Files
PaperClipAI/server/src/__tests__/agent-auth-middleware.test.ts
T
Devin FoleyandPaperclip b2e9e82f05 fix: stop remote Grok runs before continuing queued messages (#14100)
## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work.
> - The execution service owns each run and saves messages sent while it
runs.
> - Interrupt must stop the current executor before it delivers those
messages.
> - Remote Grok commands did not register the host cancellation control.
> - A cancelled task run could still write Done and prevent queue
recovery.
> - This pull request connects remote cancellation and revokes cancelled
run writes.
> - Saved input can use the existing queue admission rules after
verified cleanup.

## Linked Issues or Issue Description

**What happened?**

Interrupting a queued message marked a remote Grok run cancelled before
its sandbox stopped. The old run could still post a reply and mark the
task Done. Its saved follow-up remained deferred behind execution
recovery.

**Expected behavior**

Stop revokes run write authority and waits for verified termination.
Saved messages remain durable and enter one successor through normal
admission after cleanup.

**Steps to reproduce**

1. Run a task with `grok_local` in a remote sandbox.
2. Send a follow-up and use Interrupt while the command runs.
3. Let the old command attempt a task status update after cancellation.
4. Observe the task disposition and the saved message queue.

**Paperclip version or commit**

The gap is present in master at `d3e0f0a238`.

**Deployment mode**

Authenticated server with a Daytona sandbox.

Related work: #14028 and #14046 handle bounded continuation. #13291
covers infrastructure interruption and verified remote cleanup. #13332
addresses atomic recovery holds. This change handles direct Grok
operator cancellation and stale task writes.

## What Changed

- Register remote Grok cancellation before preparation. Keep command
ownership until the host confirms sandbox termination.
- Reuse the sandbox cancellation boundary for the direct CLI invocation.
Reject fresh attempts after cancellation and preserve workspace restore
failure evidence.
- Reject writes from cancelled task JWTs and runs with a pending stop.
Preserve diagnostic reads and existing conversation error codes.
- Recheck run authority under a database lock before task updates and
interaction responses commit.
- Preserve authorized handoffs that stop their own run. Only the
server-issued stop receipt for that request permits the final task
update.
- Add tests for hung commands, unverified stops, early cancellation,
copy-back failures, late Done, late interaction responses, authorized
handoffs, exact lease receipts, and one queue successor across
concurrent restart sweeps.
- Document the cancellation and write-authority contract.

## Verification

- Targeted adapter, cancellation-boundary, authentication,
queued-message, interaction-service, and activity-route tests passed.
The expanded run passed 214 tests; one new test had an incomplete
fixture. After correcting the fixture, all 8 selected follow-up cases
passed.
- `pnpm -r typecheck`: passed on
`179c86caf1bf0d89914a503d46e24af7e4b8c557`.
- `pnpm build`: passed on the same commit.
- `pnpm test:run`: the general-server group completed with 13,521
passed, 99 skipped, and 18 failed tests. It then stopped, so the
remaining local groups did not run. Five Slack, email, and wake-batching
failures passed on focused reruns after correcting the local
environment. The remaining 13 failures reproduce as `EACCES` on rename
in unchanged skill-cache code on macOS. Two custom-image suite setup
hooks also failed to start embedded PostgreSQL after the machine
exhausted shared-memory slots; all 31 tests in that file passed on rerun
after the local resource issue was resolved. CI covers all test groups.
- CI: 53 checks passed and 2 were skipped on the latest commit,
including the aggregate verification gate. The last server shard passed
on its single rerun after a preview-server startup timeout. The affected
file also passed locally with 28 passed and 3 skipped.
- Greptile: 5/5 on the latest commit. Both review threads are resolved.
- No live deployment or staging task mutation has been performed.

## Risks

- Stopping the sandbox can prevent file copy-back. The result preserves
workspace restore failure evidence; termination does not imply restored
files.
- If provider termination fails, the adapter keeps ownership of its
outstanding command and does not acknowledge Stop.
- The write restriction now applies to ordinary cancelled tasks. Reads
remain allowed. Task and interaction checks add a shared run-row lock to
agent mutations. An exact server-issued receipt permits the task request
that stopped its own run to complete its handoff.
- Existing terminal tasks are not reopened automatically. An operator
must correct a historical late Done before its saved queue can continue.
- No schema migration or UI change.

