✔ sessionModel: agent vars win, then the system's execution settings (9.416065ms)
✔ a pi bundle: prompt, policy, tools and manifest, 0600 in a 0700 directory (4.502321ms)
✔ a claude-code bundle adds the wrapped gate hook and the MCP config (3.49083ms)
✔ the prompt's built-in tools line is the harness's own names, as the manifest lists them (10.710252ms)
✔ the prompt names the project clones in the workspace, and none when there are none (8.739786ms)
✔ the prompt explains the review loop to a role with push_working, and reviews to one with fetch_review (6.799627ms)
✔ a bundle is written once: an existing file refuses (1.533388ms)
✔ a path with a single quote can't go into the hook command (2.085826ms)
✔ allow exits 0, a deny exits 2 with the reason on stderr (127.70801ms)
✔ a missing or wrong policy, or a bad event, exits 2 (90.923193ms)
✔ the bundle's wrapped command: a missing gate or node still blocks (1090.953417ms)
✔ claude: typed tools through MCP, the hook blocks, builtins outside --tools don't exist (743.763072ms)
✔ claude: the hook alone blocks a path outside the workspace (435.284216ms)
✔ claude: a second turn resumes the first turn's session (739.546078ms)
✔ claude adapter: --restricted is always passed (5.407259ms)
✔ claude adapter: MOSAIC_CLAUDE_BIN wins over PATH and must be an absolute executable (16.865695ms)
✔ claude adapter: a directory pin exits 2 before it runs; a pin with a space runs (5.566595ms)
✔ claude: CLAUDE.md files and auto-memory don't reach the model; without --restricted they do (699.061816ms)
✔ claude: a missing hook or MCP file refuses before claude starts (7.369295ms)
✔ pi: policy tools and typed tools pass, anything else is blocked (2.947475ms)
✔ claude: builtins map from pi names, typed tools need the mcp prefix (0.656184ms)
✔ file tool paths must resolve inside the workspace (1.309531ms)
✔ pi's own path normalisation can't be used to step out (0.934842ms)
✔ a symlink inside the workspace that points out is outside (0.772346ms)
✔ a dangling symlink is refused at any depth, in both harnesses (3.246135ms)
✔ read is checked under every spelling pi's read would open, in both harnesses (15.507954ms)
✔ other spellings cover directories, dangling links and pi's cwd (5.376084ms)
✔ a relative path climbs from the workspace's real path, in both harnesses (7.073045ms)
✔ claude path fields per tool (0.906873ms)
✔ glob patterns stay inside the workspace (0.646156ms)
✔ a path that can't be checked is blocked (0.36715ms)
✔ fetch_review's have lists the working refs before origin's, and no more than HAVE_MAX (216.111832ms)
✔ fetch_review's have orders the working refs newest first within each clone, one clone after another (136.219616ms)
✔ a runner git past its time: SIGTERM reaches its process group, and SIGKILL follows after the grace period (1809.540395ms)
✔ once a signalled runner git has closed, its group gets SIGKILL, so a process that ignored SIGTERM dies too (302.609643ms)
✔ initialize, ping and tools/list (48.266255ms)
✔ tools/call goes through the tool socket; a refusal is an isError result (34.008681ms)
✔ replies are compact JSON; over the cap a tool error names the size and the cap (40.656274ms)
✔ unknown tools and methods are JSON-RPC errors and never reach the socket (35.866741ms)
✔ a missing argument is a usage error (28.6614ms)
✔ pi: typed tools reach the socket, the gate blocks, agent_end writes the marker (376.614166ms)
✔ pi: a reply over the cap is a tool error naming the size and the cap; list_tasks asks for open tasks (312.588648ms)
✔ pi: a write through a dangling symlink is blocked, and nothing appears outside (316.270469ms)
✔ pi: a read is refused when pi would open another spelling outside (306.35043ms)
✔ pi: a relative path climbs from the real path of a workspace behind a symlink (311.156023ms)
✔ pi: a relative path climbs from the real path of a dataRoot behind a symlink (312.984608ms)
✔ pi: a missing extension refuses before any model call (5.547111ms)
✔ pi: an extension without its configuration fails pi's start (248.402336ms)
✔ founderCheck: founder variables, then a needed service without a usable token (1.330569ms)
✔ turnRequest names the sender, class, reply and decision (0.214773ms)
✔ turnRequest says a tester's message is a test and never Jason's instruction (0.246325ms)
✔ turnRequest shows the message's own sent time, a day old as well as new (0.154656ms)
✔ stateBlock names the launch state with its time and event, and the open tasks with the cap (0.254737ms)
✔ a message becomes a turn, the answer goes back as a RESULT, SIGTERM releases and exits 0 (262.974472ms)
✔ launches revoked and then restored: the turn's block reads allowed and gives the restore time (261.596103ms)
✔ a task closed on the tracker is missing from the next turn's open list (219.79141ms)
✔ a message read a day after it was sent shows the time it was sent (176.432453ms)
✔ a failed state read fails the turn: no adapter runs, the sender hears why, the message is read (123.619612ms)
✔ a SIGTERM before the claim stops the runner with exit 0 and no claim (98.11751ms)
✔ typed tools carry the runner's capability; launch goes to the host's launch socket (443.538714ms)
✔ a RESULT gets no automatic reply; failed turns reply with the reason (1300.329416ms)
✔ SIGTERM during a turn kills the turn's process group and still exits 0 (150.773417ms)
✔ founder credentials stop before the claim (20) (176.597102ms)
✔ a refused claim exits 21; an ended run's capability exits 22 (202.96215ms)
✔ the launch ending under a running session exits 22 (147.044567ms)
✔ a broker that stays unreachable exits 23 after brokerRetries polls (247.569741ms)
✔ a broker that is down at the claim exits 23, not 21 (89.654761ms)
✔ no capability, or a malformed one, on stdin exits 2 (193.655229ms)
✔ a missing or malformed policy exits 2 before the claim (121.192293ms)
✔ the PM gets launch, its task verbs and the reads (7.565464ms)
✔ list_requirements reads the requirement ids, for a task.create holder with a tracker (6.367376ms)
✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (1.402618ms)
✔ a coder with a tracker gets its own task tools and no task_reassign (1.246345ms)
✔ launch only when the business's launch block names the instance as launcher (1.307903ms)
✔ an action outside the instance's authority has no tool (1.08892ms)
✔ the review loop: push_working and request_review for the coder, fetch_review and review_verdict for the reviewer, none for the PM (4.29703ms)
✔ callTool: one JSON line out, the result back, a refusal rejects (7.82273ms)
✔ list_tasks takes the broker's filters and pages; get_task takes a task_ref (2.241939ms)
✔ callTool sends list_tasks done false unless the session names done; other tools as given (2.284692ms)
✔ replyText: compact JSON up to the cap, an error naming the size and the cap over it (0.625072ms)
ℹ tests 80
ℹ suites 0
ℹ pass 80
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 11165.016879