✔ sessionModel: the route is the instance's own vars, and a missing one refuses no-route by name (23.337182ms)
✔ a pi bundle: prompt, policy, tools and manifest, 0600 in a 0700 directory (6.585387ms)
✔ a chief-of-staff bundle lists launch among the bus tools and the manifest's typed tools (3.287391ms)
✔ a chief-of-staff bundle lists end_run, change_organization and set_name as typed and nothing untyped, for either harness (6.980242ms)
✔ the stop line names end_run for a role.end holder, and every other role keeps today's text (8.014161ms)
✔ a claude-code bundle adds the wrapped gate hook and the MCP config (2.743928ms)
✔ the prompt's built-in tools line is the harness's own names, as the manifest lists them (11.015362ms)
✔ the prompt names the project clones in the workspace, and none when there are none (7.890432ms)
✔ the prompt explains the review loop to a role with push_working, and reviews to one with fetch_review (8.058726ms)
✔ the prompt explains landing to a role with land_work, and to no other (3.609657ms)
✔ the prompt tells the pm to ask the chief of staff for sessions, and no other role (6.671238ms)
✔ a bundle is written once: an existing file refuses (1.880554ms)
✔ a path with a single quote can't go into the hook command (1.504739ms)
✔ web grant on pi: the bundle has web_search and web_fetch, and no built-in web tool (2.977459ms)
✔ no web grant on pi: no web tool in the bundle, and the gate blocks a web call (2.575856ms)
✔ web grant on claude-code: the bundle has web_search and web_fetch, and no built-in web tool (2.235129ms)
✔ no web grant on claude-code: no web tool in the bundle, and the gate blocks a web call (2.450862ms)
✔ a granted session's gate lets the typed web tools through and still blocks the built-ins (2.53027ms)
✔ search-source searxng-url: a granted role gets the typed broker tools and no harness web tool; an ungranted role gets nothing (6.938061ms)
✔ search-source searxng-bundled: a granted role gets the typed broker tools and no harness web tool; an ungranted role gets nothing (5.06774ms)
✔ search-source harness-builtin: claude-code gets WebSearch and WebFetch and no typed web tool, only when granted (9.164301ms)
✔ search-source harness-builtin is refused for a granted pi role, which has no built-in web tool (2.360098ms)
✔ allow exits 0, a deny exits 2 with the reason on stderr (159.042473ms)
✔ a missing or wrong policy, or a bad event, exits 2 (87.356795ms)
✔ the bundle's wrapped command: a missing gate or node still blocks (1086.469584ms)
✔ claude: typed tools through MCP, the hook blocks, builtins outside --tools don't exist (745.467256ms)
✔ claude: the hook alone blocks a path outside the workspace (441.374411ms)
✔ claude: a second turn resumes the first turn's session (768.329072ms)
✔ claude adapter: --restricted is always passed (5.615918ms)
✔ claude adapter: MOSAIC_CLAUDE_BIN wins over PATH and must be an absolute executable (17.088169ms)
✔ claude adapter: a directory pin exits 2 before it runs; a pin with a space runs (5.858043ms)
✔ claude: CLAUDE.md files and auto-memory don't reach the model; without --restricted they do (706.537302ms)
✔ claude: a missing hook or MCP file refuses before claude starts (7.514639ms)
✔ pi: policy tools and typed tools pass, anything else is blocked (4.468236ms)
✔ claude: builtins map from pi names, typed tools need the mcp prefix (0.969304ms)
✔ file tool paths must resolve inside the workspace (1.703266ms)
✔ pi's own path normalisation can't be used to step out (1.390893ms)
✔ a symlink inside the workspace that points out is outside (2.134134ms)
✔ a dangling symlink is refused at any depth, in both harnesses (4.135944ms)
✔ read is checked under every spelling pi's read would open, in both harnesses (20.944906ms)
✔ other spellings cover directories, dangling links and pi's cwd (1.805185ms)
✔ a relative path climbs from the workspace's real path, in both harnesses (6.751509ms)