## Model Used

OpenAI GPT-6 through Codex, with reasoning, repository inspection, code
execution, and test tools. The precise backend revision and
context-window size are not exposed in this session.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have searched GitHub for duplicate or related PRs and linked
them above
- [x] I have either (a) linked existing issues with `Fixes: #` / `Closes
#` OR (b) described the issue in-PR following the relevant issue
template
- [x] I have not referenced internal/instance-local Paperclip issues or
links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip`
URLs)
- [x] My branch name describes the change (e.g. `docs/...`, `fix/...`)
and contains no internal Paperclip ticket id or instance-derived details
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] All Paperclip CI gates are green
- [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Paperclip <noreply@paperclip.ing>
2026-09-26 17:07:07 -07:00

551 lines
20 KiB
TypeScript

import { createHash, createHmac, randomUUID } from "node:crypto";
import express from "express";
import request from "supertest";
import { afterEach, beforeEach, describe, expect, it } from "vitest";
import {
activityLog,
agentApiKeys,
agents,
boardApiKeys,
heartbeatRuns,
} from "@paperclipai/db";
import { actorMiddleware } from "../middleware/auth.js";
import { errorHandler } from "../middleware/error-handler.js";
import { createLocalAgentJwt } from "../agent-auth-jwt.js";
import { assertCompanyAccess } from "../routes/authz.js";
function hashToken(token: string) {
return createHash("sha256").update(token).digest("hex");
}
function createSelectChain(rowsForTable: (table: unknown) => unknown[]) {
return {
from(table: unknown) {
return {
where() {
return Promise.resolve(rowsForTable(table));
},
};
},
};
}
function createDbState(input: {
agent: { id: string; companyId: string; status?: string };
agentKey?: { id: string; agentId: string; companyId: string; keyHash: string; responsibleUserId?: string | null };
run?: { id: string; companyId: string; agentId: string; responsibleUserId?: string | null;
status?: string; contextSnapshot?: Record<string, unknown>; resultJson?: Record<string, unknown> };
}) {
const activity: Array<Record<string, unknown>> = [];
const agentRow = {
id: input.agent.id,
companyId: input.agent.companyId,
status: input.agent.status ?? "active",
};
const keyRow = input.agentKey
? {
id: input.agentKey.id,
agentId: input.agentKey.agentId,
companyId: input.agentKey.companyId,
keyHash: input.agentKey.keyHash,
responsibleUserId: input.agentKey.responsibleUserId ?? null,
revokedAt: null,
scopeConfig: null,
}
: null;
const runRow = input.run
? {
id: input.run.id,
companyId: input.run.companyId,
agentId: input.run.agentId,
responsibleUserId: input.run.responsibleUserId ?? null,
status: input.run.status ?? "running",
contextSnapshot: input.run.contextSnapshot ?? {},
resultJson: input.run.resultJson ?? {},
}
: null;
const db = {
select: () =>
createSelectChain((table) => {
if (table === boardApiKeys) return [];
if (table === agentApiKeys) return keyRow ? [keyRow] : [];
if (table === agents) return [agentRow];
if (table === heartbeatRuns) return runRow ? [runRow] : [];
return [];
}),
update: () => ({
set() {
return {
where() {
return Promise.resolve([]);
},
};
},
}),
insert: (table: unknown) => ({
values(values: Record<string, unknown>) {
if (table === activityLog) activity.push(values);
return Promise.resolve([]);
},
}),
} as any;
return { db, activity };
}
function createApp(db: any, deploymentMode: "authenticated" | "local_trusted" = "authenticated") {
const app = express();
app.use(express.json());
app.use(
actorMiddleware(db, {
deploymentMode,
resolveSession: async () => null,
}),
);
app.get("/actor", (req, res) => {
res.json(req.actor);
});
