✔ sessionModel: agent vars win, then the system's execution settings (12.198286ms)
✔ a pi bundle: prompt, policy, tools and manifest, 0600 in a 0700 directory (5.9262ms)
✔ a claude-code bundle adds the wrapped gate hook and the MCP config (3.233033ms)
✔ the prompt's built-in tools line is the harness's own names, as the manifest lists them (12.299329ms)
✔ the prompt names the project clones in the workspace, and none when there are none (5.575526ms)
✔ the prompt explains the review loop to a role with push_working, and reviews to one with fetch_review (6.130564ms)
✔ a bundle is written once: an existing file refuses (1.900136ms)
✔ a path with a single quote can't go into the hook command (1.385972ms)
✔ allow exits 0, a deny exits 2 with the reason on stderr (119.245875ms)
✔ a missing or wrong policy, or a bad event, exits 2 (87.771877ms)
✔ the bundle's wrapped command: a missing gate or node still blocks (1084.247453ms)
✔ claude: typed tools through MCP, the hook blocks, builtins outside --tools don't exist (726.496982ms)
✔ claude: the hook alone blocks a path outside the workspace (447.913555ms)
✔ claude: a second turn resumes the first turn's session (753.0938ms)
✔ claude adapter: --restricted is always passed (5.510195ms)
✔ claude adapter: MOSAIC_CLAUDE_BIN wins over PATH and must be an absolute executable (16.126711ms)
✔ claude adapter: a directory pin exits 2 before it runs; a pin with a space runs (5.825807ms)
✔ claude: CLAUDE.md files and auto-memory don't reach the model; without --restricted they do (699.801578ms)
✔ claude: a missing hook or MCP file refuses before claude starts (7.334068ms)
✔ pi: policy tools and typed tools pass, anything else is blocked (3.190218ms)
✔ claude: builtins map from pi names, typed tools need the mcp prefix (0.639109ms)
✔ file tool paths must resolve inside the workspace (1.269192ms)
✔ pi's own path normalisation can't be used to step out (0.950328ms)
✔ a symlink inside the workspace that points out is outside (0.90052ms)
✔ a dangling symlink is refused at any depth, in both harnesses (3.05701ms)
✔ read is checked under every spelling pi's read would open, in both harnesses (19.847482ms)
✔ other spellings cover directories, dangling links and pi's cwd (1.297455ms)
✔ a relative path climbs from the workspace's real path, in both harnesses (3.96552ms)
✔ claude path fields per tool (0.62227ms)
✔ glob patterns stay inside the workspace (0.580081ms)
✔ a path that can't be checked is blocked (0.378827ms)
✔ fetch_review's have lists the working refs before origin's, and no more than HAVE_MAX (210.844167ms)
✔ fetch_review's have orders the working refs newest first within each clone, one clone after another (136.850352ms)
✔ a runner git past its time: SIGTERM reaches its process group, and SIGKILL follows after the grace period (1809.403034ms)
✔ once a signalled runner git has closed, its group gets SIGKILL, so a process that ignored SIGTERM dies too (304.01072ms)
✔ initialize, ping and tools/list (50.777612ms)
✔ tools/call goes through the tool socket; a refusal is an isError result (35.56347ms)
✔ replies are compact JSON; over the cap a tool error names the size and the cap (35.753948ms)
✔ unknown tools and methods are JSON-RPC errors and never reach the socket (28.550769ms)
✔ a missing argument is a usage error (24.967442ms)
✔ pi: typed tools reach the socket, the gate blocks, agent_end writes the marker (372.567727ms)
✔ pi: a reply over the cap is a tool error naming the size and the cap; list_tasks asks for open tasks (308.186932ms)
✔ pi: a write through a dangling symlink is blocked, and nothing appears outside (320.494239ms)
✔ pi: a read is refused when pi would open another spelling outside (322.127336ms)
✔ pi: a relative path climbs from the real path of a workspace behind a symlink (310.997252ms)
✔ pi: a relative path climbs from the real path of a dataRoot behind a symlink (319.773223ms)
✔ pi: a missing extension refuses before any model call (5.534674ms)
✔ pi: an extension without its configuration fails pi's start (246.338405ms)
✔ founderCheck: founder variables, then a needed service without a usable token (1.396986ms)
✔ turnRequest names the sender, class, reply and decision (0.285332ms)
✔ turnRequest says a tester's message is a test and never Jason's instruction (0.292195ms)
✔ turnRequest shows the message's own sent time, a day old as well as new (0.188677ms)
✔ stateBlock names the launch state with its time and event, and the open tasks with the cap (0.284001ms)
✔ a message becomes a turn, the answer goes back as a RESULT, SIGTERM releases and exits 0 (262.059437ms)
✔ launches revoked and then restored: the turn's block reads allowed and gives the restore time (262.028564ms)
✔ a task closed on the tracker is missing from the next turn's open list (219.310707ms)
✔ a message read a day after it was sent shows the time it was sent (174.757393ms)
✔ a failed state read fails the turn: no adapter runs, the sender hears why, the message is read (123.297687ms)
✔ a SIGTERM before the claim stops the runner with exit 0 and no claim (93.947461ms)
✔ typed tools carry the runner's capability; launch goes to the host's launch socket (415.185392ms)
✔ a RESULT gets no automatic reply; failed turns reply with the reason (1291.001854ms)
✔ SIGTERM during a turn kills the turn's process group and still exits 0 (153.656166ms)
✔ founder credentials stop before the claim (20) (171.491672ms)
✔ a refused claim exits 21; an ended run's capability exits 22 (191.085969ms)
✔ the launch ending under a running session exits 22 (144.433177ms)
✔ a broker that stays unreachable exits 23 after brokerRetries polls (246.032494ms)
✔ a broker that is down at the claim exits 23, not 21 (94.490638ms)
✔ no capability, or a malformed one, on stdin exits 2 (183.289581ms)
✔ a missing or malformed policy exits 2 before the claim (149.366585ms)
✔ the PM gets launch, its task verbs and the reads (7.62434ms)
✔ list_requirements reads the requirement ids, for a task.create holder with a tracker (3.754964ms)
✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (2.214337ms)
✔ a coder with a tracker gets its own task tools and no task_reassign (4.409604ms)
✔ launch only when the business's launch block names the instance as launcher (1.873535ms)
✔ an action outside the instance's authority has no tool (1.992731ms)
✔ the review loop: push_working and request_review for the coder, fetch_review and review_verdict for the reviewer, none for the PM (4.663246ms)
✔ callTool: one JSON line out, the result back, a refusal rejects (7.938596ms)
✔ list_tasks takes the broker's filters and pages; get_task takes a task_ref (1.989143ms)
✔ callTool sends list_tasks done false unless the session names done; other tools as given (2.532793ms)
✔ replyText: compact JSON up to the cap, an error naming the size and the cap over it (0.782873ms)
ℹ tests 80
ℹ suites 0
ℹ pass 80
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 11124.299046