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 skills add minusblindfold/devloop --skill reviewgit clone --depth 1 https://github.com/minusblindfold/devloopWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/minusblindfold/devloop/review)<a href="https://agentmods.dev/skills/minusblindfold/devloop/review"><img src="https://agentmods.dev/badge/skills/minusblindfold/devloop/review/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/minusblindfold/devloop/review"><img src="https://agentmods.dev/badge/skills/minusblindfold/devloop/review.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What 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 5.1 | $0.00028 | $0.01100 |
| Opus 5.5 | $0.00011 | $0.00440 |
| Sonnet 5.5 | $0.00006 | $0.00220 |
| Haiku 4.5 | $0.00003 | $0.00110 |
Grade A, and why
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 2mo 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 — 92 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Execute each section in order. Copy the checklist and check off items as you complete them. Do not proceed past a Gate until verified.
Artifact: Write to .work/reviews/. Verify with ls before wrapping up.
Constraint: You MUST NOT modify project files. This skill produces a review artifact, not code changes.
Mode
Check .work/active/ for marker files. If exactly one exists and no $ARGUMENTS provided, auto-select that feature's slug. If multiple markers exist, list them, offer to archive any with date older than 30 days to .work/archive/, then ask. Arguments always override.
Parse $ARGUMENTS (or auto-selected slug): matches existing file in .work/reviews/ → re-entry. Matches plan + design → create (workflow mode). Set but no plan/design → create (standalone, using slug as context). Empty → detect branch; main: uncommitted only; otherwise: standalone using branch name.
You run in a forked context and cannot converse — when a choice is genuinely required (multiple current markers, ambiguous slug), stop and return the options to the caller instead of asking.
Task Progress:
- [ ] Determine mode (workflow / standalone / re-entry)
- [ ] If re-entry: read existing review, note prior findings
- [ ] Resolve rules (workflow: explicit from design specs; standalone: keyword from slug)
- [ ] Load implementation notes if present (workflow only)
- [ ] Load git diff (feature branch: main...HEAD + uncommitted; main: uncommitted only)
- [ ] Review diff against rules — cite rule, file, line for each finding
- [ ] Review diff for code quality and security issues
- [ ] Write artifact (create: new file; re-entry: append dated section)
- [ ] Gate: verify artifact with ls; update and archive active marker
- [ ] Wrap up: summarize findings, suggest fixes or confirm clean
Resolve rules and load context
Workflow mode: Read plan and design. Extract rule titles from task specs and apply the rules in devloop/rules/ whose H1 titles match. Read implementation notes matching slug — note intentional deviations (do not flag as violations).
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.
- 2mo ago First seen · 92 lines · 28 tokens per session scan A 865025df8f78
review is a skill published in the GitHub repository minusblindfold/devloop (1 stars, last pushed 3mo ago), licensed MIT. It adds 28 tokens to every session and 1,100 once invoked, about $0.0001 per session on Opus 5.5. 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
cad-reviewer-contract
Role contract preloaded at spawn by the cad-reviewer, cad-reviewer-low, cad-reviewer-medium, cad-reviewer-xhigh and cad-reviewer-max agents. Not for direct use: do not invoke it from a conversation.
spec-drift
Standalone plan-vs-code audit on any branch: runs /ship Step 8's plan-completion section from disk (hash-pinned) against an explicit plan and base. Report, JSON, exit code. Never edits code.
third-lens-review
After Claude self-pitfall + Codex on a ship-worthy/architecture/RT/security/contract change: run a third external model house (distant training distribution → different blind spots) on the patched artifact, then adversarial synthesis.
matilha-review
Use when user wants a quality review of a completed wave — will dispatch 6 parallel review agents (code-quality, UX, security, architecture, performance, docs) once Wave 3c ships.
quality-review
After a PRD, spec, or plan, before implementation: hunt perceived-quality pitfalls (silent failures, missing loading/empty states, error recovery, state drift) that make a product feel cheap. Complements pitfall-verification.
critic
Two-step self-review at every task lifecycle phase. Step 1 (this skill) runs in-context to gather session signals — files read vs grepped, user pushback, build evidence, degradation signs. Step 2 launches a forked subagent for unbiased code/plan review with fresh eyes. Combined report classifies findings as hard…