app.post("/api/routine-triggers/public/:publicId/fire", (req, res) => {
res.json({ reachedWebhook: true, actorType: req.actor.type });
});
app.post("/mcp/gateways/:gatewayPublicId", (req, res) => {
res.json({ reachedGatewayProtocol: true, actorType: req.actor.type });
});
app.get("/companies/:companyId/protected", (req, res) => {
assertCompanyAccess(req, req.params.companyId);
res.json({ ok: true });
});
app.get("/companies/:companyId/issues/:issueId", (req, res) => {
assertCompanyAccess(req, req.params.companyId);
res.json({ id: req.params.issueId, readable: true });
});
app.patch("/companies/:companyId/issues/:issueId", (req, res) => {
assertCompanyAccess(req, req.params.companyId);
res.json({ id: req.params.issueId, writable: true });
});
app.use(errorHandler);
return app;
}
function craftAgentJwtWithoutResponsibleClaim(input: {
secret: string;
agentId: string;
companyId: string;
adapterType: string;
runId: string;
expiresInSeconds?: number;
}) {
const now = Math.floor(Date.now() / 1000);
const header = { alg: "HS256", typ: "JWT" };
const claims = {
sub: input.agentId,
company_id: input.companyId,
adapter_type: input.adapterType,
run_id: input.runId,
iat: now,
exp: now + (input.expiresInSeconds ?? 3600),
iss: "paperclip",
aud: "paperclip-api",
};
const headerB64 = Buffer.from(JSON.stringify(header), "utf8").toString("base64url");
const claimsB64 = Buffer.from(JSON.stringify(claims), "utf8").toString("base64url");
const signingInput = `${headerB64}.${claimsB64}`;
// Sign with the same per-instance, per-company key the server derives. The
// instance defaults to "default" (beforeEach clears PAPERCLIP_INSTANCE_ID),
// matching the live control plane this middleware test exercises. This helper
// only omits the responsible_user_id claim — it is not a cross-instance token.
const signingKey = createHmac("sha256", input.secret).update(`jwt:default:${input.companyId}`).digest("hex");
const signature = createHmac("sha256", signingKey).update(signingInput).digest("base64url");
return `${signingInput}.${signature}`;
}
describe("agent auth middleware", () => {
const originalSecret = process.env.PAPERCLIP_AGENT_JWT_SECRET;
const originalTtl = process.env.PAPERCLIP_AGENT_JWT_TTL_SECONDS;
const originalInstanceId = process.env.PAPERCLIP_INSTANCE_ID;
beforeEach(() => {
process.env.PAPERCLIP_AGENT_JWT_SECRET = "auth-middleware-secret";
process.env.PAPERCLIP_AGENT_JWT_TTL_SECONDS = "3600";
// Pin the control-plane instance so mint/verify (and the hand-crafted
// legacy token helper) all derive keys under the "default" live instance.
delete process.env.PAPERCLIP_INSTANCE_ID;
});
afterEach(() => {
if (originalSecret === undefined) delete process.env.PAPERCLIP_AGENT_JWT_SECRET;
else process.env.PAPERCLIP_AGENT_JWT_SECRET = originalSecret;
if (originalTtl === undefined) delete process.env.PAPERCLIP_AGENT_JWT_TTL_SECONDS;
else process.env.PAPERCLIP_AGENT_JWT_TTL_SECONDS = originalTtl;
if (originalInstanceId === undefined) delete process.env.PAPERCLIP_INSTANCE_ID;
else process.env.PAPERCLIP_INSTANCE_ID = originalInstanceId;
});
it.each([
{ status: "cancelled", conversationMode: false, requested: false },
{ status: "running", conversationMode: false, requested: true },
{ status: "cancelled", conversationMode: true, requested: false },
{ status: "running", conversationMode: true, requested: true },
])("revokes writes but preserves reads for a stopped run: %j", async ({ status, conversationMode, requested }) => {
const agentId = randomUUID();
const companyId = randomUUID();
const runId = randomUUID();
const { db } = createDbState({ agent: { id: agentId, companyId }, run: {
id: runId, companyId, agentId, status, contextSnapshot: { conversationMode },
resultJson: requested ? { executionCancellation: { state: "requested" } } : {},
} });
