✔ sessionModel: agent vars win, then the system's execution settings (13.173865ms)
✔ a pi bundle: prompt, policy, tools and manifest, 0600 in a 0700 directory (4.362868ms)
✔ a chief-of-staff bundle lists launch among the bus tools and the manifest's typed tools (2.433681ms)
✔ a chief-of-staff bundle lists end_run and change_organization as typed and nothing untyped, for either harness (4.511214ms)
✔ the stop line names end_run for a role.end holder, and every other role keeps today's text (7.829306ms)
✔ a claude-code bundle adds the wrapped gate hook and the MCP config (3.079409ms)
✔ the prompt's built-in tools line is the harness's own names, as the manifest lists them (10.83531ms)
✔ the prompt names the project clones in the workspace, and none when there are none (8.374055ms)
✔ the prompt explains the review loop to a role with push_working, and reviews to one with fetch_review (6.768247ms)
✔ the prompt explains landing to a role with land_work, and to no other (3.630701ms)
✔ the prompt tells the pm to ask the chief of staff for sessions, and no other role (6.530323ms)
✔ a bundle is written once: an existing file refuses (1.783932ms)
✔ a path with a single quote can't go into the hook command (1.61449ms)
✔ web grant on pi: the bundle has web_search and web_fetch, and no built-in web tool (2.192957ms)
✔ no web grant on pi: no web tool in the bundle, and the gate blocks a web call (1.718222ms)
✔ web grant on claude-code: the bundle has web_search and web_fetch, and no built-in web tool (1.604794ms)
✔ no web grant on claude-code: no web tool in the bundle, and the gate blocks a web call (1.508416ms)
✔ a granted session's gate lets the typed web tools through and still blocks the built-ins (1.505809ms)
✔ search-source searxng-url: a granted role gets the typed broker tools and no harness web tool; an ungranted role gets nothing (3.935386ms)
✔ search-source searxng-bundled: a granted role gets the typed broker tools and no harness web tool; an ungranted role gets nothing (5.051258ms)
✔ search-source harness-builtin: claude-code gets WebSearch and WebFetch and no typed web tool, only when granted (8.661524ms)
✔ search-source harness-builtin is refused for a granted pi role, which has no built-in web tool (2.382922ms)
✔ allow exits 0, a deny exits 2 with the reason on stderr (133.679192ms)
✔ a missing or wrong policy, or a bad event, exits 2 (81.887907ms)
✔ the bundle's wrapped command: a missing gate or node still blocks (1084.410478ms)
✔ claude: typed tools through MCP, the hook blocks, builtins outside --tools don't exist (760.810147ms)
✔ claude: the hook alone blocks a path outside the workspace (415.837764ms)
✔ claude: a second turn resumes the first turn's session (690.007885ms)
✔ claude adapter: --restricted is always passed (5.130544ms)
✔ claude adapter: MOSAIC_CLAUDE_BIN wins over PATH and must be an absolute executable (19.034388ms)
✔ claude adapter: a directory pin exits 2 before it runs; a pin with a space runs (5.642229ms)
✔ claude: CLAUDE.md files and auto-memory don't reach the model; without --restricted they do (702.821836ms)
✔ claude: a missing hook or MCP file refuses before claude starts (7.306544ms)
✔ pi: policy tools and typed tools pass, anything else is blocked (4.306138ms)
✔ claude: builtins map from pi names, typed tools need the mcp prefix (0.937613ms)
✔ file tool paths must resolve inside the workspace (1.615938ms)
✔ pi's own path normalisation can't be used to step out (1.001802ms)
✔ a symlink inside the workspace that points out is outside (0.971995ms)
✔ a dangling symlink is refused at any depth, in both harnesses (2.848779ms)
✔ read is checked under every spelling pi's read would open, in both harnesses (15.539603ms)
✔ other spellings cover directories, dangling links and pi's cwd (1.416481ms)
✔ a relative path climbs from the workspace's real path, in both harnesses (4.044539ms)
✔ claude path fields per tool (0.632557ms)
✔ glob patterns stay inside the workspace (0.580051ms)
✔ a path that can't be checked is blocked (0.315965ms)
✔ fetch_review's have lists the working refs before origin's, and no more than HAVE_MAX (228.583374ms)
✔ fetch_review's have orders the working refs newest first within each clone, one clone after another (140.901189ms)
✔ a runner git past its time: SIGTERM reaches its process group, and SIGKILL follows after the grace period (1809.03326ms)
✔ once a signalled runner git has closed, its group gets SIGKILL, so a process that ignored SIGTERM dies too (304.02353ms)
✔ land_work sends one land line and resolves the host's started answer (3.393115ms)
✔ initialize, ping and tools/list (47.362017ms)
✔ tools/call goes through the tool socket; a refusal is an isError result (38.927171ms)
✔ replies are compact JSON; over the cap a tool error names the size and the cap (55.966719ms)
✔ unknown tools and methods are JSON-RPC errors and never reach the socket (32.595522ms)
✔ a missing argument is a usage error (26.544884ms)
