Mirrors DockerGitGate: build the image, docker create on the internal
network with --network-alias cred-proxy, docker cp the routes.json
into /run/cred-proxy/, attach the egress network, docker start. stop()
is idempotent.
Token values flow host env -> subprocess env -> sidecar env via
docker create -e NAME (no =VALUE on argv). The resolver fails early
with a clear pointer at the missing host env var name if any TokenRef
is unset.
Helpers (cred_proxy_container_name, cred_proxy_url) are agent-side
stable: the URL uses the network alias, not the slugged container
name, so the provisioner can write a fixed http://cred-proxy:9099/
URL regardless of which bottle is running.
Stdlib-only Python proxy: reads /run/cred-proxy/routes.json on boot,
listens on 0.0.0.0:9099, strips inbound Authorization, injects the
configured header (Bearer or token) using the route's token_env env
var, forwards over HTTPS to the upstream, and streams the response
back chunk-by-chunk (SSE-safe).
Hop-by-hop headers are stripped per RFC 7230, including anything
listed in `Connection:`. Content-Length is dropped so http.client
recomputes it on the upstream leg. Tokens never reach routes.json —
they arrive via the container's environ.
Dockerfile.cred-proxy builds on python:3.13-alpine pinned by digest;
mkdir /run/cred-proxy is baked in so docker cp can drop the route
table at start time. No pip install layer.
Smoke-tested: container boots, logs listen line, returns 404 for
unmatched paths. Full request/response cycle covered by the
integration tests in a follow-up commit.
Lifts bottle.tokens into a per-route CredProxyUpstream table, renders a
mode-600 routes.json that carries no token values or host env-var
names, and derives the {token_env: TokenRef} map the launch step will
use to forward host env values into the sidecar's environ.
Shape mirrors GitGate/PipelockProxy: abstract base does the host-side
prepare; start/stop is backend-specific. No backend wiring yet.
TokenEntry carries Kind (anthropic / github / gitea / npm), TokenRef
(name of host env var the CLI resolves at launch), and an optional Url
(required for gitea, fixed for the other kinds). Validation rejects
unknown kinds, duplicate non-gitea entries, duplicate gitea Urls, and
overlap with bottle.git hosts (where git-gate is already brokering).
No wiring yet — the field exists on Bottle but cred-proxy is the next
step. Adds tests/unit/test_manifest_tokens.py.
Make the cred-proxy a per-bottle sidecar container on the bottle's
internal docker network instead of a root-owned process inside the
agent container. The boundary becomes container namespace
separation, matching pipelock and git-gate. Update summary,
problem, goals, in-scope, architecture diagram, components,
existing code touched, external deps, and open questions; add a
"Considered alternatives" section recording the rejected
in-container shape.
Per-bottle reverse proxy that holds API tokens (Anthropic OAuth,
GitHub PAT, Gitea PAT, npm) in a root-owned process; agent gets
only URLs in its environ. AWS / SigV4 explicitly out of scope.
Squashes the demo-build arc: initial GIF + scripts, refactor to drive
recording through real cli.py, theme/timing tweaks, and the switch to
prompt-driven probes.
- README architecture diagram drops the socat/ssh image box and
the agent's ~/.ssh/config; the prose-bullets section drops the
ssh image; the manifest example swaps `ssh:` for `git:` so
someone copy-pasting it picks up the new shape.
- claude-bottle.example.json: `default` bottle's `"ssh": []` is
gone (now just an empty bottle); the gitea-dev example already
uses `git:` since the ExtraHosts work.
- PRD 0007 carries a "Superseded by PRD 0009" header at the top
with a one-paragraph block explaining why; the file stays so
the rationale of the prior design is still in-tree.
- git_gate.py: drop the now-stale shadow-route mention from a
docstring (the validator went away in the manifest layer).
- Delete tests/unit/test_ssh_gate.py and the fixture_with_ssh helpers.
- test_pipelock_yaml: drop the ssh-leak guard (structurally
impossible now); the remaining tests switch to fixture_minimal.
- test_pipelock_allowlist: rewrite the union/dedup test to
exercise an egress.allowlist that duplicates a baked default
(the property the ssh-leak assertion was hitching onto).
- test_manifest_git: shadow-route assertion becomes a legacy-ssh-
dies-with-hint assertion, since bottle.ssh is now parse-fail.
- test_orphan_cleanup: drop the SSHGate.stop idempotency check;
pipelock equivalent stays.
- test_dry_run_plan: drop assertions on the removed ssh_hosts /
ssh_gate keys.
52 unit tests pass.
