✔ sessionModel: agent vars win, then the system's execution settings (14.765749ms)
✔ a pi bundle: prompt, policy, tools and manifest, 0600 in a 0700 directory (5.731538ms)
✔ a chief-of-staff bundle lists launch among the bus tools and the manifest's typed tools (2.457181ms)
✔ a claude-code bundle adds the wrapped gate hook and the MCP config (2.403953ms)
✔ the prompt's built-in tools line is the harness's own names, as the manifest lists them (8.448232ms)
✔ the prompt names the project clones in the workspace, and none when there are none (6.323868ms)
✔ the prompt explains the review loop to a role with push_working, and reviews to one with fetch_review (8.992296ms)
✔ a bundle is written once: an existing file refuses (2.179129ms)
✔ a path with a single quote can't go into the hook command (2.05533ms)
✔ allow exits 0, a deny exits 2 with the reason on stderr (122.949032ms)
✔ a missing or wrong policy, or a bad event, exits 2 (82.270114ms)
✔ the bundle's wrapped command: a missing gate or node still blocks (1090.345033ms)
✔ claude: typed tools through MCP, the hook blocks, builtins outside --tools don't exist (780.922125ms)
✔ claude: the hook alone blocks a path outside the workspace (470.826308ms)
✔ claude: a second turn resumes the first turn's session (756.433206ms)
✔ claude adapter: --restricted is always passed (6.071969ms)
✔ claude adapter: MOSAIC_CLAUDE_BIN wins over PATH and must be an absolute executable (19.245536ms)
✔ claude adapter: a directory pin exits 2 before it runs; a pin with a space runs (7.19483ms)
✔ claude: CLAUDE.md files and auto-memory don't reach the model; without --restricted they do (728.360818ms)
✔ claude: a missing hook or MCP file refuses before claude starts (7.267484ms)
✔ pi: policy tools and typed tools pass, anything else is blocked (3.318107ms)
✔ claude: builtins map from pi names, typed tools need the mcp prefix (0.659734ms)
✔ file tool paths must resolve inside the workspace (1.068427ms)
✔ pi's own path normalisation can't be used to step out (0.885685ms)
✔ a symlink inside the workspace that points out is outside (0.821539ms)
✔ a dangling symlink is refused at any depth, in both harnesses (2.534074ms)
✔ read is checked under every spelling pi's read would open, in both harnesses (14.776276ms)
✔ other spellings cover directories, dangling links and pi's cwd (1.587952ms)
✔ a relative path climbs from the workspace's real path, in both harnesses (4.285242ms)
✔ claude path fields per tool (0.819818ms)
✔ glob patterns stay inside the workspace (0.805017ms)
✔ a path that can't be checked is blocked (0.524914ms)
✔ fetch_review's have lists the working refs before origin's, and no more than HAVE_MAX (222.698272ms)
✔ fetch_review's have orders the working refs newest first within each clone, one clone after another (133.247163ms)
✔ a runner git past its time: SIGTERM reaches its process group, and SIGKILL follows after the grace period (1809.941407ms)
✔ once a signalled runner git has closed, its group gets SIGKILL, so a process that ignored SIGTERM dies too (304.233894ms)
✔ initialize, ping and tools/list (47.535777ms)
✔ tools/call goes through the tool socket; a refusal is an isError result (34.256518ms)
✔ replies are compact JSON; over the cap a tool error names the size and the cap (38.269859ms)
✔ unknown tools and methods are JSON-RPC errors and never reach the socket (30.336371ms)
✔ a missing argument is a usage error (28.004897ms)
✔ pi: typed tools reach the socket, the gate blocks, agent_end writes the marker (380.895506ms)
✔ pi: a reply over the cap is a tool error naming the size and the cap; list_tasks asks for open tasks (351.643182ms)
✔ pi: a write through a dangling symlink is blocked, and nothing appears outside (388.300224ms)
✔ pi: a read is refused when pi would open another spelling outside (335.479106ms)
✔ pi: a relative path climbs from the real path of a workspace behind a symlink (345.378553ms)
✔ pi: a relative path climbs from the real path of a dataRoot behind a symlink (342.28412ms)
✔ pi: a missing extension refuses before any model call (6.179747ms)
✔ pi: an extension without its configuration fails pi's start (257.559933ms)
✔ founderCheck: founder variables, then a needed service without a usable token (1.439894ms)
✔ turnRequest names the sender, class, reply and decision (0.191121ms)
✔ turnRequest says a tester's message is a test and never Jason's instruction (0.195149ms)
✔ turnRequest shows the message's own sent time, a day old as well as new (0.123592ms)
✔ stateBlock names the launch state with its time and event, and the open tasks with the cap (0.230878ms)
✔ a message becomes a turn, the answer goes back as a RESULT, SIGTERM releases and exits 0 (184.993444ms)
✔ launches revoked and then restored: the turn's block reads allowed and gives the restore time (196.150497ms)
✔ a task closed on the tracker is missing from the next turn's open list (195.117059ms)
✔ a message read a day after it was sent shows the time it was sent (114.595794ms)
✔ a decision the human resolves becomes a turn citing the decision and the answer (140.337806ms)
✔ a failed state read fails the turn: no adapter runs, the sender hears why, the message is read (64.373497ms)
✔ a SIGTERM before the claim stops the runner with exit 0 and no claim (65.06167ms)
✔ typed tools carry the runner's capability; launch goes to the host's launch socket (275.816021ms)
✔ a RESULT gets no automatic reply; failed turns reply with the reason (1189.5088ms)
✔ SIGTERM during a turn kills the turn's process group and still exits 0 (92.973134ms)
✔ founder credentials stop before the claim (20) (98.36162ms)
✔ a refused claim exits 21; an ended run's capability exits 22 (93.909697ms)
✔ the launch ending under a running session exits 22 (98.032738ms)
✔ a broker that stays unreachable exits 23 after brokerRetries polls (198.928496ms)
✔ a broker that is down at the claim exits 23, not 21 (47.669801ms)
✔ no capability, or a malformed one, on stdin exits 2 (148.954023ms)
✔ a missing or malformed policy exits 2 before the claim (81.932002ms)
✔ the PM gets its task verbs and the reads, and no launch (7.990909ms)
✔ the chief of staff gets launch and send_message, and no task verbs (2.911115ms)
✔ the assistant gets send_message and the reads, no launch (3.231081ms)
✔ list_requirements reads the requirement ids, for a task.create holder with a tracker (4.537834ms)
✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (1.479233ms)
✔ a coder with a tracker gets its own task tools and no task_reassign (1.217942ms)
✔ launch only when the business's launch block names the instance as launcher (2.412717ms)
✔ an action outside the instance's authority has no tool (1.724385ms)
✔ the review loop: push_working and request_review for the coder, fetch_review and review_verdict for the reviewer, none for the PM (4.429008ms)
✔ callTool: one JSON line out, the result back, a refusal rejects (8.564741ms)
✔ list_tasks takes the broker's filters and pages; get_task takes a task_ref (2.707563ms)
✔ callTool sends list_tasks done false unless the session names done; other tools as given (2.671271ms)
✔ replyText: compact JSON up to the cap, an error naming the size and the cap over it (0.839773ms)
ℹ tests 84
ℹ suites 0
ℹ pass 84
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 10652.286247