## Why
An AI agent autonomously deleted a live task (FN-8609) via
`fn_task_delete`. Investigating that led to two audits, which found the
gate machinery is well-built and its *enforcement* well-tested — but the
*decision boundary* was unguarded and the tools that matter never
reached the gate at all.
The headline defect: all ~76 tools in `packages/cli/src/extension.ts`
are delivered through `hostExtensionPaths` → `DefaultResourceLoader`,
which never passes through `wrapToolsWithActionGate`. `fn_task_delete`
*was* classified as gated in `gating-classifications.ts` — that
classification was dead code. Roughly 15 destructive tools were ungated
even for an operator who had locked their policy all the way down.
Second: `POST /api/approvals/:id/decision` took the approver identity
from the request body, validated for shape only, with no check that the
decider was human or differed from the requester. Combined with a bash
tool whose command string was never inspected (only `params.cwd`), an
agent could read the daemon token from `~/.fusion/settings.json`, find
its own pending request, and approve it.
## Scope
This does **not** tighten defaults. The `unrestricted` preset is
byte-for-byte unchanged — `git diff` on `agent-permission-policy.ts` is
empty — and regression tests assert that an out-of-the-box install
behaves exactly as before. The bug was never "the default is too
permissive"; it was "strict policy doesn't enforce." This makes turning
security up actually work.
The one deliberate exception: the containment that stops an agent
escalating its *own* privileges (reading the daemon token / credentials,
calling the approvals API to self-approve) applies at every preset
including `unrestricted`. That is a privilege-escalation boundary rather
than a permission preference — if it only engaged under strict policy it
would not have prevented the incident that prompted this.
## What changed
8 bisectable commits:
- **Approval lifecycle** — self-approval blocked via server-derived
deciders; same-verdict replay 409s; decide re-reads and re-validates
inside the transaction; expiry TTLs; `markCompleted` ownership check;
session identity registry in core.
- **Engine gates enforce for real** — unclassified tools resolve to a
policy-governed category instead of hardcoded `allow`; missing-policy
fail-open closed; bash containment floor + exact-command approval
binding.
- **Dashboard decision routes** — stop trusting client-supplied actors
(decision, bypass-review, worktrunk → 403 on forged actors).
- **`fn serve` authenticated by default** — auto-mints a token following
the existing `fn dashboard` precedent; `--no-auth` opts out.
- **Sibling entry points closed** — user-sourced hard-cancel moves, ACP
execute-once approvals, plugin task-store gating.
- **pi-extension principal resolution** — the extension resolves the
acting principal and can withhold or policy-gate the previously ungated
destructive tools.
- **Root-cause bonus fix** — `findLatestByDedupeKey` was broken in
PostgreSQL backend mode (already-parsed jsonb fed through a string-only
parser), so approved-grant redemption **never matched in production**,
minting duplicate requests. This explains the live DB state of 17
approved / 0 completed. *(Also cherry-picked to `main` as `a9b30013bb`,
since it is an active production defect on its own.)*
- **Review follow-ups** (`627f1b1fa8`) — operator-configured
provisioning privilege and a configurable grant TTL; see below.
## Review follow-ups
**Provisioning privilege is operator-configured, not role-derived.**
`isCallerPrivileged` had gone from `caller.reportsTo == null` (every
top-level agent privileged — permanent escalation by creating a
manager-less agent) to `caller.role === "ceo"`, which swapped an
implicit rule for a magic string: any agent config can claim that role,
while an operator who genuinely wants a privileged agent had no
supported way to say so. Privilege now derives solely from
`agentProvisioning.trustedAgentIds` / `trustedRoles` and fails closed
when settings are unresolvable.
It is also no longer forwarded to `resolveAgentProvisioningPolicy` as
`isPrivileged`, because that flag short-circuits ahead of
`alwaysApproveDelete` — a trusted caller was bypassing delete approval
entirely. The policy applies the same trusted rules itself, in the right
order. The function now governs only the org-chart escape hatch (acting
outside your own direct reports).
**Grant TTL defaults to 1 hour and is configurable.** Approval →
redemption is not instantaneous: an operator approving from their phone,
an engine restart, a queued lane, or a task waiting on a worktree all
routinely exceeded 15 minutes, after which the grant expired and the
agent silently re-requested. One hour remains far short of the
"redeemable forever" hazard the TTL exists to bound. Override via
`FUSION_APPROVAL_GRANT_TTL_MS` or `configureApprovalRequestTtls()`;
invalid overrides are ignored rather than widening the window to
infinity or collapsing it to zero.
## Behavior changes requiring operator review before rollout
1. `fn serve` requires a bearer token by default (`--no-auth` opts out);
unauthenticated clients get 401.
2. Agents can no longer run withheld destructive tools
(`fn_task_delete`, `fn_task_bypass_review`,
mission/milestone/slice/feature/workflow deletes, `experiment_finalize`,
`skills_install`). Operators keep them via CLI/dashboard. **This is the
incident fix.**
3. Agents get provisioning privilege only when the operator lists them
in `agentProvisioning.trustedAgentIds` / `trustedRoles`; the
provisioning gate is now live in production. Previously-implicit
privilege (top-level position, or a `ceo` role) no longer grants
anything on its own.
