majiayu000/spellbook / skills/review-agent-harness
review-agent-harness
VerifiedReview whether a repository's coding-agent harness can reliably carry work from intent through controlled execution, verification, delivery, and learning. Use when asked to assess agent readiness, repeated agent failures, Rules/Skills/Hooks/Memory effectiveness, missing validation or recovery loops, or whether a harness repair improved later outcomes. Do not use for code-only audits, AGENTS-only audits, individual skill reliability reviews, or executing the task itself.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 13.7 KB |
| agents/openai.yaml | config | 264 B |
| evals/evals.json | config | 7.8 KB |
| references/finding-contract.md | markdown | 6.8 KB |
| references/privacy-boundary.md | markdown | 2.2 KB |
| references/repair-loop.md | markdown | 2.9 KB |
| references/review-model.md | markdown | 4.9 KB |
| references/session-adapters.md | markdown | 2.5 KB |
| scripts/collect_evidence.py | script | 21.9 KB |
| scripts/harness_common.py | script | 14.1 KB |
| scripts/render_report.py | script | 19.9 KB |
| scripts/session_adapters.py | script | 16.6 KB |
| scripts/update_ledger.py | script | 15.1 KB |
| scripts/validate_findings.py | script | 29.6 KB |
references/privacy-boundary.md
View on GitHub# Privacy Boundary
## Default Boundary
Static mode reads only the authorized target repository. It does not discover
user-home Sessions, global settings, installed Plugins, Memory, browser data,
credentials, or sibling repositories.
Session-informed mode requires explicit authorization for each file or exact
root and one named provider. Authorization to inspect Codex does not authorize
Claude Code, and a project review does not authorize global Memory bodies.
## Durable Output Rules
Never persist:
- raw Prompt, assistant response, tool input, or tool output;
- secret values, environment dumps, credential files, or private keys;
- absolute user-home paths or private cache layouts;
- stable Session, thread, task, or tool-call identifiers;
- Memory titles, paths, or bodies;
- source JSONL file names.
The adapters may retain bounded counts, anonymous aliases, sanitized first
request summaries when explicitly enabled, and coarse tool categories such as
edit or validation. They must not retain raw commands.
Durable target binding uses only collector-produced hashes of local directory
identity and bounded target metadata. It never persists the absolute target
path. Rendering and ledger updates require the caller to supply `--target` so
the scripts can recompute the binding before writing.
## Redaction And Failure
Redact secrets, home paths, UUIDs, stable ids, Markdown destinations, and
injected context before evidence enters reconciliation. The findings validator
and renderer scan reader output again.
Malformed or provider-unsupported lines remain visible as anonymous counts and
warnings. If every authorized source is unreadable, malformed, or unsupported,
mark the Session stage `unobserved` or stop; do not replace it with a broader
search.
## Authority Escalation
Ask before:
- adding a Session root after exact files were authorized;
- enabling request summaries;
- reading Memory metadata or bodies;
- reading user-global Rules, Skills, Plugins, or Hooks;
- copying evidence outside the local target or environment-owned scratch.
Separate analysis authority from mutation authority. Reading a Hook or Skill
does not authorize editing, installing, disabling, or deleting it.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |