feat(control-plane): role-scoped signed tokens so the gateway can't drive operator routes
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

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>
This commit is contained in:
2026-07-24 05:18:39 +00:00
parent 72fdb1d14b
commit 1db2a9eb67
15 changed files with 485 additions and 111 deletions
@@ -25,6 +25,7 @@ import tempfile
import unittest
from pathlib import Path
from bot_bottle.control_auth import ROLE_CLI, ROLE_GATEWAY, mint
from bot_bottle.orchestrator.client import OrchestratorClient
from bot_bottle.orchestrator.lifecycle import OrchestratorService
from bot_bottle.paths import host_control_plane_token
@@ -85,7 +86,11 @@ class TestDockerControlPlaneAuthIntegration(unittest.TestCase):
host_root=host_root,
)
cls.svc.ensure_running()
cls.token = host_control_plane_token()
# The control plane now verifies role-scoped signed tokens, not the raw
# key. Mint one of each role from the host signing key (issue #469 review).
signing_key = host_control_plane_token()
cls.cli_token = mint(ROLE_CLI, signing_key)
cls.gateway_token = mint(ROLE_GATEWAY, signing_key)
@staticmethod
def _teardown_docker(
@@ -136,11 +141,17 @@ class TestDockerControlPlaneAuthIntegration(unittest.TestCase):
status, _ = self._request("GET", "/bottles", token="not-the-real-secret")
self.assertEqual(401, status)
def test_bottles_accepts_the_real_token(self) -> None:
status, payload = self._request("GET", "/bottles", token=self.token)
def test_bottles_accepts_the_cli_token(self) -> None:
status, payload = self._request("GET", "/bottles", token=self.cli_token)
self.assertEqual(200, status)
self.assertEqual([], payload["bottles"])
def test_bottles_rejects_the_gateway_token(self) -> None:
"""A compromised gateway holds only a `gateway` token — it must not be
able to enumerate bottles (an operator route) with it (issue #469)."""
status, _ = self._request("GET", "/bottles", token=self.gateway_token)
self.assertEqual(403, status)
def test_resolve_rejects_a_caller_with_no_token(self) -> None:
"""The credential-lift attack from issue #400: an agent could POST
/resolve directly and read back the upstream tokens it's never meant
@@ -148,6 +159,13 @@ class TestDockerControlPlaneAuthIntegration(unittest.TestCase):
status, _ = self._request("POST", "/resolve")
self.assertEqual(401, status)
def test_resolve_accepts_the_gateway_token(self) -> None:
"""The gateway's own token DOES reach /resolve: with an empty body it
gets 400 (missing source_ip) rather than the 401 a rejected caller sees
— proving the role gate let the gateway token through."""
status, _ = self._request("POST", "/resolve", token=self.gateway_token)
self.assertEqual(400, status) # past the role gate; body validation fails
if __name__ == "__main__":
unittest.main()