✔ sessionModel: agent vars win, then the system's execution settings (15.458521ms)
✔ a pi bundle: prompt, policy, tools and manifest, 0600 in a 0700 directory (4.337803ms)
✔ a chief-of-staff bundle lists launch among the bus tools and the manifest's typed tools (2.301165ms)
✔ a chief-of-staff bundle lists end_run and change_organization as typed and nothing untyped, for either harness (5.034408ms)
✔ the stop line names end_run for a role.end holder, and every other role keeps today's text (6.761476ms)
✔ a claude-code bundle adds the wrapped gate hook and the MCP config (3.436434ms)
✔ the prompt's built-in tools line is the harness's own names, as the manifest lists them (9.728418ms)
✔ the prompt names the project clones in the workspace, and none when there are none (11.670151ms)
✔ the prompt explains the review loop to a role with push_working, and reviews to one with fetch_review (8.458745ms)
✔ a bundle is written once: an existing file refuses (1.388539ms)
✔ a path with a single quote can't go into the hook command (1.687168ms)
✔ web grant on pi: the bundle has web_search and web_fetch, and no built-in web tool (2.078269ms)
✔ no web grant on pi: no web tool in the bundle, and the gate blocks a web call (2.032554ms)
✔ web grant on claude-code: the bundle has web_search and web_fetch, and no built-in web tool (2.004108ms)
✔ no web grant on claude-code: no web tool in the bundle, and the gate blocks a web call (1.859927ms)
✔ a granted session's gate lets the typed web tools through and still blocks the built-ins (2.077072ms)
✔ search-source searxng-url: a granted role gets the typed broker tools and no harness web tool; an ungranted role gets nothing (4.390226ms)
✔ search-source searxng-bundled: a granted role gets the typed broker tools and no harness web tool; an ungranted role gets nothing (4.77936ms)
✔ search-source harness-builtin: claude-code gets WebSearch and WebFetch and no typed web tool, only when granted (7.362879ms)
✔ search-source harness-builtin is refused for a granted pi role, which has no built-in web tool (2.238612ms)
✔ allow exits 0, a deny exits 2 with the reason on stderr (146.387026ms)
✔ a missing or wrong policy, or a bad event, exits 2 (91.489891ms)
✔ the bundle's wrapped command: a missing gate or node still blocks (1090.700059ms)
✔ claude: typed tools through MCP, the hook blocks, builtins outside --tools don't exist (779.076339ms)
✔ claude: the hook alone blocks a path outside the workspace (438.807515ms)
✔ claude: a second turn resumes the first turn's session (767.888546ms)
✔ claude adapter: --restricted is always passed (5.027871ms)
✔ claude adapter: MOSAIC_CLAUDE_BIN wins over PATH and must be an absolute executable (15.910819ms)
✔ claude adapter: a directory pin exits 2 before it runs; a pin with a space runs (5.63641ms)
✔ claude: CLAUDE.md files and auto-memory don't reach the model; without --restricted they do (874.223745ms)
✔ claude: a missing hook or MCP file refuses before claude starts (8.846245ms)
✔ pi: policy tools and typed tools pass, anything else is blocked (3.337913ms)
✔ claude: builtins map from pi names, typed tools need the mcp prefix (0.751679ms)
✔ file tool paths must resolve inside the workspace (1.352644ms)
✔ pi's own path normalisation can't be used to step out (0.903451ms)
✔ a symlink inside the workspace that points out is outside (0.864906ms)
✔ a dangling symlink is refused at any depth, in both harnesses (4.082207ms)
✔ read is checked under every spelling pi's read would open, in both harnesses (17.149789ms)
✔ other spellings cover directories, dangling links and pi's cwd (1.833942ms)
✔ a relative path climbs from the workspace's real path, in both harnesses (4.19891ms)
✔ claude path fields per tool (0.919818ms)
✔ glob patterns stay inside the workspace (0.749633ms)
✔ a path that can't be checked is blocked (0.372993ms)
✔ fetch_review's have lists the working refs before origin's, and no more than HAVE_MAX (223.977327ms)
✔ fetch_review's have orders the working refs newest first within each clone, one clone after another (137.241227ms)
✔ a runner git past its time: SIGTERM reaches its process group, and SIGKILL follows after the grace period (1808.325105ms)
✔ once a signalled runner git has closed, its group gets SIGKILL, so a process that ignored SIGTERM dies too (304.41056ms)
✔ initialize, ping and tools/list (49.604922ms)
✔ tools/call goes through the tool socket; a refusal is an isError result (42.443376ms)
✔ replies are compact JSON; over the cap a tool error names the size and the cap (40.542276ms)
✔ unknown tools and methods are JSON-RPC errors and never reach the socket (29.668921ms)
