docs: record responses choice verification and GLM tool-stream evidence

BUILD-LOG.md

@@ -5592,3 +5592,5 @@ Fethry at c8a9bcf94, per Sage sage-4292 and mos-1616 pi-route amendment. Before:
 Fethry stopped under Mos mos-1630/Jason's T3 crash instruction. No production implementation, tests, probes, mutants or acceptance runs occurred; Sage CLEAR never arrived. Explicit route metadata per sage-4293 is recorded in agents/fethry/work/queue-332/BUILD.md; full state and resumption boundaries are in HANDOFF.md there. Safe crash checkpoint commits documentation/historical plain-text import manifest only; four unrun tests and the import-audit script remain uncommitted. No long jobs restarted, protected branch changes, merges, live changes or further thread messages.
 
 2026-10-24 — Fethry row332 P3 UNRUN implementation checkpoint: optional proxy route per sage-4325 (default Anthropic), transient preflight/bundle/spawn context and serialization split, pi private model file/provider credential requirement. Existing initial RED evidence precedes source changes; only specifically authorized old fixtures updated. GPT wire API value PENDING until separate offline CLEAR/evidence; P2 seam remains injected context, no real proxy/auth access. Not GREEN, not accepted or landed. Details agents/fethry/work/queue-332/BUILD.md.
+
+2026-10-24 — Fethry row332 P3 checkpoint ef0939e35: required explicit route restored per sage-4293/4328; openai-responses chosen by Sage sage-4331 as reasoning-based tie-break, not evidence-decided (both offline shapes passed). Rehashed pinned8.0.42 single active-tool isolated GLM turn confirms thinking/tool_stream survive to stand-in;bundle7/7,proxy-harness5/5,no skips. Real Codex auth proof remains row209 phaseA step7. P2 landing/rebase,mutants,reviews and full gate still pending; no landing claim. Plain evidence/limitations in agents/fethry/work/queue-332/BUILD.md.

agents/fethry/work/queue-332/BUILD.md

@@ -282,3 +282,18 @@ Sage's ruling: PROXY_CODEX_API=openai-responses. The evidence does not separate
 Set constant/remove pending refusal; GPT route test asserts openai-responses on mosaic-proxy-codex. Replaced temporary pending-refusal test with generated single-provider/responses/key-reference assertions per ruling. No other production change. Active plug-in docs updated.
 
 Evidence script adds glm-tool mode: one GLM turn only, same pinned verified binary and isolated setup, distinct attempt3-tool root, active built-in read tool but stand-in still returns text only. Capture tool_stream and tool names before/after proxy, never header/auth values. No actual tool execution is demanded by stand-in, so this checks wire-field survival, not arbitrary tool-call semantics. Rehash before launch, no fetch,30s catalog readiness with counts,50s timeout-k10 for one turn,own descendants restored nice19. Commit untested before cleared run; then proxy-bundle and proxy-harness each whole once. No other runs/mutants.
+
+### sage-4331 single active-tool GLM turn and two-file results
+
+Tested ef0939e35. No fetch. P1 verifyInstall rehashed existing binary/panel and matched both PIN hashes before use. One isolated glm-tool run, unshare-rn interfaces[lo],empty private HOME/auth,sanitized env. Readiness poll1 not-ready/count0;poll2 status200/count2. One pi invocation with one active builtin read tool,exit0,markerP332_OFFLINE_OK; no rerun. Outer timeout-k10/140s,inner120s,pi50s;nice19 explicitly at entry/child starts and guard restores observed16→19 three times. No production API chosen by script.
+
+Plain request/response evidence (full safe receipt evidence/glm-tool.txt):
+- pi POST/v1/chat/completions modelglm-fixture,streamtrue,tools=[read],tool_stream=true,thinking={type:enabled,clear_thinking:false},reasoning_effort=low,max_completion_tokens=16384.
+- pinned proxy200 completion stream,marker.
+- stand-in receives POST/v1/chat/completions,tools=[read],tool_stream=true and same thinking/reasoning_effort fields;max_tokens=16384. Thus zaiToolStream-derived tool_stream survives this proxy route. No tool invocation or tool-call delta sequence was requested by the stand-in; this proves field survival with active tool, not every tool-call streaming semantic.
+
+Then exactly once each whole file, fresh gXXX TMPDIR,nice19,timeout-k10/180s:
+- packages/harness/tests/proxy-bundle.test.mjs00:14:20Z:7pass,0fail,0skip,exit0.
+- packages/plugins/tests/proxy-harness.test.mjs00:14:20Z:5pass,0fail,0skip,exit0.
+
+No failures, no extra runs/mutants. Cleanup managed children0. Own-worktree/evidence-root sweep found only same six pre-existing tool services,no proxy/pi/test/timeout/chromium leftovers. Lane B free. P3 now waits for P2/row331 landing before rebase,mutants,Gladstone review and further gates; no landing/merge requested or performed.

