✔ sessionModel: agent vars win, then the system's execution settings (10.504109ms)
✔ a pi bundle: prompt, policy, tools and manifest, 0600 in a 0700 directory (4.913069ms)
✔ a claude-code bundle adds the wrapped gate hook and the MCP config (3.176849ms)
✔ the prompt's built-in tools line is the harness's own names, as the manifest lists them (12.178766ms)
✔ the prompt names the project clones in the workspace, and none when there are none (5.985201ms)
✔ the prompt explains the review loop to a role with push_working, and reviews to one with fetch_review (8.52448ms)
✔ a bundle is written once: an existing file refuses (1.839554ms)
✔ a path with a single quote can't go into the hook command (1.880907ms)
✔ allow exits 0, a deny exits 2 with the reason on stderr (138.899415ms)
✔ a missing or wrong policy, or a bad event, exits 2 (91.278048ms)
✔ the bundle's wrapped command: a missing gate or node still blocks (1104.864274ms)
✔ claude: typed tools through MCP, the hook blocks, builtins outside --tools don't exist (747.230313ms)
✔ claude: the hook alone blocks a path outside the workspace (394.5154ms)
✔ claude: a second turn resumes the first turn's session (707.563673ms)
✔ claude adapter: --restricted is always passed (5.186259ms)
✔ claude adapter: MOSAIC_CLAUDE_BIN wins over PATH and must be an absolute executable (16.440213ms)
✔ claude adapter: a directory pin exits 2 before it runs; a pin with a space runs (6.08174ms)
✔ claude: CLAUDE.md files and auto-memory don't reach the model; without --restricted they do (699.136963ms)
✔ claude: a missing hook or MCP file refuses before claude starts (7.011286ms)
✔ pi: policy tools and typed tools pass, anything else is blocked (3.385718ms)
✔ claude: builtins map from pi names, typed tools need the mcp prefix (0.720297ms)
✔ file tool paths must resolve inside the workspace (1.26506ms)
✔ pi's own path normalisation can't be used to step out (1.352752ms)
✔ a symlink inside the workspace that points out is outside (0.896351ms)
✔ a dangling symlink is refused at any depth, in both harnesses (3.230259ms)
✔ read is checked under every spelling pi's read would open, in both harnesses (19.814115ms)
✔ other spellings cover directories, dangling links and pi's cwd (2.268784ms)
✔ a relative path climbs from the workspace's real path, in both harnesses (4.894414ms)
✔ claude path fields per tool (0.975998ms)
✔ glob patterns stay inside the workspace (0.943241ms)
✔ a path that can't be checked is blocked (0.555195ms)
✔ fetch_review's have lists the working refs before origin's, and no more than HAVE_MAX (244.147082ms)
✔ initialize, ping and tools/list (54.846332ms)
✔ tools/call goes through the tool socket; a refusal is an isError result (36.818735ms)
✔ replies are compact JSON; over the cap a tool error names the size and the cap (39.421704ms)
✔ unknown tools and methods are JSON-RPC errors and never reach the socket (33.680808ms)
✔ a missing argument is a usage error (31.263414ms)
✔ pi: typed tools reach the socket, the gate blocks, agent_end writes the marker (413.14455ms)
✔ pi: a reply over the cap is a tool error naming the size and the cap; list_tasks asks for open tasks (334.8523ms)
✔ pi: a write through a dangling symlink is blocked, and nothing appears outside (304.034653ms)
✔ pi: a read is refused when pi would open another spelling outside (328.8915ms)
✔ pi: a relative path climbs from the real path of a workspace behind a symlink (306.32624ms)
✔ pi: a relative path climbs from the real path of a dataRoot behind a symlink (319.36229ms)
✔ pi: a missing extension refuses before any model call (6.356845ms)
✔ pi: an extension without its configuration fails pi's start (257.293764ms)
✔ founderCheck: founder variables, then a needed service without a usable token (1.627816ms)
✔ turnRequest names the sender, class, reply and decision (0.316014ms)
✔ a message becomes a turn, the answer goes back as a RESULT, SIGTERM releases and exits 0 (190.648382ms)
✔ a SIGTERM before the claim stops the runner with exit 0 and no claim (71.88356ms)
✔ typed tools carry the runner's capability; launch goes to the host's launch socket (301.996015ms)
✔ a RESULT gets no automatic reply; failed turns reply with the reason (1188.382793ms)
✔ SIGTERM during a turn kills the turn's process group and still exits 0 (95.218837ms)
✔ founder credentials stop before the claim (20) (91.468593ms)
✔ a refused claim exits 21; an ended run's capability exits 22 (96.325132ms)
✔ the launch ending under a running session exits 22 (100.532628ms)
✔ a broker that stays unreachable exits 23 after brokerRetries polls (207.402302ms)
✔ a broker that is down at the claim exits 23, not 21 (47.621195ms)
✔ no capability, or a malformed one, on stdin exits 2 (155.894511ms)
✔ a missing or malformed policy exits 2 before the claim (86.154824ms)
✔ the PM gets launch, its task verbs and the reads (9.457986ms)
✔ list_requirements reads the requirement ids, for a task.create holder with a tracker (7.006404ms)
✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (2.344177ms)
✔ a coder with a tracker gets its own task tools and no task_reassign (2.282287ms)
✔ launch only when the business's launch block names the instance as launcher (1.539999ms)
✔ an action outside the instance's authority has no tool (2.045868ms)
✔ the review loop: push_working and request_review for the coder, fetch_review and review_verdict for the reviewer, none for the PM (7.130472ms)
✔ callTool: one JSON line out, the result back, a refusal rejects (10.296278ms)
✔ list_tasks takes the broker's filters and pages; get_task takes a task_ref (3.526607ms)
✔ callTool sends list_tasks done false unless the session names done; other tools as given (2.955981ms)
✔ replyText: compact JSON up to the cap, an error naming the size and the cap over it (0.766571ms)
ℹ tests 70
ℹ suites 0
ℹ pass 70
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 9985.230597
