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 hyj28/claude-codex-pair --skill pair-reviewgit clone --depth 1 https://github.com/hyj28/claude-codex-pairWrote 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/hyj28/claude-codex-pair/pair-review)<a href="https://agentmods.dev/skills/hyj28/claude-codex-pair/pair-review"><img src="https://agentmods.dev/badge/skills/hyj28/claude-codex-pair/pair-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 | $0.00040 | $0.00547 |
| Opus 5 | $0.00020 | $0.00273 |
| Sonnet 5 | $0.00008 | $0.00109 |
| Haiku 4.5 | $0.00004 | $0.00055 |
Grade B, and why
pair-review scanned grade B with 1 finding 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.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
not be inherited from `~/.codex/config.toml`, which may be temporarily set to a laxer mode. How it starts
The opening of the file, as written. The whole thing — 40 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/pair-review — quick second pair of eyes on current changes
Lightweight standalone review, for when the full /pair workflow is not in play (e.g. the user or Codex wrote code directly and just wants it checked).
Arguments
/pair-review [codex] [base <branch>]
-
No args: review uncommitted changes (staged + unstaged + untracked) with a fresh Claude subagent.
-
base <branch>: reviewgit diff <branch>...HEADinstead of uncommitted changes. -
codex: ALSO run Codex's native reviewer and merge findings:- uncommitted scope:
codex exec -s read-only -c approval_policy=never review --uncommitted - base scope:
codex exec -s read-only -c approval_policy=never review --base <branch>
Pass
-s read-onlyexplicitly — reviewing never needs write access, and the sandbox must not be inherited from~/.codex/config.toml, which may be temporarily set to a laxer mode. - uncommitted scope:
Procedure
- Confirm there is actually something to review (
git status --porcelain/ the diff is non-empty). If empty, say so and stop. - Spawn a fresh-context subagent as the reviewer. It must be read-only (Read/Grep/Glob/read-only Bash; never edit). Give it the diff scope and instruct: report only real defects, each with file:line, a one-sentence claim, and a concrete failure scenario (inputs/state → wrong behavior). No style nits, no speculation. Judge the tests too, and treat a bad test as a defect: a test whose expected value is recomputed the way the code computes it, one that passes only because a collaborator was mocked out, one coupled to internals rather than a public boundary, or a diff that deletes or loosens assertions in the existing suite. Note behavior the diff adds with no test.
- If
codexwas requested, run the codex review command in parallel with a generous timeout. - Merge and dedupe findings, ranked most severe first. Note which reviewer found what only when the two disagree.
- Report the findings to the user. Do not fix anything unless the user then asks.
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 · 40 lines · 40 tokens per session scan B fb0a9fb1520c
pair-review is a skill published in the GitHub repository hyj28/claude-codex-pair (4 stars, last pushed 11d ago), licensed MIT. It adds 40 tokens to every session and 547 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
pr
Create a GitHub pull request for a verified change. Reads verify-report.md, opens a PR with title from proposal and body from spec, enforces the configured test policy. Requires the gh CLI authenticated. Run after /specclaw:verify produces a verify-report.md. For Azure DevOps PRs, use /specclaw:pr-azdo instead.
ship-pr
Run the project checks and open a pull request once they pass.
gh-address-comments
Help address review/issue comments on the open GitHub PR for the current branch using gh CLI; verify gh auth first and prompt the user to authenticate if not logged in.
github-pr-workflow
GitHub PR lifecycle: branch, commit, open, CI, merge.
gh-find-prs
Survey open Codewhale PRs and triage each for mergeability and disposition against the real landing branch.
hunk-release
Prepares, publishes, verifies, and curates Hunk releases. Use for release metadata, benchmarks, tags, publishing, release videos, backports, or recovery.