✔ sessionModel: agent vars win, then the system's execution settings (8.497804ms) ✔ a pi bundle: prompt, policy, tools and manifest, 0600 in a 0700 directory (3.417173ms) ✔ a chief-of-staff bundle lists launch among the bus tools and the manifest's typed tools (1.763837ms) ✔ a chief-of-staff bundle lists end_run and change_organization as typed and only instance.name untyped, for either harness (3.287978ms) ✔ the stop line names end_run for a role.end holder, and every other role keeps today's text (4.583317ms) ✔ a claude-code bundle adds the wrapped gate hook and the MCP config (1.579039ms) ✔ the prompt's built-in tools line is the harness's own names, as the manifest lists them (5.34955ms) ✔ the prompt names the project clones in the workspace, and none when there are none (4.448217ms) ✔ the prompt explains the review loop to a role with push_working, and reviews to one with fetch_review (4.456245ms) ✔ the prompt explains landing to a role with land_work, and to no other (2.797652ms) ✔ the prompt tells the pm to ask the chief of staff for sessions, and no other role (5.194408ms) ✔ a bundle is written once: an existing file refuses (1.037176ms) ✔ a path with a single quote can't go into the hook command (1.132466ms) ✔ web grant on pi: the bundle has web_search and web_fetch, and no built-in web tool (1.481846ms) ✔ no web grant on pi: no web tool in the bundle, and the gate blocks a web call (1.396399ms) ✔ web grant on claude-code: the bundle has web_search and web_fetch, and no built-in web tool (1.184086ms) ✔ no web grant on claude-code: no web tool in the bundle, and the gate blocks a web call (1.122596ms) ✔ a granted session's gate lets the typed web tools through and still blocks the built-ins (1.178363ms) ✔ search-source searxng-url: a granted role gets the typed broker tools and no harness web tool; an ungranted role gets nothing (3.109179ms) ✔ search-source searxng-bundled: a granted role gets the typed broker tools and no harness web tool; an ungranted role gets nothing (3.469524ms) ✔ search-source harness-builtin: claude-code gets WebSearch and WebFetch and no typed web tool, only when granted (4.507292ms) ✔ search-source harness-builtin is refused for a granted pi role, which has no built-in web tool (1.29107ms) ✔ allow exits 0, a deny exits 2 with the reason on stderr (98.070008ms) ✔ a missing or wrong policy, or a bad event, exits 2 (73.482746ms) ✔ the bundle's wrapped command: a missing gate or node still blocks (1080.530235ms) ✔ claude: typed tools through MCP, the hook blocks, builtins outside --tools don't exist (666.277172ms) ✔ claude: the hook alone blocks a path outside the workspace (388.509587ms) ✔ claude: a second turn resumes the first turn's session (639.188765ms) ✔ claude adapter: --restricted is always passed (5.11313ms) ✔ claude adapter: MOSAIC_CLAUDE_BIN wins over PATH and must be an absolute executable (15.486519ms) ✔ claude adapter: a directory pin exits 2 before it runs; a pin with a space runs (5.471386ms) ✔ claude: CLAUDE.md files and auto-memory don't reach the model; without --restricted they do (657.530212ms) ✔ claude: a missing hook or MCP file refuses before claude starts (7.663476ms) ✔ pi: policy tools and typed tools pass, anything else is blocked (2.1829ms) ✔ claude: builtins map from pi names, typed tools need the mcp prefix (0.496823ms) ✔ file tool paths must resolve inside the workspace (0.876109ms) ✔ pi's own path normalisation can't be used to step out (0.709139ms) ✔ a symlink inside the workspace that points out is outside (0.61403ms) ✔ a dangling symlink is refused at any depth, in both harnesses (2.491013ms) ✔ read is checked under every spelling pi's read would open, in both harnesses (9.336162ms) ✔ other spellings cover directories, dangling links and pi's cwd (0.987522ms) ✔ a relative path climbs from the workspace's real path, in both harnesses (3.147116ms) ✔ claude path fields per tool (0.530685ms) ✔ glob patterns stay inside the workspace (0.497403ms) ✔ a path that can't be checked is blocked (0.312816ms) ✔ fetch_review's have lists the working refs before origin's, and no more than HAVE_MAX (155.646688ms) ✔ fetch_review's have orders the working refs newest first within each clone, one clone after another (109.455163ms) ✔ a runner git past its time: SIGTERM reaches its process group, and SIGKILL follows after the grace period (1809.175968ms) ✔ once a signalled runner git has closed, its group gets SIGKILL, so a process that ignored SIGTERM dies too (304.500621ms) ✔ land_work sends one land line and resolves the host's started answer (3.051481ms) ✔ initialize, ping and tools/list (32.094104ms) ✔ tools/call goes through the tool socket; a refusal is an isError result (28.948582ms) ✔ replies are compact JSON; over the cap a tool error names the size and the cap (33.813612ms) ✔ unknown tools and methods are JSON-RPC errors and never reach the socket (26.211122ms) ✔ a missing argument is a usage error (24.200126ms) ✔ pi: typed tools reach the socket, the gate blocks, agent_end writes the marker (321.304443ms) ✔ pi: a reply over the