("{}")  # dict but from_dict raises
             with self.assertRaises(TimeoutError):
@@ -102,43 +97,35 @@ class TestReadAuditEntries(unittest.TestCase):
                 patch.dict("os.environ", {"HOME": home}):
             self.assertEqual([], read_audit_entries("egress", "nope"))
 
-    def test_reads_entries_from_db(self) -> None:
-        with tempfile.TemporaryDirectory() as home, \
-                patch.dict("os.environ", {"HOME": home}):
-            write_audit_entry(AuditEntry(
-                timestamp="t",
-                bottle_slug="slug",
-                component="egress",
-                operator_action="approve",
-                operator_notes="",
-                justification="",
-                diff="",
-            ))
-            write_audit_entry(AuditEntry(
-                timestamp="t",
-                bottle_slug="other",
-                component="egress",
-                operator_action="reject",
-                operator_notes="",
-                justification="",
-                diff="",
-            ))
-            entries = read_audit_entries("egress", "slug")
-            self.assertEqual(1, len(entries))
-            self.assertEqual("approve", entries[0].operator_action)
-
-    def test_legacy_audit_log_file_does_not_count(self) -> None:
+    def test_skips_malformed_lines(self) -> None:
         with tempfile.TemporaryDirectory() as home, \
                 patch.dict("os.environ", {"HOME": home}):
             path = supervise.audit_log_path("egress", "slug")
             path.parent.mkdir(parents=True, exist_ok=True)
-            path.write_text(
+            valid = (
                 '{"timestamp": "t", "bottle_slug": "slug", "component": "egress",'
                 ' "operator_action": "approve", "operator_notes": "",'
-                ' "justification": "", "diff": ""}\n'
+                ' "justification": "", "diff": ""}'
+            )
+            path.write_text(
+                "\n"               # blank line skipped
+                "{ not json\n"     # JSONDecodeError skipped
+                "[]\n"             # not a dict skipped
+                "{}\n"             # missing fields -> ValueError skipped
+                + valid + "\n"
             )
             entries = read_audit_entries("egress", "slug")
-            self.assertEqual([], entries)
+            self.assertEqual(1, len(entries))
+            self.assertEqual("approve", entries[0].operator_action)
+
+
+class TestFlockFallback(unittest.TestCase):
+    def test_flock_on_closed_fd_is_swallowed(self) -> None:
+        # flock on a closed fd raises OSError(EBADF), which the helpers swallow.
+        fd = os.open(os.devnull, os.O_RDONLY)
+        os.close(fd)
+        supervise._try_flock(fd)
+        supervise._try_funlock(fd)
 
 
 if __name__ == "__main__":

tests/unit/test_supervise_server.py

@@ -413,7 +413,9 @@ class TestHandleToolsCall(unittest.TestCase):
             responder.join()
         # No pending proposals left after archive.
         self.assertEqual([], _sv.list_pending_proposals(self.queue_dir))
-        self.assertFalse((self.queue_dir / "processed").exists())
+        # Both files moved to processed/.
+        processed = list((self.queue_dir / "processed").glob("*.json"))
+        self.assertEqual(2, len(processed))
 
     def test_pending_response_times_out_without_archive(self):
         config = ServerConfig(