✔ claude path fields per tool (1.034102ms)
✔ glob patterns stay inside the workspace (1.022938ms)
✔ a path that can't be checked is blocked (0.550964ms)
✔ fetch_review's have lists the working refs before origin's, and no more than HAVE_MAX (242.657674ms)
✔ fetch_review's have orders the working refs newest first within each clone, one clone after another (128.206966ms)
✔ a runner git past its time: SIGTERM reaches its process group, and SIGKILL follows after the grace period (1809.922137ms)
✔ once a signalled runner git has closed, its group gets SIGKILL, so a process that ignored SIGTERM dies too (303.098669ms)
✔ land_work sends one land line and resolves the host's started answer (3.365625ms)
✔ initialize, ping and tools/list (51.480915ms)
✔ tools/call goes through the tool socket; a refusal is an isError result (50.851116ms)
✔ replies are compact JSON; over the cap a tool error names the size and the cap (43.373394ms)
✔ unknown tools and methods are JSON-RPC errors and never reach the socket (34.933293ms)
✔ a missing argument is a usage error (30.170795ms)
✔ pi: typed tools reach the socket, the gate blocks, agent_end writes the marker (385.388717ms)
✔ pi: a reply over the cap is a tool error naming the size and the cap; list_tasks asks for open tasks (303.76173ms)
✔ pi: a write through a dangling symlink is blocked, and nothing appears outside (320.178176ms)
✔ pi: a read is refused when pi would open another spelling outside (340.918456ms)
✔ pi: a relative path climbs from the real path of a workspace behind a symlink (322.815239ms)
✔ pi: a relative path climbs from the real path of a dataRoot behind a symlink (316.608813ms)
✔ pi: a missing extension refuses before any model call (5.664457ms)
✔ pi: an extension without its configuration fails pi's start (250.377773ms)
✔ raise_decision offers domain as technical or delivery only, optional, and says where a raise goes (13.59357ms)
✔ raise_decision forwards domain to decision.raise, and a call without it sends none (61.377045ms)
✔ founderCheck: founder variables, then a needed service without a usable token (1.646273ms)
✔ turnRequest names the sender, class, reply and decision (0.28809ms)
✔ turnRequest says a tester's message is a test and never Jason's instruction (0.232117ms)
✔ turnRequest shows the message's own sent time, a day old as well as new (0.134941ms)
✔ stateBlock names the launch state with its time and event, and the open tasks with the cap (0.249763ms)
✔ a message becomes a turn, the answer goes back as a RESULT, SIGTERM releases and exits 0 (145.995551ms)
✔ launches revoked and then restored: the turn's block reads allowed and gives the restore time (199.702362ms)
✔ a task closed on the tracker is missing from the next turn's open list (170.125723ms)
✔ a message read a day after it was sent shows the time it was sent (90.618188ms)
✔ a decision the human resolves becomes a turn citing the decision and the answer (142.893542ms)
✔ a failed state read fails the turn: no adapter runs, the sender hears why, the message is read (67.842452ms)
✔ a SIGTERM before the claim stops the runner with exit 0 and no claim (54.749512ms)
✔ typed tools carry the runner's capability; launch goes to the host's launch socket (304.996781ms)
✔ end_run, organization_revision and change_organization reach the broker with the tool's arguments, and refusals keep their codes (366.912877ms)
✔ set_name reaches the broker's name.set with the tool's arguments, and refusals keep their codes (148.519631ms)
✔ a RESULT gets no automatic reply; failed turns reply with the reason (1204.037738ms)
✔ the bus's launch notice is one turn with no reply, and is not taken again (219.691351ms)
✔ SIGTERM during a turn kills the turn's process group and still exits 0 (96.836356ms)
✔ founder credentials stop before the claim (20) (91.438967ms)
✔ a refused claim exits 21; an ended run's capability exits 22 (97.127591ms)