cap is a tool error naming the size and the cap; list_tasks asks for open tasks (349.737455ms) ✔ pi: a write through a dangling symlink is blocked, and nothing appears outside (352.227811ms) ✔ pi: a read is refused when pi would open another spelling outside (329.950148ms) ✔ pi: a relative path climbs from the real path of a workspace behind a symlink (292.714031ms) ✔ pi: a relative path climbs from the real path of a dataRoot behind a symlink (299.810831ms) ✔ pi: a missing extension refuses before any model call (6.019645ms) ✔ pi: an extension without its configuration fails pi's start (400.610803ms) ✔ founderCheck: founder variables, then a needed service without a usable token (1.121345ms) ✔ turnRequest names the sender, class, reply and decision (0.184961ms) ✔ turnRequest says a tester's message is a test and never Jason's instruction (0.200568ms) ✔ turnRequest shows the message's own sent time, a day old as well as new (0.159896ms) ✔ stateBlock names the launch state with its time and event, and the open tasks with the cap (0.270294ms) ✔ a message becomes a turn, the answer goes back as a RESULT, SIGTERM releases and exits 0 (190.235892ms) ✔ launches revoked and then restored: the turn's block reads allowed and gives the restore time (199.741772ms) ✔ a task closed on the tracker is missing from the next turn's open list (196.576488ms) ✔ a message read a day after it was sent shows the time it was sent (117.838577ms) ✔ a decision the human resolves becomes a turn citing the decision and the answer (143.002243ms) ✔ a failed state read fails the turn: no adapter runs, the sender hears why, the message is read (63.459711ms) ✔ a SIGTERM before the claim stops the runner with exit 0 and no claim (52.786633ms) ✔ typed tools carry the runner's capability; launch goes to the host's launch socket (272.49669ms) ✔ end_run and change_organization reach the broker with the tool's arguments, and refusals keep their codes (283.12206ms) ✔ a RESULT gets no automatic reply; failed turns reply with the reason (1194.455722ms) ✔ the bus's launch notice is one turn with no reply, and is not taken again (215.788927ms) ✔ SIGTERM during a turn kills the turn's process group and still exits 0 (95.640569ms) ✔ founder credentials stop before the claim (20) (100.230639ms) ✔ a refused claim exits 21; an ended run's capability exits 22 (108.190264ms) ✔ the launch ending under a running session exits 22 (101.383627ms) ✔ a broker that stays unreachable exits 23 after brokerRetries polls (204.15521ms) ✔ a broker that is down at the claim exits 23, not 21 (45.514549ms) ✔ no capability, or a malformed one, on stdin exits 2 (150.409706ms) ✔ a missing or malformed policy exits 2 before the claim (82.156344ms) ✔ REQ-LIMIT-4 runner brackets its actual harness turn before the reply and read receipt (124.352586ms) ✔ every role sees read_spec, record writer follows its grant, citation schemas stay strict (10.828137ms) ✔ typed socket forwards read content/history and agent record writes, and cannot write a spec (73.427975ms) ✔ the PM gets its task verbs and the reads, and no launch (6.039348ms) ✔ send_message offers the BLOCK class, beside the classes it already had (1.985061ms) ✔ the chief of staff gets launch, send_message, end_run and change_organization, and no task verbs (1.22239ms) ✔ request_staff is the pm's alone: staff.request over message.send, roles and counts bounded (6.763808ms) ✔ end_run sends role and reason, no decision; change_organization sends revision and the organization keys (1.121518ms) ✔ the organization keys change_organization takes are the broker's ORGANIZATION_KEYS (0.072088ms) ✔ a role without role.end or organization.change gets neither tool, and each tool follows its own grant (4.918446ms) ✔ the assistant gets send_message, the reads and the web tools (it holds the web grant), no launch (0.819784ms) ✔ list_requirements reads the requirement ids, for a task.create holder with a tracker (1.617952ms) ✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (0.803697ms) ✔ a coder with a tracker gets its own task tools and no task_reassign (0.76729ms) ✔ launch only when the business's launch block names the instance as launcher (1.456861ms) ✔ an action outside the instance's authority has no tool (0.859409ms) ✔ the review loop: push_working and request_review for the coder, fetch_review and review_verdict for the reviewer, none for the PM (3.158646ms) ✔ land_work is there only with land, after push_working, with the task and the decision (1.771787ms) ✔ callTool: one JSON line out, the result back, a refusal rejects (7.4935ms) ✔ list_tasks takes the broker's filters and pages; get_task takes a task_ref (1.751176ms) ✔ callTool sends list_tasks done false unless the session names done; other tools as given (1.83011ms) ✔ replyText: compact JSON up to the cap, an error naming the size and the cap over it (0.486599ms) ✔ web_search's query maxLength is the web layer's SEARCH_MAX_QUERY_CHARS (2.077565ms) ℹ tests 110 ℹ suites 0 ℹ pass 110 ℹ fail 0 ℹ cancelled 0 ℹ skipped 0 ℹ todo 0 ℹ duration_ms 20703.590543