Claude (review-harness setup) e346b54e0f Phase C — local-Ollama LLM review wired end-to-end

Implements PROMPT.md / docs/REVIEW_PIPELINE.md Phase 2:
- internal/llm/ollama.go — real Ollama provider:
  - HealthCheck probes /api/tags + a 1-token completion + a JSON-mode
    probe ({"ok": true} round-trip), populating the model-doctor.json
    schema documented in docs/LOCAL_MODEL_SETUP.md
  - Complete + CompleteJSON via /api/chat with stream=false
  - think=false set for ALL completions (qwen3.5:latest is reasoning-
    capable but the inner-loop hot path wants direct answers, not
    reasoning traces consuming the token budget — same finding as
    the Lakehouse-Go chatd 2026-04-30 wave)
- internal/llm/review.go — Reviewer wrapper:
  - 2-attempt flow: prompt → parse → repair-prompt → parse
  - Strict JSON shape enforced; markdown fences stripped before parse
  - Severity normalized to enum; out-of-range confidence clamped
  - Per-file chunking (file-level for v0; function-level Phase D+)
  - Bounded by review-profile max_file_bytes + max_llm_chunk_chars
- pipeline.go — Phase 2 wired between static scan + report gen:
  - --enable-llm flag opts in (off by default — static-only is
    cheaper and faster)
  - Raw output ALWAYS saved to llm-findings.raw.json (forensics)
  - Normalized findings → llm-findings.normalized.json
  - LLM findings merged into the report findings list (sourced
    "llm" so consumers can filter)
  - Receipts honestly mark phase status: "ok" | "degraded" | "skipped"
- cli model doctor — real probes replace the Phase A stub.

Verified:
- model doctor: status="ok" with qwen3.5:latest + qwen3:latest both
  loaded, basic_prompt_ok=true, json_mode_ok=true
- insecure-repo with --enable-llm: 9 LLM findings; qwen3.5 correctly
  flagged SQLi, RCE, hardcoded credentials as critical with verbatim
  evidence; 27s wall for 3 chunks
- clean-repo with --enable-llm: 0 LLM findings, 4 parsed chunks, 2.8s
- self-review with --enable-llm: 77 LLM findings + 83 static; 3 of
  ~30 chunks needed retry (PROMPT.md, REPORT_SCHEMA.md,
  SCRUM_TEST_TEMPLATE.md — all eventually parsed); 5min wall

go vet + go test -short clean. Fixture stray.go now `package fixture`
so go-tooling doesn't choke on the orphan.

Phase D (validator cross-check) + Phase E (memory + diff/rules
subcommands) remain.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

2026-04-30 01:13:39 -05:00

2.5 KiB

Raw Blame History

Scrum Test — clean-repo

Generated: 2026-04-30T06:06:53.895008668Z Branch: main · Commit: 70d68757f7

Verdict

production-ready — static scan + LLM review found no issues. Re-validate after every wave.

Evidence

repo path: /home/profit/share/local-review-harness-full-md/tests/fixtures/clean-repo
file count: 4
languages: TypeScript (2), Markdown (1), JSON (1)
dependency manifests: 1 (package.json)
test files/dirs: 1

Confirmed Risks

No confirmed risks at static-scan level. (LLM review may surface more.)

Suspected Risks

None.

Blocked Checks

None.

Sprint Backlog

Sprint 0 — Reproducibility Gate

Wire just verify (or equivalent) to run the static checks before every commit/PR.
Add a CI step that fails on critical findings.

Sprint 1 — Trust Boundary Gate

Confirm auth posture for any mutation endpoint flagged as exposed.
Replace raw SQL interpolation with parameterized queries.

Sprint 2 — Memory Correctness Gate

(Phase E) Wire append-only .memory/ writes for known-risks + fixed-patterns.
Add a regression test that re-runs the harness and asserts no regression in confirmed-finding count.

Sprint 3 — Agent Loop Reality Gate

(Phase C) Wire local-Ollama LLM review.
(Phase D) Validator pass cross-checks every LLM finding against repo evidence.

Sprint 4 — Deployment Gate

Ship the harness as a single static binary (go build -o review-harness).
Document operator runbook (model setup, profile editing, output retention).

Acceptance Gates

Each gate must be testable. Format: command + verifiable post-condition.

Reproducibility: review-harness repo . exits 0; reports/latest/repo-intake.json exists with non-zero file_count.
No false positives on a clean fixture: review-harness repo tests/fixtures/clean-repo produces zero confirmed findings.
Every documented static check fires on the insecure fixture: jq '[.findings[] | .check_id] | unique | length' reports/latest/static-findings.json ≥ 8.
Receipts are honest about degraded phases: jq '[.phases[] | select(.status == "degraded")]' reports/latest/receipts.json lists every skipped/stubbed phase.

Next Commands

Re-run after fixes: review-harness repo /home/profit/share/local-review-harness-full-md/tests/fixtures/clean-repo
Generate the full Scrum bundle: review-harness scrum /home/profit/share/local-review-harness-full-md/tests/fixtures/clean-repo

2.5 KiB Raw Blame History