✔ pi: typed tools reach the socket, the gate blocks, agent_end writes the marker (381.96539ms)
✔ pi: a reply over the cap is a tool error naming the size and the cap; list_tasks asks for open tasks (316.804917ms)
✔ pi: a write through a dangling symlink is blocked, and nothing appears outside (319.350422ms)
✔ pi: a read is refused when pi would open another spelling outside (319.750244ms)
✔ pi: a relative path climbs from the real path of a workspace behind a symlink (300.372624ms)
✔ pi: a relative path climbs from the real path of a dataRoot behind a symlink (299.338608ms)
✔ pi: a missing extension refuses before any model call (6.173118ms)
✔ pi: an extension without its configuration fails pi's start (243.746931ms)
✔ founderCheck: founder variables, then a needed service without a usable token (1.495949ms)
✔ turnRequest names the sender, class, reply and decision (0.28872ms)
✔ turnRequest says a tester's message is a test and never Jason's instruction (0.230429ms)
✔ turnRequest shows the message's own sent time, a day old as well as new (0.132043ms)
✔ stateBlock names the launch state with its time and event, and the open tasks with the cap (0.247737ms)
✔ a message becomes a turn, the answer goes back as a RESULT, SIGTERM releases and exits 0 (188.7137ms)
✔ launches revoked and then restored: the turn's block reads allowed and gives the restore time (200.463126ms)
✔ a task closed on the tracker is missing from the next turn's open list (171.281054ms)
✔ a message read a day after it was sent shows the time it was sent (89.794043ms)
✔ a decision the human resolves becomes a turn citing the decision and the answer (141.499162ms)
✔ a failed state read fails the turn: no adapter runs, the sender hears why, the message is read (63.877438ms)
✔ a SIGTERM before the claim stops the runner with exit 0 and no claim (53.620453ms)
✔ typed tools carry the runner's capability; launch goes to the host's launch socket (300.032322ms)
✔ end_run and change_organization reach the broker with the tool's arguments, and refusals keep their codes (258.019818ms)
✔ a RESULT gets no automatic reply; failed turns reply with the reason (1197.975633ms)
✔ the bus's launch notice is one turn with no reply, and is not taken again (217.388112ms)
✔ SIGTERM during a turn kills the turn's process group and still exits 0 (95.466762ms)
✔ founder credentials stop before the claim (20) (92.538834ms)
✔ a refused claim exits 21; an ended run's capability exits 22 (99.019824ms)
✔ the launch ending under a running session exits 22 (99.797871ms)
✔ a broker that stays unreachable exits 23 after brokerRetries polls (207.572046ms)
✔ a broker that is down at the claim exits 23, not 21 (46.238649ms)
✔ no capability, or a malformed one, on stdin exits 2 (148.342958ms)
✔ a missing or malformed policy exits 2 before the claim (81.435155ms)
✔ every role sees read_spec, record writer follows its grant, citation schemas stay strict (15.06316ms)
✔ typed socket forwards read content/history and agent record writes, and cannot write a spec (100.966365ms)
✔ the PM gets its task verbs and the reads, and no launch (10.469566ms)
✔ the chief of staff gets launch, send_message, end_run and change_organization, and no task verbs (3.361333ms)
✔ request_staff is the pm's alone: staff.request over message.send, roles and counts bounded (12.169031ms)
✔ end_run sends role and reason, no decision; change_organization sends revision and the organization keys (1.827973ms)
✔ the organization keys change_organization takes are the broker's ORGANIZATION_KEYS (0.130076ms)
✔ a role without role.end or organization.change gets neither tool, and each tool follows its own grant (7.716749ms)
✔ the assistant gets send_message, the reads and the web tools (it holds the web grant), no launch (1.316016ms)
✔ list_requirements reads the requirement ids, for a task.create holder with a tracker (2.543133ms)
✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (1.257805ms)
✔ a coder with a tracker gets its own task tools and no task_reassign (1.061621ms)
✔ launch only when the business's launch block names the instance as launcher (2.136484ms)
✔ an action outside the instance's authority has no tool (1.620511ms)
✔ the review loop: push_working and request_review for the coder, fetch_review and review_verdict for the reviewer, none for the PM (4.669655ms)
✔ land_work is there only with land, after push_working, with the task and the decision (2.687277ms)
✔ callTool: one JSON line out, the result back, a refusal rejects (10.860682ms)
✔ list_tasks takes the broker's filters and pages; get_task takes a task_ref (2.847715ms)
✔ callTool sends list_tasks done false unless the session names done; other tools as given (3.070228ms)
✔ replyText: compact JSON up to the cap, an error naming the size and the cap over it (0.858149ms)
✔ web_search's query maxLength is the web layer's SEARCH_MAX_QUERY_CHARS (2.865374ms)
ℹ tests 108
ℹ suites 0
ℹ pass 108
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 11120.140824