const token = createLocalAgentJwt(agentId, companyId, "grok_local", runId, null);
const client = createApp(db);
const endpoint = `/companies/${companyId}/issues/${randomUUID()}`;
const write = await request(client).patch(endpoint).set("Authorization", `Bearer ${token}`).send({ status: "done" });
expect(write.status).toBe(403);
expect(write.body.code).toBe(conversationMode ? "conversation_turn_cancelled" : "agent_run_cancelled");
const read = await request(client).get(endpoint).set("Authorization", `Bearer ${token}`);
expect(read.status).toBe(200);
expect(read.body.readable).toBe(true);
});
it("keeps header-less local requests as the implicit board actor with their run id", async () => {
const runId = randomUUID();
const { db } = createDbState({ agent: { id: randomUUID(), companyId: randomUUID() } });
const res = await request(createApp(db, "local_trusted"))
.get("/actor")
.set("X-Paperclip-Run-Id", runId);
expect(res.status).toBe(200);
expect(res.body).toMatchObject({ type: "board", userId: "local-board", runId });
});
it.each([
["empty bearer token", "Bearer ", "Empty bearer token"],
["unverified token", "Bearer not-a-token", "Agent token did not verify"],
])("rejects %s instead of retaining the implicit local-board actor", async (_label, authorization, error) => {
const { db } = createDbState({ agent: { id: randomUUID(), companyId: randomUUID() } });
let commentWrites = 0;
const app = createApp(db, "local_trusted");
app.post("/comments", (_req, res) => {
commentWrites += 1;
res.status(201).json({ ok: true });
});
const res = await request(app).post("/comments").set("Authorization", authorization).send({ body: "reply" });
expect(res.status).toBe(401);
expect(res.body.error).toContain(error);
expect(commentWrites).toBe(0);
});
it.each(["authenticated", "local_trusted"] as const)(
"leaves webhook authentication to the trigger in %s mode",
async (deploymentMode) => {
const { db } = createDbState({ agent: { id: randomUUID(), companyId: randomUUID() } });
const res = await request(createApp(db, deploymentMode))
.post(`/api/routine-triggers/public/${"a".repeat(24)}/fire`)
.set("Authorization", "Bearer routine-secret")
.send({ event: "created" });
expect(res.status).toBe(200);
expect(res.body).toEqual({ reachedWebhook: true, actorType: "none" });
},
);
it.each([
`/api/routine-triggers/public/${"a".repeat(24)}/rotate-secret`,
"/api/routine-triggers/public/not-a-public-id/fire",
`/api/routine-triggers/public/${"a".repeat(24)}/fire/extra`,
])("does not bypass actor authentication for %s", async (path) => {
const { db } = createDbState({ agent: { id: randomUUID(), companyId: randomUUID() } });
const res = await request(createApp(db)).post(path).set("Authorization", "Bearer routine-secret");
expect(res.status).toBe(401);
});
it("leaves public MCP gateway bearers for the gateway protocol to validate", async () => {
const { db } = createDbState({ agent: { id: randomUUID(), companyId: randomUUID() } });
const publicId = `gw_${"a".repeat(32)}`;
const res = await request(createApp(db, "local_trusted"))
.post(`/mcp/gateways/${publicId}`)
.set("Authorization", "Bearer pcgw_runtime_token")
.send({ jsonrpc: "2.0", id: 1, method: "initialize" });
expect(res.status).toBe(200);
expect(res.body).toMatchObject({ reachedGatewayProtocol: true });
});
it("does not bypass actor authentication for lookalike MCP gateway paths", async () => {
const { db } = createDbState({ agent: { id: randomUUID(), companyId: randomUUID() } });
const res = await request(createApp(db, "local_trusted"))
.post("/mcp/gateways/not-a-public-id")
.set("Authorization", "Bearer pcgw_runtime_token")
.send({ jsonrpc: "2.0", id: 1, method: "initialize" });
expect(res.status).toBe(401);
expect(res.body.error).toContain("Agent token did not verify");
});
it.each([
["terminated", "Agent is terminated"],