4. Decision replay 409s (was 200); pending approvals expire after 24h,
approved grants after 1h (configurable); bash approvals bind per exact
command.
5. Forged/body actors on decision, bypass-review, worktrunk routes →
403; `archive-all-done` requires `{confirm:true}` (external scripts
affected).
6. `fn_secret_get` approvals grant exactly one reveal (previously
granted nothing and looped forever); ACP approvals are execute-once
(previously infinite reuse).
7. Bash containment denies token/credential/approvals-API commands in
all agent sessions at every preset.
## Verification
Independently re-run against the branch, not just self-reported:
- 5 typechecks (core, engine, cli, dashboard `tsconfig.json` +
`tsconfig.app.json`) — clean
- `pnpm lint` — clean
- `pnpm test:gate` — 379 passed
- `pnpm build --force` — green (a plain `pnpm build` skips packages as
unchanged and does **not** compile the branch)
- `pnpm check:changesets` — clean
- ~650 file-scoped tests including new negative-path suites for the
decision boundary, which previously had **zero** test coverage
`packages/engine/src/__tests__/plugin-runner.test.ts` fails 56/80 —
**verified pre-existing**, reproducing identically at base commit
`93a403af67` on `main`. Not in the merge gate.
### A mutation check that failed to fail
Worth recording, because it nearly shipped an untested security fix. The
first mutation check on the provisioning change reintroduced the `ceo`
hardcode and **all 17 tests still passed** — the tests asserted through
the policy path, which can no longer observe `isCallerPrivileged` at
all, precisely because `isPrivileged` is no longer forwarded there.
Org-chart cases that do exercise the function were added; the hardcode
now fails exactly 1 of 19, and restoring is green. A green mutation run
is only meaningful if the test can actually see the code under test.
## Known limitations (stated, not papered over)
- The bash containment floor is string-matching: a cost-raiser, not a
sandbox. Quoting, encoding, `$HOME`, symlinks, or an interpreter
one-liner can evade it. The durable protection is the decision route
refusing agent-originated deciders — the filter is the belt, not the
braces.
- Approval expiry is lazy (evaluated at decide/complete/redeem), not
swept, so an expired pending row stays visible in lists until touched.
- The extension's require-approval path returns a pending message but
cannot suspend a pi session mid-turn; engine-side pause hooks cover
engine lanes only.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Security**
* Hardened approval and permission gating with server-side decider
attribution, self-approval blocking, ownership checks, replay/race
protection, and status/TTL enforcement.
* Added fail-closed behavior for sensitive/unclassified tools and
sandbox provisioning approvals.
* Blocked credential/approval access via bash containment; plugin
destructive task operations now require explicit permission.
* **New Features**
* `fn serve` now defaults to bearer-token auth, with `--no-auth` as the
explicit opt-out.
* **Bug Fixes**
* Improved task move-source attribution (`moveSource: "user"`) and
tightened dashboard archive/bypass confirmation and operator attribution
behavior.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
128 lines
5.3 KiB
TypeScript
128 lines
5.3 KiB
TypeScript
import { homedir } from "node:os";
|
|
|
|
/*
|
|
FNXC:BashContainment 2026-07-26-12:40:
|
|
Security incident follow-up: an agent's real escalation chain was
|
|
`bash("cat ~/.fusion/settings.json")` (agents run as the same OS user, so the
|
|
0600 file holding `daemonToken` is readable) → find its own pending approval
|
|
request → `curl` the approvals decision API to approve itself. Fixing the
|
|
decision route alone is insufficient while bash can read the token.
|
|
|
|
This module is a PRIVILEGE-ESCALATION FLOOR, not a permission preference:
|
|
it holds UNCONDITIONALLY at every permission preset including the default
|
|
`unrestricted`. Ordinary bash permission gating stays policy-driven (see
|
|
agent-action-gate.ts); this floor only denies the narrow set of commands
|
|
that let an agent escalate its OWN privileges:
|
|
1. reads of the Fusion global config / daemon token (~/.fusion, any
|
|
.fusion/settings.json spelling, FUSION_DAEMON_TOKEN / daemonToken refs),
|
|
2. reads of common same-user credential stores (~/.ssh, ~/.aws, ~/.netrc,
|
|
~/.npmrc, ~/.config/gh, ~/.docker/config.json, ~/.gnupg, ~/.kube),
|
|
3. self-directed calls to the approvals API (/api/approvals, fn_token=).
