✔ sessionModel: agent vars win, then the system's execution settings (12.024945ms)
✔ a pi bundle: prompt, policy, tools and manifest, 0600 in a 0700 directory (4.631749ms)
✔ a chief-of-staff bundle lists launch among the bus tools and the manifest's typed tools (2.142244ms)
✔ a chief-of-staff bundle lists end_run and change_organization as typed and nothing untyped, for either harness (5.621667ms)
✔ the stop line names end_run for a role.end holder, and every other role keeps today's text (6.67833ms)
✔ a claude-code bundle adds the wrapped gate hook and the MCP config (1.874832ms)
✔ the prompt's built-in tools line is the harness's own names, as the manifest lists them (8.224164ms)
✔ the prompt names the project clones in the workspace, and none when there are none (5.539894ms)
✔ the prompt explains the review loop to a role with push_working, and reviews to one with fetch_review (7.060424ms)
✔ a bundle is written once: an existing file refuses (1.288212ms)
✔ a path with a single quote can't go into the hook command (1.710189ms)
✔ web grant on pi: the bundle has web_search and web_fetch, and no built-in web tool (2.382003ms)
✔ no web grant on pi: no web tool in the bundle, and the gate blocks a web call (1.790314ms)
✔ web grant on claude-code: the bundle has web_search and web_fetch, and no built-in web tool (2.393333ms)
✔ no web grant on claude-code: no web tool in the bundle, and the gate blocks a web call (1.946547ms)
✔ a granted session's gate lets the typed web tools through and still blocks the built-ins (2.19697ms)
✔ search-source searxng-url: a granted role gets the typed broker tools and no harness web tool; an ungranted role gets nothing (3.906543ms)
✔ search-source searxng-bundled: a granted role gets the typed broker tools and no harness web tool; an ungranted role gets nothing (3.731021ms)
✔ search-source harness-builtin: claude-code gets WebSearch and WebFetch and no typed web tool, only when granted (6.328019ms)
✔ search-source harness-builtin is refused for a granted pi role, which has no built-in web tool (1.768969ms)
✔ allow exits 0, a deny exits 2 with the reason on stderr (124.164703ms)
✔ a missing or wrong policy, or a bad event, exits 2 (83.767902ms)
✔ the bundle's wrapped command: a missing gate or node still blocks (1085.16975ms)
✔ claude: typed tools through MCP, the hook blocks, builtins outside --tools don't exist (734.07543ms)
✔ claude: the hook alone blocks a path outside the workspace (511.459755ms)
✔ claude: a second turn resumes the first turn's session (764.201917ms)
✔ claude adapter: --restricted is always passed (5.207261ms)
✔ claude adapter: MOSAIC_CLAUDE_BIN wins over PATH and must be an absolute executable (16.401666ms)
✔ claude adapter: a directory pin exits 2 before it runs; a pin with a space runs (5.911408ms)
✔ claude: CLAUDE.md files and auto-memory don't reach the model; without --restricted they do (742.642036ms)
✔ claude: a missing hook or MCP file refuses before claude starts (7.417089ms)
✔ pi: policy tools and typed tools pass, anything else is blocked (2.950872ms)
✔ claude: builtins map from pi names, typed tools need the mcp prefix (0.704953ms)
✔ file tool paths must resolve inside the workspace (1.12105ms)
✔ pi's own path normalisation can't be used to step out (0.937677ms)
✔ a symlink inside the workspace that points out is outside (0.83528ms)
✔ a dangling symlink is refused at any depth, in both harnesses (3.504414ms)
✔ read is checked under every spelling pi's read would open, in both harnesses (21.929963ms)
✔ other spellings cover directories, dangling links and pi's cwd (1.528923ms)
✔ a relative path climbs from the workspace's real path, in both harnesses (5.008239ms)
✔ claude path fields per tool (0.939615ms)
✔ glob patterns stay inside the workspace (0.938786ms)
✔ a path that can't be checked is blocked (0.516534ms)
✔ fetch_review's have lists the working refs before origin's, and no more than HAVE_MAX (214.012448ms)
✔ fetch_review's have orders the working refs newest first within each clone, one clone after another (120.77238ms)
✔ a runner git past its time: SIGTERM reaches its process group, and SIGKILL follows after the grace period (1805.503066ms)
✔ once a signalled runner git has closed, its group gets SIGKILL, so a process that ignored SIGTERM dies too (304.045306ms)
✔ initialize, ping and tools/list (45.954339ms)
✔ tools/call goes through the tool socket; a refusal is an isError result (36.951715ms)
✔ replies are compact JSON; over the cap a tool error names the size and the cap (36.162225ms)
✔ unknown tools and methods are JSON-RPC errors and never reach the socket (32.138418ms)
