Run passed.

[luci] cache npm: exact
[luci] step 1/5: export NPM_CONFIG_CACHE="$PWD/.npm-cache"
[luci] step 1/5 done in 0s
[luci] step 2/5: npm ci --prefer-offline
npm warn deprecated node-domexception@1.0.0: Use your platform's native DOMException instead

added 230 packages, and audited 231 packages in 10s

47 packages are looking for funding
  run `npm fund` for details

found 0 vulnerabilities
[luci] step 2/5 done in 11s
[luci] step 3/5: npm run check

> will@0.1.0 check
> biome check src scripts web deploy rolldown.config.ts playwright.config.ts package.json tsconfig.json && tsc --noEmit

Checked 79 files in 241ms. No fixes applied.
[luci] step 3/5 done in 13s
[luci] step 4/5: npm test

> will@0.1.0 test
> node --test src

✔ explicit fake wins even with a key (1.882781ms)
✔ no env at all falls back to fake (0.264818ms)
✔ key alone gives the real session (0.252941ms)
✔ explicit real without a key fails loudly (0.737572ms)
✔ explicit real with a key is real (0.292715ms)
✔ thinking level defaults to medium (0.834455ms)
✔ ZAI_THINKING_LEVEL overrides the default (0.445556ms)
✔ unknown ZAI_THINKING_LEVEL fails loudly (2.153026ms)
✔ wake queue persists across reopen; heartbeat coalesces; done compacts (17.972562ms)
✔ wake queue defer reorders wake with backoff (5.237818ms)
{"ts":"2026-09-26T20:52:29.288Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
✔ spool drains idempotently into the running service (5298.936816ms)
✔ agent state persists mutations atomically (4.860316ms)
{"ts":"2026-09-26T20:52:36.761Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
{"ts":"2026-09-26T20:52:36.768Z","level":"warn","component":"agent","message":"running with a FAKE session (no model calls; set ZAI_API_KEY for real)"}
{"ts":"2026-09-26T20:52:36.769Z","level":"warn","component":"agent","message":"worktree not ready","error":"worktree missing and no git url configured"}
{"ts":"2026-09-26T20:52:36.846Z","level":"info","component":"telegram","message":"bot identified","username":"will_bot"}
{"ts":"2026-09-26T20:52:37.926Z","level":"debug","component":"agent","message":"spool drained","events":2}
{"ts":"2026-09-26T20:52:38.015Z","level":"info","component":"agent","message":"agent stopped cleanly"}
✔ end to end: mention reaches the resident session and readiness is healthy (6417.200406ms)
✔ chat turn completes, budget spent, wake removed (15.646047ms)
✔ boot context appears only on the first turn (5.491376ms)
✔ model failure retries with backoff, then defers after three attempts (3.969017ms)
✔ turn timeout enqueues exactly one recovery wake and completes the original (56.525273ms)
✔ budget cap blocks ordinary wakes but reserve admits recovery wakes (6.176693ms)
✔ storage stop pauses ordinary wakes, recovery continues (5.443534ms)
✔ system prompt is exactly the lexically ordered personality (6.050952ms)
✔ wake prompts render per-kind templates with facts (1.064545ms)
✔ facts block omits missing values instead of rendering undefined (0.407009ms)
✔ all coding tools are bound to the editable worktree (11.579446ms)
✔ worktree: clone, ff-only when clean, preserved when dirty or ahead (400.106196ms)
✔ readWorktreeState: fresh head and dirty flag without fetch or merge (134.113917ms)
{"ts":"2026-09-26T20:52:52.091Z","level":"debug","component":"ciwatch","message":"ci poll","runs":1,"wakes":0}
{"ts":"2026-09-26T20:52:52.099Z","level":"debug","component":"ciwatch","message":"ci poll","runs":2,"wakes":1}
{"ts":"2026-09-26T20:52:52.102Z","level":"debug","component":"ciwatch","message":"ci poll","runs":2,"wakes":0}
{"ts":"2026-09-26T20:52:52.105Z","level":"debug","component":"ciwatch","message":"ci poll","runs":3,"wakes":0}
✔ ci watch enqueues one failure wake per failed run of a tracked sha (65.169389ms)