Delete claude_bottle/ssh_gate.py, the DockerSSHGate sidecar,
and the provision_ssh provisioner (~/.ssh/config + ssh-agent
wiring). Unwire the gate from the abstract BottleBackend
(provision orchestration drops the ssh step,
_validate_ssh_entries goes away) and from the Docker backend
(prepare/launch lose the `gate` kwarg, bottle_plan drops the
gate_plan field, dry-run JSON drops the ssh_hosts / ssh_gate
keys, y/N preflight drops the ssh-hosts block). cli/info now
prints declared git remotes instead of ssh hosts. pipelock's
docstring picks up the git-gate framing now that there's no
PRD-0007 boundary to call out.
BREAKING (dry-run JSON): the `ssh_hosts` and `ssh_gate` keys
are gone from `start --dry-run --format=json`. Consumers should
read `git_remotes` / `git_gate` instead.
Drop the SshEntry dataclass, the Bottle.ssh field, the shadow-
route validator, and the SSH-only _opt_port helper. A legacy
bottle.ssh key now parse-fails with a one-line hint pointing at
bottle.git (PRD 0008), which is the replacement.
BREAKING: manifests carrying bottle.ssh will not load. Migration
is per-entry: drop the ssh entry, add a git entry with a Name +
full Upstream URL + IdentityFile.
ssh-gate was built for non-git SSH (PRD 0007), but every
upstream currently declared in any bottle is a git remote, and
those now flow through git-gate (PRD 0008) with credential
isolation, gitleaks scanning, and `insteadOf` URL rewrites.
ssh-gate is left doing L4 forwarding with no gating value over
git-gate's path; carrying it means a redundant sidecar lifecycle,
a shadow-route validator between bottle.ssh and bottle.git, and
a third place to keep an SSH identity in sync.
Goal is straightforward deletion: bottle.ssh becomes a parse
error pointing at bottle.git, the SshEntry / SSHGate / socat
provisioner / pipelock allowlist branch all go away, and PRD
0007 carries a "Superseded by PRD 0009" header so the rationale
of the prior design stays in the tree.
Consolidates oauth-token-exposure-to-claude.md and
tea-token-isolation-via-proxy.md into agent-credential-proxy-landscape.md,
adding a May-2026 survey of existing tools (Docker AI Sandboxes,
Cloudflare Sandbox Auth, Infisical Agent Vault, nono, Aembit, LiteLLM
CVE-2026-42208, Portkey, Helicone, etc.) and a build-vs-adopt verdict.
Adds secret-minimization-over-dlp.md explaining why pipelock's body
DLP and gitleaks's pre-receive scan cannot stop encoding/splitting
exfil, and why moving credentials out of the bottle (the git-gate
pattern, generalized) is the only robust answer.
Updates git-secret-scanning-hardening.md's reference to point at
the new consolidated landscape doc.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
- example manifest swaps the gitea-dev bottle from ssh: to git:
and shows ExtraHosts pinning gitea.dideric.is to its Tailscale IP
- README's git-gate paragraph names the field and the case it
solves (upstream resolvable on the host but not from the gate
container's default DNS)
- PRD 0008's manifest-field bullet mentions the field for parity
GitGateUpstream carries each entry's extra_hosts; a new
git_gate_aggregate_extra_hosts() merges them into one map for the
gate container's /etc/hosts. Same host -> same IP is harmless
duplication; same host -> different IPs is a manifest bug
(/etc/hosts is per-container, not per-upstream) and dies with
the conflicting upstream names.
DockerGitGate.start passes one --add-host host:ip per merged
entry on docker create. Empty map (the default) emits no flags
and is a no-op for bottles that don't need DNS overrides.
Optional `ExtraHosts: { hostname: ip }` map per git entry. The
docker backend will surface these to the gate sidecar via
--add-host so the gate can resolve upstreams whose default
container DNS doesn't point at the reachable IP (e.g.
Tailscale-only hosts with a public DNS A record pointed
elsewhere). The agent-side insteadOf rewrite still keys off
the original hostname, so the manifest's Upstream URL stays
human-readable.
Architecture diagram + bullet now reflect that the gate fronts
every git operation, not just push: pre-receive gitleaks-gates
the push path; an access-hook refreshes from upstream before each
upload-pack so fetch / clone / pull / ls-remote see whatever the
upstream has at that moment (fail-closed if unreachable).
A pair of integration tests against a real sshd-based "upstream"
sibling container that prove every operation through the gate is
observably equivalent to the same operation against the upstream:
- test_clone_and_refetch_reflect_upstream: clone via gate
returns the upstream's current commit; an out-of-band commit
on the upstream shows up via the gate on the next ls-remote.