✔ a missing argument is a usage error (28.312474ms)
✔ pi: typed tools reach the socket, the gate blocks, agent_end writes the marker (362.559037ms)
✔ pi: a reply over the cap is a tool error naming the size and the cap; list_tasks asks for open tasks (315.051088ms)
✔ pi: a write through a dangling symlink is blocked, and nothing appears outside (355.647201ms)
✔ pi: a read is refused when pi would open another spelling outside (362.960148ms)
✔ pi: a relative path climbs from the real path of a workspace behind a symlink (324.758584ms)
✔ pi: a relative path climbs from the real path of a dataRoot behind a symlink (300.654974ms)
✔ pi: a missing extension refuses before any model call (5.634793ms)
✔ pi: an extension without its configuration fails pi's start (254.723128ms)
✔ founderCheck: founder variables, then a needed service without a usable token (1.019746ms)
✔ turnRequest names the sender, class, reply and decision (0.254245ms)
✔ turnRequest says a tester's message is a test and never Jason's instruction (0.232449ms)
✔ turnRequest shows the message's own sent time, a day old as well as new (0.145251ms)
✔ stateBlock names the launch state with its time and event, and the open tasks with the cap (0.21098ms)
✔ a message becomes a turn, the answer goes back as a RESULT, SIGTERM releases and exits 0 (257.975956ms)
✔ launches revoked and then restored: the turn's block reads allowed and gives the restore time (265.839568ms)
✔ a task closed on the tracker is missing from the next turn's open list (233.440834ms)
✔ a message read a day after it was sent shows the time it was sent (174.175295ms)
✔ a decision the human resolves becomes a turn citing the decision and the answer (180.304495ms)
✔ a failed state read fails the turn: no adapter runs, the sender hears why, the message is read (148.331562ms)
✔ a SIGTERM before the claim stops the runner with exit 0 and no claim (105.136993ms)
✔ typed tools carry the runner's capability; launch goes to the host's launch socket (429.501014ms)
✔ end_run and change_organization reach the broker with the tool's arguments, and refusals keep their codes (424.019026ms)
✔ a RESULT gets no automatic reply; failed turns reply with the reason (1319.830773ms)
✔ the bus's launch notice is one turn with no reply, and is not taken again (306.490743ms)
✔ SIGTERM during a turn kills the turn's process group and still exits 0 (151.95878ms)
✔ founder credentials stop before the claim (20) (166.788034ms)
✔ a refused claim exits 21; an ended run's capability exits 22 (193.571729ms)
✔ the launch ending under a running session exits 22 (142.606392ms)
✔ a broker that stays unreachable exits 23 after brokerRetries polls (246.842523ms)
✔ a broker that is down at the claim exits 23, not 21 (91.080107ms)
✔ no capability, or a malformed one, on stdin exits 2 (191.25947ms)
✔ a missing or malformed policy exits 2 before the claim (124.31863ms)
✔ the PM gets its task verbs and the reads, and no launch (10.148378ms)
✔ the chief of staff gets launch, send_message, end_run and change_organization, and no task verbs (2.436336ms)
✔ end_run sends role and reason, no decision; change_organization sends revision and the organization keys (2.705063ms)
✔ the organization keys change_organization takes are the broker's ORGANIZATION_KEYS (0.171775ms)
✔ a role without role.end or organization.change gets neither tool, and each tool follows its own grant (9.202439ms)
✔ the assistant gets send_message, the reads and the web tools (it holds the web grant), no launch (2.559048ms)
✔ list_requirements reads the requirement ids, for a task.create holder with a tracker (3.732564ms)
✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (1.690735ms)
✔ a coder with a tracker gets its own task tools and no task_reassign (1.495285ms)
✔ launch only when the business's launch block names the instance as launcher (2.378713ms)
✔ an action outside the instance's authority has no tool (1.762047ms)
✔ the review loop: push_working and request_review for the coder, fetch_review and review_verdict for the reviewer, none for the PM (5.154138ms)
✔ callTool: one JSON line out, the result back, a refusal rejects (9.657511ms)
✔ list_tasks takes the broker's filters and pages; get_task takes a task_ref (2.398156ms)
✔ callTool sends list_tasks done false unless the session names done; other tools as given (2.420346ms)
✔ replyText: compact JSON up to the cap, an error naming the size and the cap over it (0.778744ms)
✔ web_search's query maxLength is the web layer's SEARCH_MAX_QUERY_CHARS (2.430289ms)
ℹ tests 101
ℹ suites 0
ℹ pass 101
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 12128.599003