✔ sessionModel: agent vars win, then the system's execution settings (12.150642ms)
✔ a pi bundle: prompt, policy, tools and manifest, 0600 in a 0700 directory (5.832684ms)
✔ a chief-of-staff bundle lists launch among the bus tools and the manifest's typed tools (3.169465ms)
✔ a chief-of-staff bundle lists end_run and change_organization as typed and only instance.name untyped, for either harness (5.925184ms)
✔ the stop line names end_run for a role.end holder, and every other role keeps today's text (8.409349ms)
✔ a claude-code bundle adds the wrapped gate hook and the MCP config (2.393608ms)
✔ the prompt's built-in tools line is the harness's own names, as the manifest lists them (8.599794ms)
✔ the prompt names the project clones in the workspace, and none when there are none (8.300119ms)
✔ the prompt explains the review loop to a role with push_working, and reviews to one with fetch_review (6.040157ms)
✔ the prompt explains landing to a role with land_work, and to no other (3.597515ms)
✔ the prompt tells the pm to ask the chief of staff for sessions, and no other role (6.349545ms)
✔ a bundle is written once: an existing file refuses (1.400862ms)
✔ a path with a single quote can't go into the hook command (1.488588ms)
✔ web grant on pi: the bundle has web_search and web_fetch, and no built-in web tool (3.018221ms)
✔ no web grant on pi: no web tool in the bundle, and the gate blocks a web call (2.680396ms)
✔ web grant on claude-code: the bundle has web_search and web_fetch, and no built-in web tool (2.5541ms)
✔ no web grant on claude-code: no web tool in the bundle, and the gate blocks a web call (2.433623ms)
✔ a granted session's gate lets the typed web tools through and still blocks the built-ins (2.518287ms)
✔ search-source searxng-url: a granted role gets the typed broker tools and no harness web tool; an ungranted role gets nothing (6.15817ms)
✔ search-source searxng-bundled: a granted role gets the typed broker tools and no harness web tool; an ungranted role gets nothing (4.427589ms)
✔ search-source harness-builtin: claude-code gets WebSearch and WebFetch and no typed web tool, only when granted (5.815856ms)
✔ search-source harness-builtin is refused for a granted pi role, which has no built-in web tool (1.604703ms)
✔ allow exits 0, a deny exits 2 with the reason on stderr (129.065093ms)
✔ a missing or wrong policy, or a bad event, exits 2 (85.220311ms)
✔ the bundle's wrapped command: a missing gate or node still blocks (1088.085323ms)
✔ claude: typed tools through MCP, the hook blocks, builtins outside --tools don't exist (1007.331912ms)
✔ claude: the hook alone blocks a path outside the workspace (543.444588ms)
✔ claude: a second turn resumes the first turn's session (743.346743ms)
✔ claude adapter: --restricted is always passed (4.832311ms)
✔ claude adapter: MOSAIC_CLAUDE_BIN wins over PATH and must be an absolute executable (15.235088ms)
✔ claude adapter: a directory pin exits 2 before it runs; a pin with a space runs (5.821409ms)
✔ claude: CLAUDE.md files and auto-memory don't reach the model; without --restricted they do (736.793582ms)
✔ claude: a missing hook or MCP file refuses before claude starts (8.346391ms)
✔ pi: policy tools and typed tools pass, anything else is blocked (3.310977ms)
✔ claude: builtins map from pi names, typed tools need the mcp prefix (0.7263ms)
✔ file tool paths must resolve inside the workspace (1.386761ms)
✔ pi's own path normalisation can't be used to step out (0.894715ms)
✔ a symlink inside the workspace that points out is outside (0.889261ms)
✔ a dangling symlink is refused at any depth, in both harnesses (3.706897ms)
✔ read is checked under every spelling pi's read would open, in both harnesses (21.388536ms)
✔ other spellings cover directories, dangling links and pi's cwd (2.217038ms)
✔ a relative path climbs from the workspace's real path, in both harnesses (7.762969ms)
✔ claude path fields per tool (1.004671ms)
✔ glob patterns stay inside the workspace (0.901303ms)
✔ a path that can't be checked is blocked (0.562692ms)
✔ fetch_review's have lists the working refs before origin's, and no more than HAVE_MAX (215.080399ms)
✔ fetch_review's have orders the working refs newest first within each clone, one clone after another (136.859919ms)
✔ a runner git past its time: SIGTERM reaches its process group, and SIGKILL follows after the grace period (1815.50753ms)
✔ once a signalled runner git has closed, its group gets SIGKILL, so a process that ignored SIGTERM dies too (303.942492ms)
✔ land_work sends one land line and resolves the host's started answer (3.155397ms)
✔ initialize, ping and tools/list (54.952023ms)
✔ tools/call goes through the tool socket; a refusal is an isError result (34.777752ms)
✔ replies are compact JSON; over the cap a tool error names the size and the cap (40.677573ms)
✔ unknown tools and methods are JSON-RPC errors and never reach the socket (31.504495ms)
✔ a missing argument is a usage error (26.12639ms)
✔ pi: typed tools reach the socket, the gate blocks, agent_end writes the marker (368.02621ms)