agents/fethry/work/queue-332/api-choice/proxy-bundle.txt

@@ -0,0 +1,19 @@
+00:14:20Z
+nice=19
+✔ P3 Anthropic route uses messages at proxy root and the declared model id (12.070131ms)
+✔ P3 omitted route refuses proxy-route-missing before model config generation (2.197623ms)
+✔ P3 GLM route uses completions at proxy /v1 with explicit z.ai compatibility (3.390562ms)
+✔ P3 GPT route selects a stack-owned provider instead of pi Codex OAuth (4.240948ms)
+✔ P3 a declared Codex route generates only the single build-time responses provider (3.37448ms)
+✔ P3 models file is private, references the one session key, and stores no executable values or keys (3.305299ms)
+✔ P3 every launch regenerates proxy port, declared models and own key without caching (6.287121ms)
+ℹ tests 7
+ℹ suites 0
+ℹ pass 7
+ℹ fail 0
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 128.759611
+exit=0
+00:14:20Z

agents/fethry/work/queue-332/api-choice/proxy-harness.txt

@@ -0,0 +1,17 @@
+00:14:20Z
+nice=19
+✔ P3 Claude env receives the session proxy URL and key, without another agent or host credential (3.096286ms)
+✔ P3 Claude preflight accepts a proxy key without a host provider credential (1.305258ms)
+✔ P3 pi refuses no-session-credential with no provider key (1.23137ms)
+✔ P3 pi refuses no-session-credential with only a Claude OAuth token (3.659409ms)
+✔ P3 pi accepts its own provider key or a session proxy key (0.590987ms)
+ℹ tests 5
+ℹ suites 0
+ℹ pass 5
+ℹ fail 0
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 68.648537
+exit=0
+00:14:20Z

agents/fethry/work/queue-332/evidence/glm-tool.txt

@@ -0,0 +1,8 @@
+{"phase":"isolation","interfaces":["lo"],"enteredNice":19,"nice":19,"version":"8.0.42","binarySha256":"74661f8cf949977870231359e8bf4344e07946766a673c08fa25a3e5f72fc95e","panelSha256":"8c9c6b7d3fb311e6aad5d06e1993e53a2d28389e640448a1bc6304ed20b057b5","reverified":true}
+{"phase":"readiness","poll":1,"status":"not-ready","modelCount":0}
+{"phase":"readiness","poll":2,"status":200,"modelCount":2}
+{"phase":"catalog","modelIds":["gpt-fixture","glm-fixture"]}
+{"phase":"turn","name":"glm-completions-tool","api":"openai-completions","exit":0,"signal":null,"marker":true,"piRequests":[{"method":"POST","path":"/v1/chat/completions","keys":["max_completion_tokens","messages","model","reasoning_effort","store","stream","stream_options","thinking","tool_stream","tools"],"model":"glm-fixture","stream":true,"roles":["developer","user"],"inputTypes":"undefined","thinking":{"type":"enabled","clear_thinking":false},"reasoning_effort":"low","tool_stream":true,"toolNames":["read"],"stream_options":{"include_usage":true},"max_completion_tokens":16384}],"proxyResponses":[{"request":0,"status":200,"events":[],"marker":true}],"upstreamRequests":[{"method":"POST","path":"/v1/chat/completions","keys":["max_tokens","messages","model","reasoning_effort","store","stream","stream_options","thinking","tool_stream","tools"],"model":"glm-fixture","stream":true,"roles":["developer","user"],"inputTypes":"undefined","thinking":{"type":"enabled","clear_thinking":false},"reasoning_effort":"low","tool_stream":true,"toolNames":["read"],"stream_options":{"include_usage":true},"max_tokens":16384}],"nice":19}
+{"phase":"conclusion","limitation":"Stand-in openai-compatibility credential only; no Codex OAuth credential or real-auth turn. This script never selects the production API."}
+{"phase":"nice-corrections","restoredTo":19,"observedBeforeRestore":{"16":3}}
+{"phase":"cleanup","remainingManagedChildren":0}

docs/SESSIONS.md

@@ -1715,3 +1715,5 @@ are never rewritten or removed; corrections are new entries.
 2026-10-24 | fethry (65673bd9-3bfa-412b-85d5-9b364187d8b6) | row332 route required sage-4328 |9563fbc8b proxy-launch6/0,launch-settings18/0,bundle6/1(expected GPT PENDING),launcher35/0,zero skips; grep positive controls/old assertion preservation recorded; lane B free00:02:27Z UTC10-25,no test leftovers,no probes/fetches/mutants.
 
 2026-10-24 | fethry (65673bd9-3bfa-412b-85d5-9b364187d8b6) | row332 pinned offline evidence sage-4329/4330 | One verified fetch; first setup0turns wrong provider shape; statically diagnosed bundled panel schema then one corrected isolated attempt: both GPT shapes+GLM exit0/marker,thinking retained,tool_stream untested(no tools); GPT choice inconclusive/PENDING,no real auth; process cleanup0,lane B free.
+
+2026-10-24 | fethry (65673bd9-3bfa-412b-85d5-9b364187d8b6) | row332 sage-4331 API choice and GLM tool field | ef0939e35 responses constant per lead reasoning tie-break;rehash/no-fetch isolated GLM one turn0/tool_stream retained;bundle7/0/0skip,proxy-harness5/0/0skip;lane B free00:14:20Z UTC10-25,no test leftovers;wait row331 before further work.