|
|
|
|
HONESTY / LIMITATIONS (do not oversell this): matching happens on a
|
|
normalized copy of the command STRING (quotes/backslashes stripped, $HOME and
|
|
the literal home directory folded to ~, lowercased). That defeats the trivial
|
|
bypasses (`cat ~/.fus''ion/settings.json`, `"$HOME"/.fusion`, absolute home
|
|
paths) but NOT determined indirection: base64-built paths, variable
|
|
concatenation (a=fus; b=ion; cat ~/.$a$b/...), glob spelling (~/.f*sion),
|
|
symlinks created first, python/node one-liners, or reading via a spawned
|
|
interpreter. String inspection of shell cannot be sound. This floor raises
|
|
the cost of the KNOWN escalation chain and creates an audit-visible denial;
|
|
real containment requires OS-level isolation (separate user / sandbox),
|
|
which is out of scope here. Deny messages are deliberately explicit so a
|
|
probing agent's attempts are visible in session logs.
|
|
*/
|
|
|
|
export interface BashContainmentVerdict {
|
|
allowed: boolean;
|
|
/** Stable rule id for logs/tests when denied. */
|
|
rule?: string;
|
|
/** Human-readable denial reason (safe to surface to the agent). */
|
|
reason?: string;
|
|
}
|
|
|
|
interface ContainmentRule {
|
|
id: string;
|
|
pattern: RegExp;
|
|
reason: string;
|
|
}
|
|
|
|
/*
|
|
FNXC:BashContainment 2026-07-26-12:40:
|
|
Rules match the NORMALIZED command (see normalizeCommand). Home-anchored
|
|
patterns use `~/.<dir>` because normalization folds $HOME/${HOME}/absolute
|
|
home spellings to `~`. `/users/<name>/` and `/home/<name>/` cover OTHER
|
|
users' homes which normalization cannot fold.
|
|
*/
|
|
const RULES: readonly ContainmentRule[] = [
|
|
{
|
|
id: "fusion-global-dir",
|
|
pattern: /(?:~|\/users\/[^/\s]+|\/home\/[^/\s]+)\/\.fusion\b/,
|
|
reason: "access to the global Fusion directory (daemon token / global settings) is not permitted from agent sessions",
|
|
},
|
|
{
|
|
id: "fusion-settings-file",
|
|
pattern: /\.fusion\/settings\.json/,
|
|
reason: "access to Fusion settings.json is not permitted from agent sessions",
|
|
},
|
|
{
|
|
id: "fusion-daemon-token",
|
|
pattern: /fusion_daemon_token|fusion_dashboard_token|daemontoken/,
|
|
reason: "referencing the Fusion daemon token is not permitted from agent sessions",
|
|
},
|
|
{
|
|
id: "credential-store",
|
|
pattern: /(?:~|\/users\/[^/\s]+|\/home\/[^/\s]+)\/(?:\.ssh|\.aws|\.netrc|\.npmrc|\.gnupg|\.kube|\.config\/gh|\.docker\/config\.json)\b/,
|
|
reason: "access to user credential stores is not permitted from agent sessions",
|
|
},
|
|
{
|
|
id: "approvals-api",
|
|
pattern: /\/api\/approvals|fn_token=/,
|
|
reason: "calling the Fusion approvals API from a shell is not permitted from agent sessions (approvals are decided by the operator)",
|
|
},
|
|
];
|
|
|
|
function escapeRegExp(value: string): string {
|
|
return value.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
|
|
}
|
|
|
|
const HOME_DIR = homedir();
|
|
const HOME_PATTERN = new RegExp(escapeRegExp(HOME_DIR), "gi");
|
|
|
|
/**
|
|
* FNXC:BashContainment 2026-07-26-12:40:
|
|
* Normalization defeats quote-splitting and $HOME spellings only. Keep this
|
|
* pure and dependency-free so it is trivially unit-testable.
|
|
*/
|
|
export function normalizeBashCommandForContainment(command: string): string {
|
|
let normalized = command.replace(/["'\\]/g, "");
|
|
normalized = normalized.replace(/\$\{home\}/gi, "~").replace(/\$home\b/gi, "~");
|
|
if (HOME_DIR && HOME_DIR !== "/") {
|
|
normalized = normalized.replace(HOME_PATTERN, "~");
|
|
}
|
|
return normalized.toLowerCase();
|
|
}
|
|
|
|
/** Evaluate the unconditional containment floor for one bash command string. */
|
|
export function evaluateBashContainment(command: string): BashContainmentVerdict {
|
|
if (typeof command !== "string" || command.trim() === "") {
|
|
return { allowed: true };
|
|
}
|
|
const normalized = normalizeBashCommandForContainment(command);
|
|
for (const rule of RULES) {
|
|
if (rule.pattern.test(normalized)) {
|
|
return { allowed: false, rule: rule.id, reason: rule.reason };
|
|
}
|
|
}
|
|
return { allowed: true };
|
|
}
|
|
|
|
/** Stable message shown to the agent on denial. */
|
|
export function buildBashContainmentDenialMessage(verdict: BashContainmentVerdict): string {
|
|
return (
|
|
`Command blocked by Fusion privilege-escalation containment (${verdict.rule ?? "containment"}): ` +
|
|
`${verdict.reason ?? "not permitted"}. This boundary applies at every permission preset; ` +
|
|
`do not attempt to work around it — ask the operator instead.`
|
|
);
|
|
}
|