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

148 lines
5.4 KiB
JSON

{
"generated_at": "2026-04-30T06:06:33.240219171Z",
"findings": [
{
"id": "",
"title": "Hardcoded file path for secrets",
"severity": "high",
"status": "suspected",
"file": "src/handler.go",
"line_hint": "10",
"evidence": "const HARDCODED_PATH = \"/home/profit/secrets/key.pem\"",
"reason": "Hardcoding a file path for a private key exposes secrets and prevents proper secret management.",
"suggested_fix": "Move the path to an environment variable or a configuration file outside the source code.",
"source": "llm",
"confidence": 1,
"check_id": "llm.review"
},
{
"id": "",
"title": "Hardcoded server IP address",
"severity": "medium",
"status": "suspected",
"file": "src/handler.go",
"line_hint": "11",
"evidence": "const SERVER_IP = \"192.168.1.176\"",
"reason": "Hardcoding an IP address reduces portability and may leak internal network topology.",
"suggested_fix": "Read the server IP from an environment variable.",
"source": "llm",
"confidence": 1,
"check_id": "llm.review"
},
{
"id": "",
"title": "SQL Injection vulnerability",
"severity": "critical",
"status": "suspected",
"file": "src/handler.go",
"line_hint": "15-16",
"evidence": "q := fmt.Sprintf(\"SELECT * FROM users WHERE name = '%s'\", name)\ndb.Query(q)",
"reason": "Using string formatting to construct SQL queries directly exposes the application to SQL injection attacks.",
"suggested_fix": "Use parameterized queries with placeholders instead of string formatting.",
"source": "llm",
"confidence": 1,
"check_id": "llm.review"
},
{
"id": "",
"title": "Unsafe shell command execution",
"severity": "critical",
"status": "suspected",
"file": "src/handler.go",
"line_hint": "19-20",
"evidence": "exec.Command(\"bash\", \"-c\", cmd).Run()",
"reason": "Executing arbitrary shell commands without validation allows for remote code execution.",
"suggested_fix": "Validate and sanitize the input command strictly, or avoid using shell execution entirely.",
"source": "llm",
"confidence": 1,
"check_id": "llm.review"
},
{
"id": "",
"title": "Hardcoded API key",
"severity": "critical",
"status": "suspected",
"file": "src/handler.go",
"line_hint": "23",
"evidence": "const API_KEY = \"sk-1234567890abcdefABCDEFGHIJKLMNOPQRSTUV\"",
"reason": "Hardcoding an API key in source code exposes sensitive credentials to anyone with access to the repository.",
"suggested_fix": "Store the API key in a secure environment variable or secrets manager.",
"source": "llm",
"confidence": 1,
"check_id": "llm.review"
},
{
"id": "",
"title": "CORS misconfiguration allows cross-origin attacks",
"severity": "high",
"status": "suspected",
"file": "src/server.js",
"line_hint": "2",
"evidence": "res.setHeader(\"Access-Control-Allow-Origin\", \"*\");",
"reason": "Allowing all origins (*) exposes the API to cross-site request forgery and data theft from any website.",
"suggested_fix": "Restrict Access-Control-Allow-Origin to specific trusted domains or use credentials with a specific origin.",
"source": "llm",
"confidence": 1,
"check_id": "llm.review"
},
{
"id": "",
"title": "Hardcoded AWS access key in source code",
"severity": "critical",
"status": "suspected",
"file": "src/server.js",
"line_hint": "5",
"evidence": "const AWS_KEY = \"AKIAIOSFODNN7EXAMPLE\";",
"reason": "Hardcoded credentials in source code pose a severe security risk as they can be easily leaked and misused.",
"suggested_fix": "Use environment variables or a secure secrets manager to store AWS credentials.",
"source": "llm",
"confidence": 1,
"check_id": "llm.review"
},
{
"id": "",
"title": "Missing authentication on user creation endpoint",
"severity": "high",
"status": "suspected",
"file": "src/server.js",
"line_hint": "7",
"evidence": "app.post(\"/api/users\", function(req, res) { /* no auth */ });",
"reason": "The /api/users endpoint lacks authentication, allowing anyone to create or modify user accounts.",
"suggested_fix": "Implement authentication middleware to verify user identity before allowing POST requests.",
"source": "llm",
"confidence": 1,
"check_id": "llm.review"
},
{
"id": "",
"title": "Missing authentication on admin deletion endpoint",
"severity": "critical",
"status": "suspected",
"file": "src/server.js",
"line_hint": "8",
"evidence": "app.delete(\"/api/admin\", function(req, res) { /* no auth */ });",
"reason": "The /api/admin endpoint lacks authentication, allowing unauthenticated users to delete administrative resources.",
"suggested_fix": "Implement strict authentication and authorization checks for all admin endpoints.",
"source": "llm",
"confidence": 1,
"check_id": "llm.review"
}
],
"summary": {
"total": 9,
"confirmed": 0,
"suspected": 9,
"rejected": 0,
"critical": 5,
"high": 3,
"medium": 1,
"low": 0,
"by_source": {
"llm": 9
},
"by_check": {
"llm.review": 9
}
}
}