Privileged applier that LANDS meta-optimize / corpus-audit patches the user approved, with a fresh landing review and human approval.

MITAuto-check: notes

Install Meta Apply

skills CLI
$ npx skills add wanshuiyin/Auto-claude-code-research-in-sleep --skill meta-apply -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install wanshuiyin/Auto-claude-code-research-in-sleep meta-apply --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/wanshuiyin/Auto-claude-code-research-in-sleep.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/skills-codex/meta-apply .claude/skills/meta-apply && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
meta-apply
GitHub stars
17k
Used in
1 other repo
Token cost
~2.1k tokens
SKILL.md length
921 words
Files
1
Skills in repo
26
Repo updated
First seen
Licence
MIT

At a glance

Privileged applier that LANDS meta-optimize / corpus-audit patches the user approved, with a fresh landing review and human approval.

  • Works in 4 steps: Load staging + resolve the helper → Jury-at-landing for each requested patch → Land the survivors (Write/Edit only —… → …
  • The user says meta apply
  • SKILL.md covers The acquittal is generated…, The non-negotiable rules…, Workflow and Provenance is a receipt, not…, plus 2 more sections
  • Calls python3

What it does

Meta Apply is an agent skill from wanshuiyin/Auto-claude-code-research-in-sleep. Privileged applier that LANDS meta-optimize / corpus-audit patches the user approved, with a fresh landing review and human approval. Base Codex review is same-family provisional. Use when the user says "meta apply", "/meta-apply", "land the staged patches", "应用优化", after a /meta-optimize run.

Its SKILL.md is about 2.1k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

The repository describes itself as: ARIS ⚔️ (Auto-Research-In-Sleep) — Lightweight Markdown-only skills for autonomous ML research: cross-model review loops, idea discovery, and experiment automation. No framework… The licence is MIT.

When your agent uses it

  • The user says meta apply
  • Land the staged patches
  • After a /meta-optimize run

Example prompts

  • “meta apply”
  • “/meta-apply”
  • “land the staged patches”
  • “/meta-apply”

Requirements

  • Python 3
  • Pre-approved tools (allowed-tools): Bash(*), Read, Write, Edit, Grep, Glob

Workflow steps

4 steps, taken from the step headings in SKILL.md.

  1. Load staging + resolve the helper
  2. Jury-at-landing for each requested patch
  3. Land the survivors (Write/Edit only — never Bash)
  4. Report

What it can do on your machine

Read from SKILL.md and the folder at commit 26b95cf. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves these tools, so the agent can use them without asking each time:

    • Bash(*)
    • Read
    • Write
    • Edit
    • Grep
    • Glob

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • python3

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Meta Apply loads about 2.1k tokens when it runs. Until then it costs about 76 tokens; SKILL.md has 921 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~76
When it runs · the whole SKILL.md, loaded when a task matches
~2.1k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check: notes

The automated check noted patterns worth knowing about, such as sudo or a known installer.

  • NotePre-approves every shell command (allowed-tools: Bash)SKILL.md
    allowed-tools: Bash(*), Read, Write, Edit, Grep, Glob

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from wanshuiyin/Auto-claude-code-research-in-sleep at commit 26b95cf, republished under its MIT licence (© wanshuiyin). 921 words, ~2,076 tokens.

Download SKILL.mdSave it as .claude/skills/meta-apply/SKILL.md (or your agent's skills folder).
name
meta-apply
description
Privileged applier that LANDS meta-optimize / corpus-audit patches the user approved, with a fresh landing review and human approval. Base Codex review is same-family provisional. Use when the user says "meta apply", "/meta-apply", "land the staged patches", "应用优化", after a /meta-optimize run.
allowed-tools
Bash(*), Read, Write, Edit, Grep, Glob
argument-hint
[patch-number-or-all]

Meta-Apply: the privileged landing gate for self-modification patches

Codex assurance: a base landing review records review_independence: same-family and acceptance_status: provisional via stamp-provisional; the artifact is not auto-curatable. Only an overlay or deterministic verifier may produce accepted authorization.

This skill exists to enforce a privilege boundary. Producers like /meta-optimize (and, later, corpus-audit) are read-only — no Write/Edit, no apply step; they can only stage candidate patches under .aris/meta/pending/. This skill is the only place a staged patch becomes a real change to the corpus. Splitting "propose" from "land" across two skills with different tool grants is what makes "a loop cannot apply its own patch" structural rather than a sentence the producer is asked to obey.

It is human-invoked only. It runs when the user explicitly types /meta-apply after reading the producer's REPORT. That invocation IS the landing authorization (the locked 2026-05-30 decision: the human stays at the landing gate).

