Files
bot-bottle/bot_bottle/control_auth.py
T
didericis-claude 1db2a9eb67
test / integration-docker (pull_request) Successful in 17s
tracker-policy-pr / check-pr (pull_request) Successful in 16s
test / unit (pull_request) Successful in 39s
lint / lint (push) Successful in 56s
test / integration-firecracker (pull_request) Successful in 3m21s
test / coverage (pull_request) Successful in 19s
test / publish-infra (pull_request) Has been skipped
feat(control-plane): role-scoped signed tokens so the gateway can't drive operator routes
Review follow-up on #469: the data plane held the same control-plane secret
that authorizes every route, so a compromised egress/git-gate could queue a
supervise proposal AND approve it (or rewrite policy, read injected tokens) —
the (source_ip, identity_token) checks attribute the *bottle*, not the caller.

Replace the single shared bearer secret with role-scoped, HMAC-signed tokens
(compact HS256 JWTs, stdlib-only — no new dependency):

  * new `control_auth` mints/verifies `{role}` tokens; roles are `gateway`
    (data plane) and `cli` (host operator/launcher).
  * the orchestrator holds only the signing *key* and verifies; `dispatch`
    gates each route by role — `gateway` reaches /resolve + /supervise/
    {propose,poll}, everything else is `cli`-only (401 unauthenticated,
    403 wrong role).
  * the gateway is handed a pre-minted `gateway` token it cannot rewrite into
    `cli`; the host CLI mints its own `cli` token from the host key.
  * `gateway_init` scopes the signing key to the orchestrator process and the
    gateway token to the data-plane daemons, so even in the combined infra
    container a compromised data-plane daemon never sees the key.

Launchers (docker gateway + infra, macOS infra) inject the minted token(s);
Firecracker stays open behind its nft boundary. Open mode (no key) still grants
full `cli` access — the fail-visible fallback for tests.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-24 05:22:56 +00:00

101 lines
3.9 KiB
Python

"""Role-scoped control-plane credentials (issue #469 review follow-up).
The control plane no longer trusts a single shared bearer secret for every
route. Instead the orchestrator holds a *signing key* and issues short,
HMAC-signed tokens (compact HS256 JWTs) that embed a **role** naming the kind
of caller:
* ``gateway`` — the data plane (egress / git-gate / supervise). Restricted to
the agent-facing routes it actually needs (``/resolve``,
``/supervise/propose``, ``/supervise/poll``).
* ``cli`` — the host operator / launcher. Full access to the mutating and
operator routes (launch/teardown, policy, ``/supervise/respond``, …).
Only the orchestrator (and the host CLI, which shares the host trust domain)
holds the signing key; the gateway is handed a pre-minted ``gateway`` token it
cannot rewrite into a ``cli`` token. So a compromised data-plane process can no
longer approve its own supervise proposals or drive operator routes — it can
only present the ``gateway`` role it was issued (control_plane rejects it on
operator routes with 403).
Stdlib-only (HMAC-SHA256 over a JSON payload); no JWT dependency — the project
carries no runtime pip deps. Tokens are **signed, not encrypted** (the role is
not a secret) and **long-lived** (parity with the static token they replace;
the security win is the unforgeable role claim, not rotation).
"""
from __future__ import annotations
import base64
import binascii
import hashlib
import hmac
import json
ROLE_GATEWAY = "gateway"
ROLE_CLI = "cli"
ROLES: frozenset[str] = frozenset({ROLE_GATEWAY, ROLE_CLI})
_ALG = "HS256"
def _b64url_encode(raw: bytes) -> str:
return base64.urlsafe_b64encode(raw).rstrip(b"=").decode("ascii")
def _b64url_decode(text: str) -> bytes:
padded = text + "=" * (-len(text) % 4)
return base64.urlsafe_b64decode(padded.encode("ascii"))
def _sign(secret: str, signing_input: str) -> str:
mac = hmac.new(secret.encode("utf-8"), signing_input.encode("ascii"), hashlib.sha256)
return _b64url_encode(mac.digest())
# The fixed, canonical JOSE header — the same for every token we mint.
_HEADER_SEGMENT = _b64url_encode(
json.dumps({"alg": _ALG, "typ": "JWT"}, separators=(",", ":")).encode("utf-8")
)
def mint(role: str, secret: str) -> str:
"""A compact HS256 token asserting `role`, signed with `secret`.
Raises ValueError for an unknown role (mint only what the control plane will
accept) or an empty signing key (an unsigned credential is never valid)."""
if role not in ROLES:
raise ValueError(f"unknown control-plane role {role!r}")
if not secret:
raise ValueError("cannot mint a control-plane token without a signing key")
payload = _b64url_encode(json.dumps({"role": role}, separators=(",", ":")).encode("utf-8"))
signing_input = f"{_HEADER_SEGMENT}.{payload}"
return f"{signing_input}.{_sign(secret, signing_input)}"
def verify(token: str, secret: str) -> str | None:
"""The role a valid `token` carries, or None if it is malformed, wrongly
signed, or names an unknown role. Constant-time signature check; rejects any
header whose alg isn't HS256 (no alg-confusion / `none`)."""
if not token or not secret:
return None
parts = token.split(".")
if len(parts) != 3:
return None
header_b64, payload_b64, sig = parts
expected = _sign(secret, f"{header_b64}.{payload_b64}")
if not hmac.compare_digest(sig, expected):
return None
try:
header = json.loads(_b64url_decode(header_b64))
payload = json.loads(_b64url_decode(payload_b64))
except (ValueError, binascii.Error):
return None
if not isinstance(header, dict) or header.get("alg") != _ALG:
return None
role = payload.get("role") if isinstance(payload, dict) else None
return role if isinstance(role, str) and role in ROLES else None
__all__ = ["ROLE_GATEWAY", "ROLE_CLI", "ROLES", "mint", "verify"]