✔ sessionModel: agent vars win, then the system's execution settings (9.23967ms)
✔ a pi bundle: prompt, policy, tools and manifest, 0600 in a 0700 directory (4.656947ms)
✔ a claude-code bundle adds the wrapped gate hook and the MCP config (2.749835ms)
✔ the prompt's built-in tools line is the harness's own names, as the manifest lists them (8.9138ms)
✔ the prompt names the project clones in the workspace, and none when there are none (7.764195ms)
✔ the prompt explains the review loop to a role with push_working, and reviews to one with fetch_review (11.61429ms)
✔ a bundle is written once: an existing file refuses (1.598347ms)
✔ a path with a single quote can't go into the hook command (1.822379ms)
✔ allow exits 0, a deny exits 2 with the reason on stderr (114.321641ms)
✔ a missing or wrong policy, or a bad event, exits 2 (74.75146ms)
✔ the bundle's wrapped command: a missing gate or node still blocks (1088.649921ms)
✔ claude: typed tools through MCP, the hook blocks, builtins outside --tools don't exist (727.056511ms)
✔ claude: the hook alone blocks a path outside the workspace (444.005447ms)
✔ claude: a second turn resumes the first turn's session (733.995696ms)
✔ claude adapter: --restricted is always passed (5.150568ms)
✔ claude adapter: MOSAIC_CLAUDE_BIN wins over PATH and must be an absolute executable (15.597407ms)
✔ claude adapter: a directory pin exits 2 before it runs; a pin with a space runs (5.490688ms)
✔ claude: CLAUDE.md files and auto-memory don't reach the model; without --restricted they do (708.548281ms)
✔ claude: a missing hook or MCP file refuses before claude starts (7.305153ms)
✔ pi: policy tools and typed tools pass, anything else is blocked (4.113385ms)
✔ claude: builtins map from pi names, typed tools need the mcp prefix (0.957744ms)
✔ file tool paths must resolve inside the workspace (1.792419ms)
✔ pi's own path normalisation can't be used to step out (1.054324ms)
✔ a symlink inside the workspace that points out is outside (0.842073ms)
✔ a dangling symlink is refused at any depth, in both harnesses (2.985568ms)
✔ read is checked under every spelling pi's read would open, in both harnesses (15.561249ms)
✔ other spellings cover directories, dangling links and pi's cwd (3.065696ms)
✔ a relative path climbs from the workspace's real path, in both harnesses (4.920804ms)
✔ claude path fields per tool (0.819422ms)
✔ glob patterns stay inside the workspace (0.650068ms)
✔ a path that can't be checked is blocked (0.498297ms)
✔ fetch_review's have lists the working refs before origin's, and no more than HAVE_MAX (196.907236ms)
✔ fetch_review's have orders the working refs newest first within each clone, one clone after another (128.017375ms)
✔ a runner git past its time: SIGTERM reaches its process group, and SIGKILL follows after the grace period (1808.695365ms)
✔ once a signalled runner git has closed, its group gets SIGKILL, so a process that ignored SIGTERM dies too (304.059235ms)
✔ initialize, ping and tools/list (42.907079ms)
✔ tools/call goes through the tool socket; a refusal is an isError result (33.538444ms)
✔ replies are compact JSON; over the cap a tool error names the size and the cap (33.35114ms)
✔ unknown tools and methods are JSON-RPC errors and never reach the socket (29.819411ms)
✔ a missing argument is a usage error (26.727141ms)
✔ pi: typed tools reach the socket, the gate blocks, agent_end writes the marker (344.164678ms)
✔ pi: a reply over the cap is a tool error naming the size and the cap; list_tasks asks for open tasks (320.127151ms)
✔ pi: a write through a dangling symlink is blocked, and nothing appears outside (316.952633ms)
✔ pi: a read is refused when pi would open another spelling outside (306.989546ms)
✔ pi: a relative path climbs from the real path of a workspace behind a symlink (316.846185ms)
✔ pi: a relative path climbs from the real path of a dataRoot behind a symlink (302.800607ms)
✔ pi: a missing extension refuses before any model call (5.511976ms)
✔ pi: an extension without its configuration fails pi's start (246.951826ms)
✔ founderCheck: founder variables, then a needed service without a usable token (1.276872ms)
✔ turnRequest names the sender, class, reply and decision (0.21173ms)
✔ turnRequest says a tester's message is a test and never Jason's instruction (0.266439ms)
✔ turnRequest shows the message's own sent time, a day old as well as new (0.164593ms)
✔ stateBlock names the launch state with its time and event, and the open tasks with the cap (0.212003ms)
✔ a message becomes a turn, the answer goes back as a RESULT, SIGTERM releases and exits 0 (254.732765ms)
✔ launches revoked and then restored: the turn's block reads allowed and gives the restore time (266.81338ms)
✔ a task closed on the tracker is missing from the next turn's open list (219.484603ms)
✔ a message read a day after it was sent shows the time it was sent (170.602975ms)
✔ a failed state read fails the turn: no adapter runs, the sender hears why, the message is read (126.150055ms)
✔ a SIGTERM before the claim stops the runner with exit 0 and no claim (89.685582ms)
✔ typed tools carry the runner's capability; launch goes to the host's launch socket (411.218361ms)
✔ a RESULT gets no automatic reply; failed turns reply with the reason (1302.071296ms)
✔ SIGTERM during a turn kills the turn's process group and still exits 0 (143.545572ms)
✔ founder credentials stop before the claim (20) (174.686151ms)
✔ a refused claim exits 21; an ended run's capability exits 22 (196.894541ms)
✔ the launch ending under a running session exits 22 (148.983956ms)
✔ a broker that stays unreachable exits 23 after brokerRetries polls (248.075219ms)
✔ a broker that is down at the claim exits 23, not 21 (96.218352ms)
✔ no capability, or a malformed one, on stdin exits 2 (201.299274ms)
✔ a missing or malformed policy exits 2 before the claim (139.120672ms)
✔ the PM gets launch, its task verbs and the reads (9.876082ms)
✔ list_requirements reads the requirement ids, for a task.create holder with a tracker (7.282816ms)
✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (1.357888ms)
✔ a coder with a tracker gets its own task tools and no task_reassign (1.630208ms)
✔ launch only when the business's launch block names the instance as launcher (1.748203ms)
✔ an action outside the instance's authority has no tool (1.823946ms)
✔ the review loop: push_working and request_review for the coder, fetch_review and review_verdict for the reviewer, none for the PM (5.074969ms)
✔ callTool: one JSON line out, the result back, a refusal rejects (7.402439ms)
✔ list_tasks takes the broker's filters and pages; get_task takes a task_ref (1.965411ms)
✔ callTool sends list_tasks done false unless the session names done; other tools as given (2.143392ms)
✔ replyText: compact JSON up to the cap, an error naming the size and the cap over it (0.600679ms)
ℹ tests 80
ℹ suites 0
ℹ pass 80
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 11106.16061