The acquittal is generated HERE, not relayed

The earlier design trusted a jury_verdict: pass field in the producer's manifest. That is forgeable — the producer writes that field, so a buggy/misaligned producer could stage a killed patch labelled "pass." So this skill does not trust any producer-written verdict. It runs the fresh landing reviewer itself, at landing time, on the actual staged diff (fresh reviewer, paths-only). The verdict is produced where it is consumed, by the privileged human-invoked skill — nothing to forge.

The non-negotiable rules (jury-at-landing, reject-default)

For each staged patch the user asks to land, in order — any failure ⇒ skip & report, never silently apply:

  1. The human named THIS patch. Apply only patches the user listed (/meta-apply 1,3 or all); default to applying nothing.
  2. Fresh landing review PASS, obtained now. Spawn a fresh gpt-6-astra reviewer via spawn_agent (reasoning_effort: ultra, read-only, paths-only per reviewer-independence.md) on the staged .diff + its target. Ask: does this change improve the harness without regressions; PASS or KILL + one-line reason. Include the scope-limits block from review-scope-limits.md: this jury judges ARIS's own mechanism, so an over-defensive KILL permanently blocks a good patch. The block bans proposing new hash binding — it is not a reason to KILL a patch that touches the existing provenance stamp. KILL ⇒ refuse. The human cannot override a KILL — they may only pick among reviewer-PASSED survivors.
  3. Record the review class honestly. Base Codex review is same-family and lands only with stamp-provisional; it can complete this explicit human-invoked operation but does not authorize future auto-curation. A Claude/Gemini overlay or deterministic verifier uses strict stamp and may record accepted. See skill-governance.md.

Workflow

Step 0: Load staging + resolve the helper
bash
PENDING=".aris/meta/pending"
[ -d "$PENDING" ] || { echo "Nothing staged. Run /meta-optimize first."; exit 0; }
echo "Staged:"; cat "$PENDING/manifest.jsonl"

Resolve provenance.py through the Codex manifest:

bash
if [ -z "${ARIS_REPO:-}" ] && [ -f .aris/installed-skills-codex.txt ]; then
  ARIS_REPO=$(awk -F'\t' '$1=="repo_root"{print $2; exit}' .aris/installed-skills-codex.txt 2>/dev/null) || true
fi
PROVENANCE=""
[ -n "${ARIS_REPO:-}" ] && [ -f "$ARIS_REPO/tools/provenance.py" ] && PROVENANCE="$ARIS_REPO/tools/provenance.py"
[ -z "$PROVENANCE" ] && [ -f tools/provenance.py ] && PROVENANCE="tools/provenance.py"
[ -n "$PROVENANCE" ] || { echo "ERROR: provenance.py unresolved" >&2; exit 1; }
Step 1: Jury-at-landing for each requested patch

For every patch the user asked to land, read its staged .diff and target, then spawn the fresh reviewer jury (Rule 2) — paths-only, no producer reasoning, no prior-round context. Record {patch, jury_verdict, jury_review_id, one_line_reason}. Print a one-line result per patch (PASS → eligible / KILL → refused: <reason>).

The producer may have written an advisory pre-screen into the manifest to help the human read the REPORT — ignore it for the landing decision. Only this fresh verdict counts.

Show full SKILL.md (423 more words)Show less
Step 2: Land the survivors (Write/Edit only — never Bash)

For each patch that PASSED Step 1 and was named by the user:

  1. Back up the target to .aris/meta/backups/<date>/<target> (use the Write tool to copy contents; corpus paths are not Bash-writable when corpus_write_guard is active — and the applier should use Write/Edit for corpus mutation anyway).
  2. Apply the diff by Edit/Write on the target corpus file.
  3. Stamp provenance on the changed file. Base Codex uses:
    bash
    python3 "$PROVENANCE" stamp-provisional "$TARGET" --author "$AUTHOR" \
      --reviewer "$JURY_MODEL" --verdict-id "$JURY_REVIEW_ID"
    This records review_independence: same-family and acceptance_status: provisional; is_auto_curatable remains false. If the active overlay produced a cross-family result, use strict stamp instead.
  4. Log to .aris/meta/optimizations.jsonl: {ts, patch, target, author_model, reviewer_model, jury_review_id, applied: true}.
Step 3: Report

Per patch: LANDED <target> (+ backup path + provenance sidecar) or REFUSED <patch>: <reason>. Remove landed patches from .aris/meta/pending/. Remind the user a landed patch is revertable from its backup, and to test the changed skill next run.

Provenance is a receipt, not an acquittal of correctness

A stamp records that a change passed a process (fresh landing review + human landing), not that it is correct. To prevent "approved-but-wrong with a stamp that vouches for it" (false-authority laundering — worse than no stamp, because a later auto-curator reads it as evidence):

  • The stamp carries verdict_id (auditable review) + content_hash (a later hand-edit invalidates it).
  • Recommended (not yet built): a TTL forcing re-review of long-lived auto-authored artifacts, and a behavioral auditor that REVOKES a stamp when a landed skill misbehaves. Track as follow-up; never treat a stamp as permanent truth.

