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 i9wa4/dotfiles --skill subagent-reviewgit clone --depth 1 https://github.com/i9wa4/dotfilesWrote 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/i9wa4/dotfiles/subagent-review)<a href="https://agentmods.dev/skills/i9wa4/dotfiles/subagent-review"><img src="https://agentmods.dev/badge/skills/i9wa4/dotfiles/subagent-review.svg" alt="Measured on agentmods" 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 | $0.00049 | $0.00498 |
| Opus 5 | $0.00024 | $0.00249 |
| Sonnet 5 | $0.00010 | $0.00100 |
| Haiku 4.5 | $0.00005 | $0.00050 |
Grade A, and why
subagent-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 8d 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.
What it actually says
Subagent Review
1. Ledger
- Before review, guardian freezes target, criteria, scope, and the design/public boundary. Broadening is recorded, never inferred.
- Guardian owns a never-reused ledger (
F-001); use the contract for required fields and reports. - Guardian alone decides materiality and convergence; packets supply observations, evidence, and risk signals. Nits never gate rework. Invalid packets, missing evidence, authority gaps, cap exhaustion, and criterion changes are guardian-recorded states.
2. Review to Convergence
- Run cheapest verification. Guardian defines the required perspective set; critic is guardian-specified and never self-selects. Waves are parallel.
- Deduplicate. Guardian MUST request critic for substantive review; critic returns its set plus self-review; guardian aggregates both sets. Reviewers never edit, commit, push, approve, or decide materiality. See the packet schema and blind-projection contract in contract.
- Each round reconciles every prior ID: recurrence, regression, new discovery, closure, supersession, accepted risk, or unchanged; record changed criteria.
- Review the full change, then deduplicate material findings and freeze IDs, closure conditions, and reproducible verification in one executor batch. After rework, verify and review only that set unless scope/criteria changed.
- Repeat batched fix → verification → diff-review until closure, then final
confirmation. Cap: initial plus two failed reworks. At cap/unavailable
evidence,
BLOCKEDincludes IDs, gap, authority needed, and next owner. Quarantine invalid packets before a round; they consume no rework slot.
3. Round Report
Each round follows the contract: frozen scope/criteria/targets, ledger, evidence, batch, verdict, and next action.
What ships with it
13 files 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.
- evals/eval.yaml 520 B
- evals/tasks/negative-trigger-1.yaml 457 B
- evals/tasks/negative-trigger-2.yaml 445 B
- evals/tasks/positive-trigger-1.yaml 411 B
- evals/tasks/positive-trigger-2.yaml 475 B
- evals/tasks/positive-trigger-3.yaml 639 B
- references/review-topology.md 6.1 KB
- references/reviewer-architecture.md 1.5 KB
- references/reviewer-code.md 1.4 KB
- references/reviewer-data.md 1.5 KB
- references/reviewer-historian.md 2.5 KB
- references/reviewer-qa.md 1.4 KB
- references/reviewer-security.md 1.4 KB
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.
- 8d ago First seen · 45 lines · 49 tokens per session scan A f826695e8bbe
subagent-review is a skill published in the GitHub repository i9wa4/dotfiles (11 stars, last pushed today), licensed MIT. It adds 49 tokens to every session and 498 once invoked, about $0.0002 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 2026-08-30.
Other skills, from other repositories
pr-style
This skill should be used when creating or editing a GitHub pull request title or description, regardless of tool (gh CLI, GitHub MCP tools, API). Defines PR title format and body format with Why/What/Notes sections, and prompts for the Jira ticket number.
drift-check
Pre-PR advisory check for deviations from the project spec. Read-only analysis.
pi-review
A process for getting an independent review of a code or design change from a separate pi review session. The review uses either the sol or luna model and checks the repository or supplied document.
audit
Eight audit modes (codebase — owns "audit the codebase" — docs/process, performance, threat-model, motion, SEO, debt). Triggers "nuclear review", "whole codebase review", "adversarial audit", "fable audit", "correctness audit", "audit the docs", "doc drift", "process audit", "perf audit", "why is it slow", "bundle…
agent-review
Conducts independent, read-only code reviews of branches, diffs, commits, working trees, or pull requests using external CLI agents (Codex, Claude, agy) or a synthesized review workflow. Focuses strictly on actionable defects and evidence-based findings. Use when reviewing code changes, requesting a second opinion on…
review
Review the local unstaged/staged diff with Darkroom's checklist, or summarize active-PR feedback. Triggers "review my changes", "check this diff", "PR comments", "summarize PR feedback". Native /code-review handles a diff or PR.