✔ a missing argument is a usage error (28.316605ms)
✔ pi: typed tools reach the socket, the gate blocks, agent_end writes the marker (380.37814ms)
✔ pi: a reply over the cap is a tool error naming the size and the cap; list_tasks asks for open tasks (309.053636ms)
✔ pi: a write through a dangling symlink is blocked, and nothing appears outside (314.70051ms)
✔ pi: a read is refused when pi would open another spelling outside (308.593312ms)
✔ pi: a relative path climbs from the real path of a workspace behind a symlink (317.101316ms)
✔ pi: a relative path climbs from the real path of a dataRoot behind a symlink (312.982995ms)
✔ pi: a missing extension refuses before any model call (6.013207ms)
✔ pi: an extension without its configuration fails pi's start (257.365829ms)
✔ founderCheck: founder variables, then a needed service without a usable token (1.600081ms)
✔ turnRequest names the sender, class, reply and decision (0.303113ms)
✔ turnRequest says a tester's message is a test and never Jason's instruction (0.28957ms)
✔ turnRequest shows the message's own sent time, a day old as well as new (0.191283ms)
✔ stateBlock names the launch state with its time and event, and the open tasks with the cap (0.344892ms)
✔ a message becomes a turn, the answer goes back as a RESULT, SIGTERM releases and exits 0 (320.442502ms)
✔ launches revoked and then restored: the turn's block reads allowed and gives the restore time (259.049691ms)
✔ a task closed on the tracker is missing from the next turn's open list (236.612331ms)
✔ a message read a day after it was sent shows the time it was sent (175.910538ms)
✔ a decision the human resolves becomes a turn citing the decision and the answer (213.795351ms)
✔ a failed state read fails the turn: no adapter runs, the sender hears why, the message is read (156.077518ms)
✔ a SIGTERM before the claim stops the runner with exit 0 and no claim (101.402454ms)
✔ typed tools carry the runner's capability; launch goes to the host's launch socket (443.479086ms)
✔ end_run and change_organization reach the broker with the tool's arguments, and refusals keep their codes (470.970009ms)
✔ a RESULT gets no automatic reply; failed turns reply with the reason (1363.12064ms)
✔ the bus's launch notice is one turn with no reply, and is not taken again (250.642454ms)
✔ SIGTERM during a turn kills the turn's process group and still exits 0 (149.321568ms)
✔ founder credentials stop before the claim (20) (171.022493ms)
✔ a refused claim exits 21; an ended run's capability exits 22 (200.708848ms)
✔ the launch ending under a running session exits 22 (144.180725ms)
✔ a broker that stays unreachable exits 23 after brokerRetries polls (277.195885ms)
✔ a broker that is down at the claim exits 23, not 21 (95.982613ms)
✔ no capability, or a malformed one, on stdin exits 2 (197.595385ms)
✔ a missing or malformed policy exits 2 before the claim (127.26467ms)
✔ the PM gets its task verbs and the reads, and no launch (10.875432ms)
✔ the chief of staff gets launch, send_message, end_run and change_organization, and no task verbs (3.073372ms)
✔ end_run sends role and reason, no decision; change_organization sends revision and the organization keys (3.638296ms)
✔ the organization keys change_organization takes are the broker's ORGANIZATION_KEYS (0.177943ms)
✔ a role without role.end or organization.change gets neither tool, and each tool follows its own grant (11.389755ms)
✔ the assistant gets send_message, the reads and the web tools (it holds the web grant), no launch (1.96292ms)
✔ list_requirements reads the requirement ids, for a task.create holder with a tracker (3.204711ms)
✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (1.140932ms)
✔ a coder with a tracker gets its own task tools and no task_reassign (1.037012ms)
✔ launch only when the business's launch block names the instance as launcher (2.096763ms)
✔ an action outside the instance's authority has no tool (1.537378ms)
✔ the review loop: push_working and request_review for the coder, fetch_review and review_verdict for the reviewer, none for the PM (3.719353ms)
✔ callTool: one JSON line out, the result back, a refusal rejects (9.54809ms)
✔ list_tasks takes the broker's filters and pages; get_task takes a task_ref (2.331828ms)
✔ callTool sends list_tasks done false unless the session names done; other tools as given (3.014416ms)
✔ replyText: compact JSON up to the cap, an error naming the size and the cap over it (0.984143ms)
✔ web_search's query maxLength is the web layer's SEARCH_MAX_QUERY_CHARS (3.042694ms)
ℹ tests 101
ℹ suites 0
ℹ pass 101
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 12274.879471