Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/hhamja/claude-code-harness/loop-reviewnpx skills add hhamja/claude-code-harness --skill loop-reviewgit clone --depth 1 https://github.com/hhamja/claude-code-harnessWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00059 | $0.00895 |
| Opus 5 | $0.00030 | $0.00447 |
| Sonnet 5 | $0.00012 | $0.00179 |
| Haiku 4.5 | $0.00006 | $0.00089 |
Grade A, and why
loop-review scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 32 lines — stays where its author put it; the contents beside it link to each section on GitHub.
loop-review — comprehensive harness review
Review a target project's agent/loop harness against the control plane (${CLAUDE_PLUGIN_ROOT}/docs/loop-control-plane.md) — not just whether the architecture exists, but whether its gates actually hold. Orchestrator-worker + maker≠checker: you (the main agent) are the orchestrator only — you spawn read-only reviewers, adversarially reproduce their findings, synthesize, and write the report. You never grade your own reading.
Complementary to /loopy:loop-audit (grades an initialized loop's process) and loop-run --verify-only (grades the product). This one grades the architecture and its enforcement, and works on any project — including one with no .claude/loop/.
Target path ($ARGUMENTS): default ..
Procedure
- Fan out (parallel, read-only). Spawn together, each returning one report and modifying nothing:
loop-architect— "diagnose the harness at<path>— ETCLOVG coverage + L0–L5 maturity". The structured spine.design-critic— "adversarially red-team the harness at<path>: refute its compliance claims — gate bypasses, forgeable approvals, rubber-stamps, script bugs".- If
<path>/.claude/loop/exists, alsoauditor(process adherence). For a large harness, add a fewexplorerscouts split by cutting plane (hooks/CI, gate scripts, agents, state templates) so coverage does not rest on one context.
- Reproduce, don't trust (§3). A single checker rubber-stamps. For each
design-critichole a command can settle (e.g. a gate that should deny), reproduce it YOURSELF in a fresh temp dir (mktemp -d— never against the target's own files) by feeding the input to the actual script. Promote only reproduced holes to CONFIRMED; leave the rest PLAUSIBLE. Drop any that fail to reproduce. - Synthesize (you write everything — maker≠checker for the review itself). Merge into ONE report: an ETCLOVG table (PASS/PARTIAL/MISSING + cited evidence),
maturity: L<n>/5with the single capping requirement, holes ranked by exploitability (CONFIRMED/PLAUSIBLE, each with repro or reason), and priority fixes in build order (§9 — enforcement holes and verifier/holdout before observability before governance polish). OVERWRITE<path>/harness-review.mdwith a one-line timestamp header. It is human-facing — commit it likereview.md, never gitignore. - Next step. Point at the top fix. If the target has no loop, suggest
/loopy:loop-init.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 3d ago First seen · 32 lines · 0 tokens per session scan A 6b1b11f6c3ec
loop-review is a skill published in the GitHub repository hhamja/claude-code-harness (1 stars, last pushed 1mo ago), licensed MIT. It adds 59 tokens to every session and 895 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
git-commit-convention
写 git 提交信息、整理提交历史时使用——Conventional Commits 的格式、类型选择和拆分原则。.
delegate-to-deepseek
默认把中等及以下、批量、重复或机械任务作为完整逻辑单元派给 DeepSeek，并由主 Agent 独立验收。适用于批量改文件、扫日志、翻译、ETL、脚本、测试、文档、CRUD、单领域重构、单组件或单 endpoint。主 Agent 可基于上下文和失败代价调整派工策略；用户显式指令、安全、权限、隐私边界和派工后验证不可突破。DEEPSEEKMODE=off 时跳过。.
code-review-checklist
审查代码改动（review diff / PR）时使用——按固定清单过正确性、边界、错误处理和测试盲区，避免只看顺眼不顺眼。.
sruja-project
Procedural workflows for working with the Sruja codebase. Teaches AI editors how to add components, validate changes, and follow patterns.
grill-me
This skill ensures AI and human share a design concept before writing code.
sruja-harness
Verification harness for any AI agent skill. Use with any other skills; before marking a task done, run verify-task. On failure, record a correction learning. Ensures lint, tests, drift, and intent checks pass regardless of which skill generated the code.