Key Rules

  • Human-invoked only. Never run as a side-effect of another skill or a hook.
  • Jury-at-landing, reject-default, no override. The binding verdict is produced HERE on the staged diff; never trust a producer-written verdict; the human picks among survivors, never resurrects a KILL.
  • Never promote provisional to accepted. Base Codex always uses stamp-provisional; only an overlay or deterministic verifier may use strict stamp.
  • Corpus mutation goes through Write/Edit (reviewable, attributable), not Bash. The corpus_write_guard hook (if installed) additionally denies Bash corpus writes — it does NOT gate Write/Edit, so it does not by itself stop this skill from editing the corpus; the jury-at-landing + stamp discipline above is what governs Write/Edit mutations (that discipline is procedure, not a hook-enforced mechanism).
  • Back up before every mutation. Reversible by construction.
  • Only land staged patches. Applies what producers staged in .aris/meta/pending/; invents nothing of its own.

Review Tracing

Save each landing-jury reviewer call's trace per review-tracing.md to .aris/traces/meta-apply/<date>_run<NN>/ — the acquittal that landed a corpus change must be forensically recoverable.

© wanshuiyin, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in skills/skills-codex/meta-apply of wanshuiyin/Auto-claude-code-research-in-sleep.

Open the folder on GitHubat commit 26b95cf

Used in 1 other repository

We found 1 copy of this SKILL.md (exact, near-identical or edited) in other folders, from 1 other GitHub owner. This page covers the copy in wanshuiyin/Auto-claude-code-research-in-sleep, which our catalogue first saw on October 7, 2026.

Compare with similar skills

Meta Apply next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Meta Apply compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Meta Apply this skillwanshuiyin/Auto-claude-code-research-in-sleep17k1 repos~2.1kAutomated safety check: NotesMIT
Landing Optimizeraaron-he-zhu/aaron-marketing-skills2.9k—~3.3kAutomated safety check: PassApache-2.0
Landingalirezarezvani/claude-skills28k—~3.8kAutomated safety check: PassMIT
SQL Optimizationgithub/awesome-copilot40k2 repos~2.3kAutomated safety check: PassMIT
Landing Optimizeraiskillstore/marketplace433—~2.3kAutomated safety check: PassApache-2.0
Agent Performance Optimizerruvnet/ruflo74k2 repos~3.6kAutomated safety check: PassMIT

Similar skills

  • Landing Optimizer

    aaron-he-zhu/aaron-marketing-skills

    A skill your agent uses when the user asks to "optimize our landing page for influencer traffic", "fix our promo-code landing page", or "improve conversion from a creator campaign"; produces a…

    2.9k GitHub stars~3.3k tokensUpdated today
    Marketing & SEOAuto-check passed
  • Landing

    alirezarezvani/claude-skills

    Generates a premium single-page HTML landing page with 3D CSS animations, GSAP scroll effects, and mouse-parallax depth.

    28k GitHub stars~3.8k tokensUpdated 1 mo ago
    Frontend & DesignAuto-check passed
  • SQL Optimization

    github/awesome-copilot

    Official

    Universal SQL performance optimization assistant for comprehensive query tuning, indexing strategies, and database performance analysis across all SQL databases (MySQL, PostgreSQL, SQL Server…

    40k GitHub starsUsed in 2 repos~2.3k tokens
    DatabasesAuto-check passed
  • Landing Optimizer

    aiskillstore/marketplace

    A skill your agent uses when the user asks to "optimize our landing page for influencer traffic", "fix our promo-code landing page", or "improve conversion from a creator campaign"; produces a…

    433 GitHub stars~2.3k tokensUpdated yesterday
    Marketing & SEOAuto-check passed
  • Agent skill for performance-optimizer - invoke with $agent-performance-optimizer

    74k GitHub starsUsed in 2 repos~3.6k tokens
    Auto-check passed
  • Database Optimizer

    davila7/claude-code-templates

    Expert database optimizer specializing in modern performance tuning, query optimization, and scalable architectures.

    33k GitHub starsUsed in 8 repos~2.5k tokens
    DatabasesAuto-check passed

More from wanshuiyin/Auto-claude-code-research-in-sleep

All 26 skills in this repo
  • Academic Poster Builder

    wanshuiyin/Auto-claude-code-research-in-sleep

    Builds an academic conference poster as a single HTML and CSS file with measurement-based gates, real paper figures and a print-ready PDF rendered through headless Chromium.