- test_push_through_gate_lands_on_upstream: a clean push routed
through the gate lands on the upstream's bare repo.
The upstream container is a tiny inline-built alpine image with
openssh-server, a `git` user (passwd -u so sshd doesn't reject
the locked account), and a baked bare repo seeded with one
commit. Host keys are baked in at build so the test can pin
KnownHostKey on the manifest entry before the container starts.
While wiring this up the access-hook gained a one-shot HEAD
sync: `git init --bare` defaults HEAD to refs/heads/master, and
upstreams that use main would leave the bare repo's HEAD
unresolvable — clones came through but the working tree was
empty. The hook now does a `rev-parse --verify HEAD` check
after the first fetch and runs `ls-remote --symref` to repoint
HEAD if it doesn't resolve. One extra round-trip on first
fetch only.
The agent's ~/.gitconfig now uses insteadOf (not pushInsteadOf),
so every git operation against a declared upstream — push, fetch,
clone, pull, ls-remote — routes through the gate. Matches the
gate's now-bidirectional design: fetch is mirrored via the
access-hook, push is gated via gitleaks.
The gate is now a transparent mirror, not push-only. Per-repo
init now runs `git remote add --mirror=fetch origin <url>` so a
later `git fetch origin` mirrors the upstream's full ref graph at
canonical paths. The pre-receive hook forwards accepted refs via
`git push origin` (renamed from upstream).
New: an access-hook script wired via `git daemon --access-hook`
runs `git fetch origin --prune` against the real upstream before
every upload-pack request (clone, fetch, pull, ls-remote). On
upstream error the hook exits non-zero — the agent's fetch fails
rather than the gate serving stale data.
The pre-existing smoke test (ls-remote against unreachable
upstream returns refs) had to invert: under the bidirectional
design any ls-remote success is necessarily a success against
the upstream, so the unreachable-upstream case now correctly
fails closed.
The gate now fronts every git operation, not just push. Fetch
(clone, pull, ls-remote) is mirrored via git daemon's
--access-hook running 'git fetch origin --prune' against the
real upstream before each upload-pack; fail-closed if upstream
is unreachable so the agent never serves stale data.
Push path is unchanged in concept (gitleaks gate → forward) but
the hook now pushes to 'origin' rather than 'upstream', matching
the remote name the entrypoint configures.
Bumps the sidecar count from two to up to three; the diagram and
bullet list now cover the git-gate alongside pipelock and ssh-gate,
including the ~/.gitconfig pushInsteadOf wiring that fires the
agent's git push through the gate.
Two integration tests against a real Docker daemon:
- test_ls_remote_succeeds_against_fresh_gate: a freshly-started
gate has its empty bare repo exported via git daemon; ls-remote
from a sibling container on the internal network returns no
refs and exits 0.
- test_push_with_secret_is_rejected: the PRD 0008 success
criterion — a push containing an AKIA-shaped synthetic that
trips gitleaks's aws-access-token rule is rejected by the
pre-receive hook with a non-zero exit on the client and a
gitleaks rejection in the response.
Dockerfile.git-gate switches base to zricethezav/gitleaks (alpine
3.22 + gitleaks v8.30.1, pinned by digest) since gitleaks isn't
packaged for alpine, and adds git-daemon (the sub-package the
listener needs; the core git binary in the base doesn't include
the daemon).
DockerBottleBackend now instantiates a DockerGitGate alongside
DockerPipelockProxy and DockerSSHGate; the prepare step lifts
bottle.git into a GitGatePlan stored on DockerBottlePlan, and
launch starts/stops the sidecar in the same ExitStack as the
other two (only when bottle.git is non-empty).
bottle_plan.print now surfaces git remotes and per-upstream gate
forwards in the y/N preflight; to_dict adds git_remotes and
git_gate keys to the dry-run JSON payload for CLI consumers.
PRD: docs/prds/0008-git-gate.md
provision_git now does two things: copy the host cwd's .git (when
--cwd is set, existing behavior) and write ~/.gitconfig with
pushInsteadOf rules for each bottle.git entry. A 'git push <real
upstream URL>' from inside the agent transparently rewrites to
'git://<gate>/<name>.git' so the gate gets first crack at the
incoming refs.
pushInsteadOf (not insteadOf) keeps fetch on the original URL —
v1 of the git-gate is push-only scope per PRD 0008. The render
helper is exposed for testing without docker.
Dockerfile.git-gate builds a small alpine image with git,
openssh-client, and gitleaks; the directory layout the entrypoint
and per-upstream cp's expect is pre-created in the image so docker
cp can target paths beneath /etc/git-gate and /git-gate/creds at
container-create time (cp doesn't create intermediate dirs).
