Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/UniM0cha/self-improving-skillsnpx agentmods add skills/unim0cha/self-improving-skills/self-improving-skills-reviewWrote 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/unim0cha/self-improving-skills/self-improving-skills-review)<a href="https://agentmods.dev/skills/unim0cha/self-improving-skills/self-improving-skills-review"><img src="https://agentmods.dev/badge/skills/unim0cha/self-improving-skills/self-improving-skills-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/unim0cha/self-improving-skills/self-improving-skills-review"><img src="https://agentmods.dev/badge/skills/unim0cha/self-improving-skills/self-improving-skills-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 | $0.00068 | $0.01309 |
| Opus 5.5 | $0.00027 | $0.00524 |
| Sonnet 5.5 | $0.00014 | $0.00262 |
| Haiku 4.5 | $0.00007 | $0.00131 |
Grade A, and why
self-improving-skills-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 today.
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 — 119 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Codex Self-Improvement Review
Use this skill to decide whether a recent interaction should update Codex's reusable skills. Treat skills as procedural memory: they record how to do a class of task, not one-off facts.
Review Policy
Inspect the current thread evidence before writing anything. Then walk this ladder and act on the EARLIEST rung that applies (Hermes review-priority order):
- Patch a skill that was in play this session. If a skill was loaded or
invoked (
$skill-name) during this thread and it turned out wrong, stale, or incomplete — that is the skill to fix first. It routed the work, so its gaps are exactly what future sessions will hit. - Extend an existing relevant skill. Prefer adding the gotcha, corrected step, or example to a skill that already governs this class of task over creating a new narrow sibling.
- Embed the corrected preference in the governing skill. User corrections about workflow, tone, formatting, or verification belong in the body of the skill that governs that class of task — so the next session starts already corrected.
- Create a new skill — last resort. Only when the learning is class-level and likely to recur, and no existing skill covers it.
Additional rules:
- Name anti-patterns (a bad name means the knowledge is not class-level —
fall back up the ladder or capture nothing): a PR number, an error string, a
session label like
fix-X/debug-Y, a bare library name with no technique, anything tied to one instance. - Duplicates: report, don't merge. If you notice two skills covering the
same ground, note it in your output — consolidation is the
$codex-skill-curatorpass's job, not the review's. - Do not encode transient errors that resolved within the session (if a retry worked, the lesson — if any — is the retry pattern, not the failure), missing binaries, one-off paths, temporary outages, or negative claims like "tool X does not work". Setup failures are capturable only as the FIX (install command, config, env var).
- Never edit bundled/system/admin skills. Work in user or repo skill roots only.
- Keep changes small, make a backup first, and report the diff summary.
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.
- today First seen · 119 lines · 68 tokens per session scan A ac9d1c6b4666
self-improving-skills-review is a skill published in the GitHub repository UniM0cha/self-improving-skills (51 stars, last pushed 1mo ago), licensed MIT. It adds 68 tokens to every session and 1,309 once invoked, about $0.0003 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 2026-11-07.
self-improving-skills-review compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| self-improving-skills-reviewUniM0cha/self-improving-skills | Skill | 51 | 1mo ago | MIT | A | 68 | Codex |
| notion-artifact-captureAtlasOmnia/hermes-custom-pack | Skill | 76 | 11d ago | MIT | A | 82 | Claude Code, Codex |
| libretranslate-markdown-i18npedroiff0/awesome-skills | Skill | 1 | 15d ago | MIT | A | 127 | Claude Code, Codex |
| gh-issuesSafeAI-Lab-X/ClawKeeper | Skill | 1.0k | 2mo ago | none | A | 116 | Claude Code |
| openclaw-pr-maintainerSafeAI-Lab-X/ClawKeeper | Skill | 1.0k | 2mo ago | none | A | 84 | Claude Code, Codex |
| grill-with-docspedroiff0/awesome-skills | Skill | 1 | 15d ago | MIT | A | 36 | Claude Code, Codex |
| google-workspaceAtlasOmnia/hermes-custom-pack | Skill | 76 | 11d ago | MIT | A | 24 | Claude Code, Codex |