    17k GitHub starsUsed in 1 repo~4.5k tokens
    Auto-check: notes
  • Proof Run Orchestrator

    wanshuiyin/Auto-claude-code-research-in-sleep

    Runs a mathematical proof project as a stateful pipeline of run directories: a local attempt first, then a manual GPT Pro handoff package, with an optional DeepSeek audit.

    17k GitHub starsUsed in 1 repo~4.7k tokens
    Auto-check passed
  • Render HTML

    wanshuiyin/Auto-claude-code-research-in-sleep

    Render an ARIS Markdown / JSON artifact (IDEAREPORT, AUTOREVIEW, KILLARGUMENT, PAPERPLAN, research-wiki state, etc.) into a single-file HTML view designed for human reading.

    17k GitHub starsUsed in 1 repo~5.4k tokens
    Auto-check: notes
  • Experiment Audit

    wanshuiyin/Auto-claude-code-research-in-sleep

    Audit experiment integrity before claiming results. An agent skill from wanshuiyin/Auto-claude-code-research-in-sleep.

    17k GitHub starsUsed in 1 repo~2.7k tokens
    Auto-check: notes
  • Integrity Forensics

    wanshuiyin/Auto-claude-code-research-in-sleep

    Run the Anti-Autoresearch integrity-forensics DETERMINISTIC slice (numeric core + rules-only reporter) against a paper via a SHA-pinned thin launcher, then convert the verdict into a typed policy…

    17k GitHub starsUsed in 1 repo~1.5k tokens
    Auto-check passed
  • Interview Cheatsheet

    wanshuiyin/Auto-claude-code-research-in-sleep

    Generate a long-form Chinese interview-prep cheat sheet on a specific ML/LLM topic — formulas with derivations, from-scratch PyTorch code, comparison tables, and 25 高频面试题 (L1 必会 / L2 进阶 / L3 顶级 lab).

    17k GitHub starsUsed in 1 repo~3.3k tokens
    Auto-check: notes

Questions about Meta Apply

What does Meta Apply do?

Privileged applier that LANDS meta-optimize / corpus-audit patches the user approved, with a fresh landing review and human approval. Meta Apply is an agent skill from wanshuiyin/Auto-claude-code-research-in-sleep. Privileged applier that LANDS meta-optimize / corpus-audit patches the user approved, with a fresh landing review and human approval.

When should I use Meta Apply?

Meta Apply fits situations like: the user says meta apply; land the staged patches; after a /meta-optimize run.

How do I install Meta Apply in Claude Code?

Run `npx skills add wanshuiyin/Auto-claude-code-research-in-sleep --skill meta-apply -a claude-code`. Or copy the skill folder (skills/skills-codex/meta-apply in wanshuiyin/Auto-claude-code-research-in-sleep) into .claude/skills/meta-apply in your project. Claude Code loads it when a task matches its description.

How do I install Meta Apply in Codex?

Run `npx skills add wanshuiyin/Auto-claude-code-research-in-sleep --skill meta-apply -a codex`. Or copy the skill folder (skills/skills-codex/meta-apply in wanshuiyin/Auto-claude-code-research-in-sleep) into .agents/skills/meta-apply in your project. Codex loads it when a task matches its description.

Can I use Meta Apply in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add wanshuiyin/Auto-claude-code-research-in-sleep --skill meta-apply -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/meta-apply, .gemini/skills/meta-apply, .github/skills/meta-apply and .opencode/skills/meta-apply in your project.

What does Meta Apply need to run?

Going by SKILL.md and its folder, Meta Apply needs the command-line tools its instructions call (python3). Our summary lists: Python 3. Its frontmatter pre-approves these tools: Bash(*), Read, Write, Edit, Grep, Glob.

Does Meta Apply access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Meta Apply safe to install?

Our automated static check of SKILL.md found notes only (pre-approves every shell command (allowed-tools: bash)), nothing it rates as a warning. It is not a guarantee. Review the folder before installing.

What licence does Meta Apply use?

Meta Apply is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Meta Apply use?

About 2.1k tokens (SKILL.md is roughly 8.3k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Meta Apply?

Skills that share tags, products or a category with Meta Apply: Landing Optimizer (aaron-he-zhu/aaron-marketing-skills, 2.9k stars), Landing (alirezarezvani/claude-skills, 28k stars), SQL Optimization (github/awesome-copilot, 40k stars) and Landing Optimizer (aiskillstore/marketplace, 433 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Meta Apply?

wanshuiyin (a GitHub user) maintains it in wanshuiyin/Auto-claude-code-research-in-sleep, which has 17,205 GitHub stars. The repository holds 26 skills in this directory. The repository was last updated on October 7, 2026.

Source: wanshuiyin/Auto-claude-code-research-in-sleep on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.