DockerGitGate.start mirrors DockerSSHGate's shape: build, create,
cp the rendered entrypoint + hook + per-upstream identity files
(plus a known_hosts file synthesized from KnownHostKey when set),
attach the egress network, start. build_image gains an optional
dockerfile= argument so the gate can build from its own
Dockerfile in the shared context.
PRD: docs/prds/0008-git-gate.md
Mirrors the SSHGate/PipelockProxy shape: a host-side prepare that
lifts bottle.git into a tuple of GitGateUpstreams and renders two
shell scripts under stage_dir — the gate's entrypoint (which
initializes a bare repo per upstream and execs git daemon
--enable=receive-pack) and the shared pre-receive hook
(gitleaks-scan, then forward each accepted ref to the real
upstream using the per-repo credential).
Failure in either hook phase aborts the push so the agent sees a
real rejection, not a silent success. KnownHostKey absence is
fail-closed: the hook refuses to forward without a pinned key
rather than TOFU-trusting the upstream from inside the gate.
PRD: docs/prds/0008-git-gate.md
Each entry pairs a Name (local alias the gate exposes) with an
ssh:// Upstream URL, an IdentityFile the gate uses to push to
that upstream, and an optional KnownHostKey for upstream
host-key pinning. The Upstream URL is parsed at construction
into UpstreamUser/Host/Port/Path so downstream code doesn't
re-parse.
Two cross-validation rules: Names must be unique within a
bottle (each maps to a distinct bare repo), and no git entry's
(host, port) may overlap an ssh entry's (Hostname, Port) — the
same upstream reachable two ways would let a misbehaving agent
route around the gitleaks-bearing git-gate via the L4 ssh-gate.
PRD: docs/prds/0008-git-gate.md
Per-bottle sidecar that fronts the agent's git remotes, runs
gitleaks via a pre-receive hook, and only forwards to the real
upstream on a clean scan. Upstream push credentials live in the
gate, not the agent — so a misbehaving agent cannot push a
secret-bearing commit past it.
Pipelock's BIP-39 seed-phrase scanner fires on Anthropic Messages API
bodies because user-authored conversation text can hit 12 consecutive
BIP-39 dictionary words that pass the checksum, returning a 403
`blocked: request body contains secret: BIP-39 Seed Phrase` that the
Claude CLI surfaces as `Please run /login`. Pipelock's `suppress`
section only covers git/file findings, not the inline body scanner,
so the recommended treatment for LLM endpoints is
`tls_interception.passthrough_domains`: CONNECT is still allowlist-
gated, but the body is not MITM'd. The existing body-scan integration
test moves to `raw.githubusercontent.com` so it still pins TLS body
DLP on non-passthrough'd hosts.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Threat-models the case where a credential ends up in a tracked
file and is git-pushed to a public remote — the secret is
compromised the instant the push lands (events API, scrapers),
not at merge time. Recommends gitleaks as the smallest-blast-
radius layer to add: Go binary, MIT, offline, scans full history,
hookable from the existing .githooks/.
No code or workflow change; just the research note.
Bug: git fetch failed with "connect to host
claude-bottle-ssh-gate-implementer port 30009: Connection refused".
OpenSSH treats a URL-supplied port (the user's remote was
ssh://git@gitea.dideric.is:30009/...) as overriding the
~/.ssh/config Port directive, so even though the config wrote
Port 30000 the agent dialed :30009 — where nothing was listening
because the gate had been assigned BASE_LISTEN_PORT + index.
Fix: the gate's listen port now equals the upstream port. Same
script, same socat, just port = entry.Port. Two entries on the
same upstream port are rejected at prepare time (the gate is one
container with a flat port space).
Re-smoked: probe nc github.com via the gate at :22, banner came
back as expected.
PRD 0007 updated to record the design refinement.
PRD 0007: the launch ExitStack calls gate.stop on every failure
path, so an early bring-up error (where the gate container was
never created) must not raise from teardown. Mirrors the existing
DockerPipelockProxy.stop assertion.
The orphan-container enumeration in cleanup.py already covers
ssh-gate containers via its `claude-bottle-` name prefix filter —
no code change there.
PRD 0007: SSH traffic now flows through the per-agent ssh-gate
sidecar, so pipelock should know nothing about bottle.ssh.
Removed:
- pipelock_bottle_ssh_hostnames, _trusted_domains, _ip_cidrs.
- The trusted_domains / ssrf blocks built from ssh entries.