✔ pi: a reply over the cap is a tool error naming the size and the cap; list_tasks asks for open tasks (468.4893ms)
✔ pi: a write through a dangling symlink is blocked, and nothing appears outside (524.558151ms)
✔ pi: a read is refused when pi would open another spelling outside (322.547441ms)
✔ pi: a relative path climbs from the real path of a workspace behind a symlink (305.376786ms)
✔ pi: a relative path climbs from the real path of a dataRoot behind a symlink (293.829928ms)
✔ pi: a missing extension refuses before any model call (5.965505ms)
✔ pi: an extension without its configuration fails pi's start (241.410515ms)
✔ raise_decision offers domain as technical or delivery only, optional, and says where a raise goes (12.926762ms)
✔ raise_decision forwards domain to decision.raise, and a call without it sends none (196.479266ms)
✔ founderCheck: founder variables, then a needed service without a usable token (1.443789ms)
✔ turnRequest names the sender, class, reply and decision (0.200527ms)
✔ turnRequest says a tester's message is a test and never Jason's instruction (0.299781ms)
✔ turnRequest shows the message's own sent time, a day old as well as new (0.200439ms)
✔ stateBlock names the launch state with its time and event, and the open tasks with the cap (0.228865ms)
✔ a message becomes a turn, the answer goes back as a RESULT, SIGTERM releases and exits 0 (303.571055ms)
✔ launches revoked and then restored: the turn's block reads allowed and gives the restore time (279.167814ms)
✔ a task closed on the tracker is missing from the next turn's open list (543.673825ms)
✔ a message read a day after it was sent shows the time it was sent (190.701354ms)
✔ a decision the human resolves becomes a turn citing the decision and the answer (239.74407ms)
✔ a failed state read fails the turn: no adapter runs, the sender hears why, the message is read (167.86111ms)
✔ a SIGTERM before the claim stops the runner with exit 0 and no claim (106.995451ms)
✔ typed tools carry the runner's capability; launch goes to the host's launch socket (391.808669ms)
✔ end_run and change_organization reach the broker with the tool's arguments, and refusals keep their codes (441.595851ms)
✔ a RESULT gets no automatic reply; failed turns reply with the reason (1306.130456ms)
✔ the bus's launch notice is one turn with no reply, and is not taken again (254.795084ms)
✔ SIGTERM during a turn kills the turn's process group and still exits 0 (153.343913ms)
✔ founder credentials stop before the claim (20) (214.605731ms)
✔ a refused claim exits 21; an ended run's capability exits 22 (196.846984ms)
✔ the launch ending under a running session exits 22 (144.320258ms)
✔ a broker that stays unreachable exits 23 after brokerRetries polls (249.217522ms)
✔ a broker that is down at the claim exits 23, not 21 (113.483347ms)
✔ no capability, or a malformed one, on stdin exits 2 (197.841353ms)
✔ a missing or malformed policy exits 2 before the claim (133.753894ms)
✔ every role sees read_spec, record writer follows its grant, citation schemas stay strict (18.661529ms)
✔ typed socket forwards read content/history and agent record writes, and cannot write a spec (228.958275ms)
✔ the PM gets its task verbs and the reads, and no launch (10.214747ms)
✔ the chief of staff gets launch, send_message, end_run and change_organization, and no task verbs (2.880895ms)
✔ request_staff is the pm's alone: staff.request over message.send, roles and counts bounded (11.337387ms)
✔ end_run sends role and reason, no decision; change_organization sends revision and the organization keys (1.778426ms)
✔ the organization keys change_organization takes are the broker's ORGANIZATION_KEYS (0.142265ms)
✔ a role without role.end or organization.change gets neither tool, and each tool follows its own grant (6.450945ms)
✔ the assistant gets send_message, the reads and the web tools (it holds the web grant), no launch (1.12375ms)
✔ list_requirements reads the requirement ids, for a task.create holder with a tracker (1.924292ms)
✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (0.930188ms)
✔ a coder with a tracker gets its own task tools and no task_reassign (0.938301ms)
✔ launch only when the business's launch block names the instance as launcher (2.156942ms)
✔ an action outside the instance's authority has no tool (1.015733ms)
✔ the review loop: push_working and request_review for the coder, fetch_review and review_verdict for the reviewer, none for the PM (4.105196ms)
✔ land_work is there only with land, after push_working, with the task and the decision (2.324557ms)
✔ callTool: one JSON line out, the result back, a refusal rejects (14.190711ms)
✔ list_tasks takes the broker's filters and pages; get_task takes a task_ref (3.503434ms)
✔ callTool sends list_tasks done false unless the session names done; other tools as given (3.011195ms)
✔ replyText: compact JSON up to the cap, an error naming the size and the cap over it (0.625689ms)
✔ web_search's query maxLength is the web layer's SEARCH_MAX_QUERY_CHARS (2.370115ms)
ℹ tests 110
ℹ suites 0
ℹ pass 110
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 12523.448508