["pending_approval", "Agent is pending approval"],
])("rejects a %s agent JWT instead of retaining local-board", async (status, error) => {
const companyId = randomUUID();
const agentId = randomUUID();
const runId = randomUUID();
const { db } = createDbState({ agent: { id: agentId, companyId, status } });
const token = createLocalAgentJwt(agentId, companyId, "codex_local", runId, "user-1");
const res = await request(createApp(db, "local_trusted"))
.get("/actor")
.set("Authorization", `Bearer ${token}`);
expect(res.status).toBe(401);
expect(res.body.error).toContain(error);
});
it("rejects an agent JWT when the agent record belongs to another company", async () => {
const companyId = randomUUID();
const agentId = randomUUID();
const runId = randomUUID();
const { db } = createDbState({ agent: { id: agentId, companyId: randomUUID() } });
const token = createLocalAgentJwt(agentId, companyId, "codex_local", runId, "user-1");
const res = await request(createApp(db, "local_trusted"))
.get("/actor")
.set("Authorization", `Bearer ${token}`);
expect(res.status).toBe(401);
expect(res.body.error).toContain("missing or belongs to another company");
});
it("reports an expired agent JWT specifically", async () => {
const companyId = randomUUID();
const agentId = randomUUID();
const runId = randomUUID();
const { db } = createDbState({ agent: { id: agentId, companyId } });
const token = craftAgentJwtWithoutResponsibleClaim({
secret: process.env.PAPERCLIP_AGENT_JWT_SECRET!,
agentId,
companyId,
adapterType: "codex_local",
runId,
expiresInSeconds: -1,
});
const res = await request(createApp(db, "local_trusted"))
.get("/actor")
.set("Authorization", `Bearer ${token}`);
expect(res.status).toBe(401);
expect(res.body.error).toContain("Expired agent token");
});
it("uses the signed responsible_user_id claim and keeps the signed run id authoritative", async () => {
const companyId = randomUUID();
const agentId = randomUUID();
const runId = randomUUID();
const { db } = createDbState({
agent: { id: agentId, companyId },
run: { id: runId, companyId, agentId, responsibleUserId: "user-row" },
});
const token = createLocalAgentJwt(agentId, companyId, "codex_local", runId, "user-claim");
const res = await request(createApp(db))
.get("/actor")
.set("Authorization", `Bearer ${token}`)
.set("X-Paperclip-Run-Id", runId);
expect(res.status).toBe(200);
expect(res.body).toMatchObject({
type: "agent",
agentId,
companyId,
runId,
onBehalfOfUserId: "user-claim",
source: "agent_jwt",
});
});
it("preserves signed skill_test JWT scope on the request actor", async () => {
const companyId = randomUUID();
const agentId = randomUUID();
const runId = randomUUID();
const issueId = randomUUID();
const { db } = createDbState({
agent: { id: agentId, companyId },
run: { id: runId, companyId, agentId, responsibleUserId: "user-claim" },
});
const token = createLocalAgentJwt(agentId, companyId, "codex_local", runId, "user-claim", {
kind: "skill_test",
issueId,
});
const res = await request(createApp(db))
.get("/actor")
.set("Authorization", `Bearer ${token}`)
.set("X-Paperclip-Run-Id", runId);
expect(res.status).toBe(200);
expect(res.body).toMatchObject({
type: "agent",
agentId,
companyId,
keyScope: { kind: "skill_test", issueId },
source: "agent_jwt",
});
});
it("rejects mismatched run headers for agent JWTs and audits the spoof attempt", async () => {
const companyId = randomUUID();
const agentId = randomUUID();
const runId = randomUUID();
const spoofedRunId = randomUUID();
const { db, activity } = createDbState({
agent: { id: agentId, companyId },
run: { id: runId, companyId, agentId, responsibleUserId: "user-claim" },
});
const token = createLocalAgentJwt(agentId, companyId, "codex_local", runId, "user-claim");
const res = await request(createApp(db))
.get("/actor")