{"ts":"2026-09-26T20:52:52.239Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
{"ts":"2026-09-26T20:52:52.265Z","level":"info","component":"deploy","message":"deploy events imported","count":2}
✔ deploy events poll mirrors into service and wakes on failures only (5182.344902ms)
{"ts":"2026-09-26T20:52:57.404Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
✔ chat_send queues outbound; recall queries service; maintenance ops gated (5145.113104ms)
{"ts":"2026-09-26T20:53:02.551Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
{"ts":"2026-09-26T20:53:02.568Z","level":"warn","component":"loop","message":"storage threshold transition","from":"ok","to":"warn"}
✔ storage watch: transitions emit deterministic outbox notice (5138.327042ms)
✔ readiness server reports 503 until local init completes (11.807513ms)
✔ fake telegram sanity (60.119468ms)
{"ts":"2026-09-26T20:53:08.093Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
{"ts":"2026-09-26T20:53:08.104Z","level":"info","component":"telegram","message":"bot identified","username":"will_bot"}
✔ mention wakes, chatter archives without wake, other chats ignored (5218.208432ms)
{"ts":"2026-09-26T20:53:13.291Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
{"ts":"2026-09-26T20:53:13.297Z","level":"info","component":"telegram","message":"bot identified","username":"will_bot"}
✔ media message downloads into service store and references digest (5163.614195ms)
{"ts":"2026-09-26T20:53:18.451Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
{"ts":"2026-09-26T20:53:18.464Z","level":"info","component":"telegram","message":"bot identified","username":"will_bot"}
✔ /hide via reply tombstones deterministically without a model wake (5180.191265ms)
{"ts":"2026-09-26T20:53:23.635Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
✔ outbox delivers at-least-once and mirrors outbound to service (5143.505253ms)
✔ sums usage across assistant messages of one wake (2.815965ms)
✔ splits by provider/model and tolerates junk (0.492421ms)
✔ non-array input yields no rows (0.274968ms)
{"ts":"2026-09-26T20:53:30.718Z","level":"warn","component":"zai","message":"zai retry","path":"/paas/v4/chat/completions","status":500,"attempt":1,"delay":500}
✔ rest client retries retryable failures (565.730939ms)
✔ rest client surfaces 4xx without retry (21.278003ms)
{"ts":"2026-09-26T20:53:31.374Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
✔ zai_analyze_media answers, caches by digest, second call is a hit (5191.641351ms)
{"ts":"2026-09-26T20:53:36.556Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
✔ zai_transcribe and zai_parse_document return explicit results (5181.725438ms)
{"ts":"2026-09-26T20:53:41.747Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
✔ generation tools store bytes and return digests; limits fail explicitly (8482.955341ms)
{"ts":"2026-09-26T20:53:50.218Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
✔ curated tool set carries stable names and one-line descriptions (5118.035723ms)
{"ts":"2026-09-26T20:53:55.595Z","level":"info","component":"zai-mcp","message":"vision mcp connected"}
✔ mcp bridge lists and calls tools over stdio (408.71615ms)
{"ts":"2026-09-26T20:53:56.061Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
✔ healthz ok, index served, docs manifest served (8217.605064ms)
{"ts":"2026-09-26T20:54:04.256Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
✔ path traversal is rejected (5133.771228ms)
{"ts":"2026-09-26T20:54:09.391Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
✔ audit ingest -> timeline -> SSE replay with Last-Event-ID (5193.149964ms)
{"ts":"2026-09-26T20:54:14.582Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
✔ media upload, meta, public download round-trip (5149.899245ms)
{"ts":"2026-09-26T20:54:19.733Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
