41b0a99ed2
Surfaced by today's untracked-files audit. None of these are accidents —
multiple are referenced by name in CLAUDE.md and memory files but were
never added.
Categories:
- docs/PHASE_AUDIT_GUIDE.md (106 LOC) — Claude Code phase audit guidance
- ops/systemd/lakehouse-langfuse-bridge.service — Langfuse bridge unit
- package.json — top-level npm manifest
- scripts/e2e_pipeline_check.sh + production_smoke.sh — real test scripts
- reports/kimi/audit-last-week*.md — the "Two reports live" CLAUDE.md cites
- tests/multi-agent/scenarios/ — 44 staffing scenarios (cutover decision A)
- tests/multi-agent/playbooks/ — 102 playbook records
- tests/battery/, tests/agent_test/PRD.md, tests/real-world/* — real tests
- sidecar/sidecar/{lab_ui,pipeline_lab}.py — 888 LOC dev-only UIs that
remain in service post-sidecar-drop (commit ba928b1 explicitly kept them)
Sensitivity check: scenarios use synthetic company names ("Heritage Foods",
"Cornerstone Fabrication"); audit reports describe code findings only;
no PII or secrets surfaced.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
80 lines
3.8 KiB
JSON
80 lines
3.8 KiB
JSON
[
|
||
{
|
||
"date": "2026-04-21",
|
||
"client": "Riverfront Steel",
|
||
"cities": "Toledo",
|
||
"states": "OH",
|
||
"events_total": 5,
|
||
"events_ok": 0,
|
||
"checkpoint_count": 2,
|
||
"model": "gpt-oss:20b",
|
||
"cloud": false,
|
||
"lesson": "** \nBefore any tool call, run the JSON payload through a linter and escape all single quotes or use parameterized queries to avoid syntax errors. Validate the entire JSON structure—especially braces and commas—before sending it to the executor. Pre‑fetch the required pool data for each role so that the `pool` field is populated, preventing the “pool=?” placeholder that caused the failures. This will eliminate the repeated JSON syntax failures seen in the Forklift Operator expansion and Warehouse Associate misplacement.",
|
||
"checkpoints": [
|
||
{
|
||
"after": "12:15",
|
||
"risk": "JSON syntax error in tool calls",
|
||
"hint": "For the next Forklift Operator expansion, escape single quotes in SQL query or use a parameterized query; validate JSON with a linter before execution."
|
||
},
|
||
{
|
||
"after": "15:45",
|
||
"risk": "Warehouse Associate JSON error",
|
||
"hint": "Escape quotes in SQL query; close JSON braces before sending to executor."
|
||
}
|
||
],
|
||
"created_at": "2026-04-21T00:43:44.018Z",
|
||
"file": "2026-04-21_Riverfront_Steel_1776732224018.json"
|
||
},
|
||
{
|
||
"date": "2026-04-21",
|
||
"client": "Riverfront Steel",
|
||
"cities": "Toledo",
|
||
"states": "OH",
|
||
"events_total": 5,
|
||
"events_ok": 1,
|
||
"checkpoint_count": 2,
|
||
"model": "gpt-oss:20b",
|
||
"cloud": false,
|
||
"lesson": "** \nBefore any baseline, recurring, or emergency fill, query the pool size and turn count; missing data causes the job to fail. Replicate the expansion logic that pulls pool and turns for all event types. If the GPT‑OSS model is unavailable, switch to a local fallback or log a warning instead of returning empty risk text. Validate that gaps are accounted for before committing the fill to avoid single‑gap failures.",
|
||
"checkpoints": [
|
||
{
|
||
"after": "12:15",
|
||
"risk": "generate returned empty text from gpt-oss:20b: {\"text\":\"\",\"model\":\"gpt-oss:20b\",",
|
||
"hint": "(T3 unavailable)"
|
||
},
|
||
{
|
||
"after": "15:45",
|
||
"risk": "generate returned empty text from gpt-oss:20b: {\"text\":\"\",\"model\":\"gpt-oss:20b\",",
|
||
"hint": "(T3 unavailable)"
|
||
}
|
||
],
|
||
"created_at": "2026-04-21T00:34:20.521Z",
|
||
"file": "2026-04-21_Riverfront_Steel_1776731660521.json"
|
||
},
|
||
{
|
||
"date": "2026-04-21",
|
||
"client": "Riverfront Steel",
|
||
"cities": "Toledo",
|
||
"states": "OH",
|
||
"events_total": 5,
|
||
"events_ok": 0,
|
||
"checkpoint_count": 2,
|
||
"model": "gpt-oss:20b",
|
||
"cloud": false,
|
||
"lesson": "** \nBefore any event, pre‑fetch the full pool roster and skill certification data for Toledo, OH; the missing pool data caused every shift to fail. Verify forklift operator certifications and tool compatibility ahead of time, as the misplacement risk highlighted a skill gap. Ensure the risk‑generation model (gpt‑oss:20b) is online or have a manual fallback; the empty response after the expansion shows a T3 unavailability that halted risk assessment. Apply these checks for baseline, recurring, expansion, emergency, and misplacement events to avoid the single‑gap failure pattern.",
|
||
"checkpoints": [
|
||
{
|
||
"after": "12:15",
|
||
"risk": "generate returned empty text from gpt-oss:20b: {\"text\":\"\",\"model\":\"gpt-oss:20b\",",
|
||
"hint": "(T3 unavailable)"
|
||
},
|
||
{
|
||
"after": "15:45",
|
||
"risk": "Forklift Operator skill gap",
|
||
"hint": "Verify forklift operator certification and tool compatibility for Toledo shift."
|
||
}
|
||
],
|
||
"created_at": "2026-04-21T00:39:54.327Z",
|
||
"file": "2026-04-21_Riverfront_Steel_1776731994327.json"
|
||
}
|
||
] |