✔ the launch ending under a running session exits 22 (98.595902ms)
✔ a broker that stays unreachable exits 23 after brokerRetries polls (203.291607ms)
✔ a broker that is down at the claim exits 23, not 21 (57.758738ms)
✔ no capability, or a malformed one, on stdin exits 2 (153.31098ms)
✔ a missing or malformed policy exits 2 before the claim (87.507225ms)
✔ REQ-LIMIT-4 runner brackets its actual harness turn before the reply and read receipt (145.848422ms)
✔ REQ-LIMIT-4 a runner at its limit stops at the end of its turn: RESULT and read first, then exit 22 and a limit-check revoke (134.860129ms)
✔ landingRequest is the fixed template, with the decision and task ref and nothing else (0.13704ms)
✔ each landing entry is one turn before any message, with the state block and the fixed request; it sends nothing and reads nothing (146.175143ms)
✔ an empty landing takes no landing turn: the message is turn 1 (144.386349ms)
✔ a run in a landing turn isn't released for a second held decision until its landing turns end and it has read its messages (563.666815ms)
✔ a malformed landing exits 2 before the claim, with no turn (492.376644ms)
✔ row 65 a session.json without a turnEnd of marker or exit, or a harnessEnv of strings, exits 2 before the claim (216.819555ms)
✔ row 65 harnessEnv reaches the adapter, and an exit harness needs no turn marker (118.768837ms)
✔ every role sees read_spec, record writer follows its grant, citation schemas stay strict (15.695501ms)
✔ typed socket forwards read content/history and agent record writes, and cannot write a spec (109.984777ms)
✔ the PM gets its task verbs and the reads, and no launch (13.052729ms)
✔ send_message offers the BLOCK class, beside the classes it already had (2.818852ms)
✔ the chief of staff gets launch, send_message, end_run, change_organization and set_name, and no task verbs (2.325564ms)
✔ request_staff is the pm's alone: staff.request over message.send, roles and counts bounded (14.218219ms)
✔ end_run sends role and reason, no decision; change_organization sends revision and the organization keys (1.916153ms)
✔ set_name sends instance and name to name.set under instance.name, only for a holder of that grant (7.484961ms)
✔ the organization keys change_organization takes are the broker's ORGANIZATION_KEYS and the settings' SETTINGS_KEYS (0.095466ms)
✔ a role without role.end or organization.change gets none of the three tools, and each follows its own grant (6.595301ms)
✔ no role's tools send instances or instances.set, and change_organization takes no instance write keys (8.425303ms)
✔ the assistant gets send_message, the reads and the web tools (it holds the web grant), no launch (1.780498ms)
✔ list_requirements reads the requirement ids, for a task.create holder with a tracker (3.061507ms)
✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (1.585479ms)
✔ a coder with a tracker gets its own task tools and no task_reassign (1.16723ms)
✔ launch only when the business's launch block names the instance as launcher (1.883282ms)
✔ an action outside the instance's authority has no tool (0.971597ms)
✔ the review loop: push_working and request_review for the coder, fetch_review and review_verdict for the reviewer, none for the PM (3.920361ms)
✔ land_work is there only with land, after push_working, with the task and the decision (2.923093ms)
✔ callTool: one JSON line out, the result back, a refusal rejects (10.71244ms)
✔ list_tasks takes the broker's filters and pages; get_task takes a task_ref (3.17027ms)
✔ callTool sends list_tasks done false unless the session names done; other tools as given (3.396759ms)
✔ replyText: compact JSON up to the cap, an error naming the size and the cap over it (0.859713ms)
✔ web_search's query maxLength is the web layer's SEARCH_MAX_QUERY_CHARS (3.351086ms)
ℹ tests 123
ℹ suites 0
ℹ pass 123
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 11352.551913