✔ chat ingest, hide via internal API, tombstone filters public chat (5139.294869ms)
{"ts":"2026-09-26T20:54:24.870Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
✔ status reports schema, counts and storage; recall works over HTTP (5134.990012ms)
{"ts":"2026-09-26T20:54:30.004Z","level":"info","component":"service","message":"service listening","public":"127.0.0.1:0","internal":"127.0.0.1:0"}
✔ usage view aggregates turn_usage rows into daily cost (5124.322169ms)
✔ audit ingest is idempotent by event id and assigns seqs (12.978311ms)
✔ recall searches beyond the newest page in hot data and archives (39.252964ms)
✔ recall orders overlapping archives and late hot rows globally (35.324945ms)
✔ recall has identical Unicode case matching before and after archival (22.279522ms)
✔ recall tie ordering is limit-prefix consistent before and after archival (12.982412ms)
✔ chat ingest is idempotent and hide tombstones (3.997028ms)
✔ deployment events mirror idempotently (3.320414ms)
✔ media store is content-addressed and idempotent (5.639207ms)
✔ analysis cache upserts by (digest, capability, model, schemaVersion) (3.645182ms)
✔ archive cycle moves old audit rows exactly once and cold reads serve them (13.685891ms)
✔ crash between publish and delete does not duplicate rows (exactly once) (10.551236ms)
✔ archive cycle moves chat after one year and tombstones keep filtering (14.741575ms)
✔ quick check reports ok and jobs are recorded (5.78979ms)
✔ vacuum runs without error (3.447118ms)
✔ cost uses the price table per MTok (2.901307ms)
✔ unknown model keeps tokens but marks cost unavailable (0.498666ms)
✔ days are grouped by UTC date and sorted ascending (1.494601ms)
✔ projection extrapolates today across the remaining month (0.361944ms)
✔ projection is null without a priced day today (0.294986ms)
✔ ordinary calls stop at cap minus reserve (1.876501ms)
✔ recovery calls draw from the reserve only (0.299649ms)
✔ zero cap disables limiting (0.263133ms)
✔ state rolls over on a new day (1.885042ms)
✔ property: total allowed recovery calls per day equals reserve (25.7982ms)
✔ default budget exists and is sane (0.402683ms)
✔ cutoff subtracts hot days (1.998327ms)
✔ only rows strictly older than the cutoff archive (1.349258ms)
✔ rows group by month (0.378989ms)
✔ sidecar overlap selection skips non-overlapping segments (0.467886ms)
✔ property: every row lands in exactly one month group (17.304103ms)
✔ mention of the bot wakes (3.315912ms)
✔ reply to the bot wakes (0.438297ms)
✔ ordinary chatter never wakes (0.275792ms)
✔ mention inside a reply of someone else does not wake (0.389964ms)
✔ /hide via reply is deterministic, never wakes the model (0.387824ms)
✔ /hide with explicit id targets that id (0.446292ms)
✔ other commands wake (0.285228ms)
✔ caption mention on media wakes (0.400668ms)
✔ sender of the target or oliver may hide (0.389446ms)
✔ toTgMessage extracts largest photo and documents (1.834959ms)
✔ failure wakes outrank chat which outranks heartbeat (3.108301ms)
✔ FIFO inside a class (0.344475ms)
✔ deferred wakes are skipped until their time (0.348169ms)
✔ heartbeat coalescing predicate (0.440798ms)
✔ heartbeat hour is claimed at most once and never replayed (0.37637ms)
✔ property: ordering is stable sort by (priority, seq) (14.51246ms)
ℹ tests 98
ℹ suites 0
ℹ pass 98
ℹ fail 0
ℹ cancelled 0
ℹ skipped 0
ℹ todo 0
ℹ duration_ms 127982.473125
[luci] step 4/5 done in 128s
[luci] step 5/5: npm run coverage

> will@0.1.0 coverage
> node --test --experimental-test-coverage --test-coverage-include='src/**/*.ts' "src/**/*.test.ts" "src/**/*.int.test.ts" 2>&1 | node scripts/coverage-ratchet.mjs

coverage-ratchet: 91.4% >= floor 91.2% (ratchet raised, commit it)
[luci] step 5/5 done in 127s

Complete raw log