.set("Authorization", `Bearer ${token}`)
.set("X-Paperclip-Run-Id", spoofedRunId);
expect(res.status).toBe(422);
expect(res.body.code).toBe("agent_jwt_run_id_mismatch");
expect(activity).toHaveLength(1);
expect(activity[0]).toMatchObject({
companyId,
actorType: "agent",
actorId: agentId,
action: "auth.agent_jwt_run_header_mismatch",
entityType: "heartbeat_run",
entityId: runId,
runId,
details: { claimRunId: runId, headerRunId: spoofedRunId },
});
});
it("falls back to the run row responsible user for legacy claim-less agent JWTs", async () => {
const companyId = randomUUID();
const agentId = randomUUID();
const runId = randomUUID();
const { db } = createDbState({
agent: { id: agentId, companyId },
run: { id: runId, companyId, agentId, responsibleUserId: "user-legacy" },
});
const token = craftAgentJwtWithoutResponsibleClaim({
secret: process.env.PAPERCLIP_AGENT_JWT_SECRET!,
agentId,
companyId,
adapterType: "codex_local",
runId,
});
const res = await request(createApp(db))
.get("/actor")
.set("Authorization", `Bearer ${token}`);
expect(res.status).toBe(200);
expect(res.body).toMatchObject({
type: "agent",
runId,
onBehalfOfUserId: "user-legacy",
source: "agent_jwt",
});
});
it("rejects fork-minted run JWTs before issue reads or writes reach live issue data", async () => {
const companyId = randomUUID();
const agentId = randomUUID();
const runId = randomUUID();
const issueId = randomUUID();
const { db } = createDbState({
agent: { id: agentId, companyId },
run: { id: runId, companyId, agentId, responsibleUserId: "user-claim" },
});
process.env.PAPERCLIP_INSTANCE_ID = "pap-12899-worktree";
const forkToken = createLocalAgentJwt(agentId, companyId, "codex_local", runId, "user-claim");
expect(forkToken).not.toBeNull();
process.env.PAPERCLIP_INSTANCE_ID = "default";
const app = createApp(db);
const readRes = await request(app)
.get(`/companies/${companyId}/issues/${issueId}`)
.set("Authorization", `Bearer ${forkToken}`)
.set("X-Paperclip-Run-Id", runId);
const writeRes = await request(app)
.patch(`/companies/${companyId}/issues/${issueId}`)
.set("Authorization", `Bearer ${forkToken}`)
.set("X-Paperclip-Run-Id", runId)
.send({ title: "should not write" });
expect(readRes.status).toBe(401);
expect(writeRes.status).toBe(401);
});
it("populates agent-key actors from the key responsible user binding", async () => {
const companyId = randomUUID();
const agentId = randomUUID();
const token = "pcp_test_agent_key";
const { db } = createDbState({
agent: { id: agentId, companyId },
agentKey: {
id: randomUUID(),
agentId,
companyId,
keyHash: hashToken(token),
responsibleUserId: "user-key",
},
});
const res = await request(createApp(db))
.get("/actor")
.set("Authorization", `Bearer ${token}`);
expect(res.status).toBe(200);
expect(res.body).toMatchObject({
type: "agent",
agentId,
companyId,
onBehalfOfUserId: "user-key",
source: "agent_key",
});
});
it("rejects agent keys that lack a responsible user binding and audits the denial", async () => {
const companyId = randomUUID();
const agentId = randomUUID();
const keyId = randomUUID();
const token = "pcp_test_agent_key_without_user";
const { db, activity } = createDbState({
agent: { id: agentId, companyId },
agentKey: {
id: keyId,
agentId,
companyId,
keyHash: hashToken(token),
responsibleUserId: null,
},
});
const res = await request(createApp(db))
.get(`/companies/${companyId}/protected`)
.set("Authorization", `Bearer ${token}`);
expect(res.status).toBe(403);
expect(res.body.code).toBe("RESPONSIBLE_USER_UNAVAILABLE");
expect(activity).toHaveLength(1);
expect(activity[0]).toMatchObject({
companyId,
actorType: "agent",
actorId: agentId,
action: "auth.agent_key_missing_responsible_user",
entityType: "agent_api_key",
entityId: keyId,
details: { method: "GET", url: `/companies/${companyId}/protected` },
});
});
});