- pipelock_proxy_host_port — its last caller (the ssh provisioner)
is gone.
- is_ipv4_literal — only used to classify ssh hostnames into
trusted_domains vs ssrf.ip_allowlist, both of which are gone.
api_allowlist now derives solely from baked-in defaults +
bottle.egress.allowlist. Tests updated to pin the new shape and
assert ssh hostnames do NOT leak into pipelock's config.
PRD 0007: stop tunneling ssh through pipelock. Each Host block in
the agent's ~/.ssh/config now points at the gate container + the
per-entry listen port; HostKeyAlias preserves host-key validation
against the real upstream name, and CheckHostIP=no skips the
resolved-IP path (which would otherwise hit the gate's IP).
known_hosts collapses to a single entry per upstream keyed on the
alias.
The pipelock_proxy_host_port import is gone from this module; the
function itself becomes dead code and gets removed alongside the
broader pipelock SSH carve-outs in the next commit.
PRD 0007: thread the DockerSSHGate through the bottle lifecycle.
- DockerBottlePlan gains gate_plan: SSHGatePlan.
- prepare.resolve_plan accepts a gate and renders its entrypoint
script next to the pipelock yaml.
- launch.launch starts the gate sidecar after pipelock (so it's on
the same internal + egress networks) and registers its stop in
the ExitStack. Skipped when the bottle has no ssh entries.
- DockerBottleBackend instantiates DockerSSHGate alongside the
pipelock proxy.
- bottle_plan.print + to_dict surface the upstream table so
--dry-run shows the per-host listen-port mapping.
ssh_config provisioning still points at pipelock; that swap lands
in the next commit so this one stays a pure wiring change.
PRD 0007: Docker-specific start/stop for the SSH egress gate.
Mirrors DockerPipelockProxy: docker create on the internal
network with /bin/sh entrypoint, docker cp the staged entrypoint
script in, attach to the egress network, docker start. Image is
alpine/socat pinned by digest — self-sufficient at boot so the
gate's agent-facing leg can stay on the --internal network.
Not yet wired into the bottle launch path; that lands next.
First piece of PRD 0007: the per-agent SSH egress gate that will
let pipelock stop seeing SSH traffic. This commit only lands the
backend-agnostic surface — the SSHGate ABC, SSHGatePlan, the
listen-port assignment (BASE_LISTEN_PORT + index), and the
entrypoint-script renderer. Backend wiring lands in follow-up
commits.
The gate's agent-facing leg sits on the `--internal` network, so
the forwarder image cannot rely on apk/apt at startup. Surfaced
by the DNS spike — a placeholder using `apk add socat` died
silently and gave a false-negative DNS-on-internal result.
Spike: container on a `--internal` user-defined network resolves
another container's name via the embedded resolver at 127.0.0.11
and reaches it over TCP, while egress to the public internet
remains blocked. The PRD's design assumption holds — no design
change needed.
PRD 0006 enabled pipelock's native TLS interception, which broke
git fetch over SSH from inside the agent: pipelock's SNI gate
rejects the SSH banner that follows CONNECT. Document the
architectural fix — a dedicated per-agent TCP-forwarder sidecar
built from bottle.ssh entries — so pipelock can stay maximally
strict on the HTTPS path with no SSH carve-outs.
The directory carries this session's scheduled-tasks lock file and
agent-memory cache; both are per-user state, not project artifacts.
Stops `git status` from listing it on every command.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Fourth and final step of PRD 0006. Two new end-to-end tests pin
the two paths through pipelock's tls_interception layer.
- test_pipelock_blocks_secret_https_post: posts a GitHub-PAT-shaped
body to api.anthropic.com over HTTPS through the bottle. With
pipelock now bumping the CONNECT and seeing the decrypted body,
it returns 403 with the documented `blocked: request body
contains secret: GitHub Token` body. The probe is a single curl
invocation — curl natively does CONNECT through HTTPS_PROXY, the
agent's trust store now contains pipelock's CA, no hand-rolled
TLS in the test.
- test_pipelock_allows_normal_https: GETs git's README from
raw.githubusercontent.com (a baked-in allowlist host). 200 +
non-zero body length proves the full chain works:
pipelock_tls_init → docker cp of CA into sidecar → bumped CONNECT
→ provision_ca installed CA in agent → curl trusts pipelock's
bumped leaf → body forwarded back through the tunnel.
- test_pipelock_sidecar_smoke: pre-existing direct-start smoke
test updated to call pipelock_tls_init and populate the CA
paths on the plan. (The full launch flow does this in launch.py;
this test exercises the proxy class in isolation.)
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
didericis