跳到正文
cccskills
免费在 GitHub 公开

repomix-packer

Use when packaging a codebase to a single AI-friendly file for LLM analysis — local or remote, XML/Markdown/JSON, token counting, gitignore filtering, peer-side `repomix` CLI.

中文简介正在准备中。当前显示原文说明。

查看安装方法

包含的文件(2)

  • SKILL.md5.3 KB
  • evals/triggers.json1.1 KB

SKILL.md(原文)

安装前可以查看交给 AI 代理的指令内容。

Pinned upstream: repomix CLI (npm: repomix, brew: repomix). Re-verify per minor bump. Repomix is an optional dependency — this skill never installs it silently.

repomix-packer

Wraps the upstream yamadashy/repomix CLI for codebase-snapshot workflows: pack a local or remote repo into a single XML / Markdown / JSON file with token counts and secret detection, then feed it to an LLM for review, audit, or migration scoping.

When to use

  • Producing an LLM-ingestible snapshot of a repo (or a sub-tree) for review or audit.
  • Comparing two branches by packaging each and diffing the snapshots.
  • Pulling a remote third-party library into context without cloning.
  • Pre-flighting a token budget before sending a codebase to an LLM.

Do NOT use when:

  • You only need a few specific files — read them directly with view.
  • The snapshot will only feed a non-text format (PDF, image, audio) — route to markitdown.
  • The repo is sensitive and --no-security-check would be needed — STOP, route to a human.

Procedure: Snapshot a repo for LLM review

1. Inspect: verify repomix is installed (peer-side)

repomix --version

If the binary is missing, surface one of the install recipes and STOP — do not install silently:

# npm (preferred for project-local installs)
npm install -g repomix

# Homebrew (macOS / Linux)
brew install repomix

2. Decide local vs remote

# Local: pack the current directory.
repomix

# Remote shorthand: owner/repo
npx repomix --remote owner/repo

# Remote URL with a pinned commit
npx repomix --remote https://github.com/owner/repo/commit/<sha>

3. Filter the snapshot to the smallest useful slice

# Include patterns
repomix --include "src/**/*.php,*.md"

# Add ignore patterns on top of .gitignore
repomix -i "tests/**,*.test.js"

# Strip comments to save tokens
repomix --remove-comments

4. Pick the output format and destination

repomix --style markdown -o snapshot.md   # human-readable
repomix --style xml -o snapshot.xml       # default; clearest separators for LLMs
repomix --style json -o snapshot.json     # programmatic post-processing
repomix --copy                            # also copy to clipboard

5. Verify token budget and secrets

Repomix prints per-file and total token counts and runs Secretlint on the output. Check the totals against the target LLM context window:

ModelApprox context
Claude Sonnet 4.5~200K tokens
GPT-4 family~128K tokens
GPT-3.5~16K tokens

If Secretlint flags anything, STOP — sanitize the input or add the offending paths to .repomixignore before re-packing. Never use --no-security-check on an unfamiliar codebase.

6. Hand the snapshot to the consumer skill

Most workflows that call this skill pass the snapshot to:

Cite the snapshot path so the consumer skill can read it.

Output format

  1. The repomix invocation (one shell line, with all filters and the output path).
  2. The output file path + format + total token count.
  3. Any Secretlint findings, verbatim. Empty section if none.

Gotcha

  • --copy puts the entire snapshot on the clipboard — surprising on large repos. Prefer -o <path> for anything > a few KB.
  • --no-gitignore plus a wildcard include can pull in .env, vendor/, node_modules/ — never combine without a tight --include first.
  • Remote npx repomix --remote owner/repo defaults to the latest commit on the default branch — pin a commit SHA when reproducing a previous snapshot.
  • Token counts are LLM-tokenizer estimates, not exact — leave a 10–15% headroom under the model's documented context window.

Do NOT

  • Do NOT run repomix --no-security-check on an unfamiliar codebase.
  • Do NOT install repomix silently — surface the recipe and let the consumer install it.
  • Do NOT commit repomix-output.* artifacts — add the pattern to .gitignore.
  • Do NOT package .env, key material, or .git/ — adjust .repomixignore first.
  • Do NOT vendor repomix into the repo — it is a peer-side CLI.

Auto-trigger keywords

  • repomix
  • pack codebase
  • repository snapshot
  • llm context bundle
  • codebase to single file

Provenance

  • Upstream tool: https://github.com/yamadashy/repomix (MIT).
  • Adopted from: an external reference (internal provenance, redacted) — wrapper-style adoption, no upstream code vendored.
  • Provenance registry: agents/settings/contexts/skills-provenance.yml (entry: repomix).
  • Iron-Law floor: non-destructive-by-default, missing-tool-handling, tool-safety.

评价

还没有评价。欢迎分享你的使用感受。

评论

无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。

还没有评论。来写第一条吧。

提交后会立即公开。请勿填写邮箱、电话等个人信息。URL 最多 2 个。

0 / 1000 字

此表单受 reCAPTCHA 保护,适用 Google 的隐私权政策和服务条款。

同一仓库的技能

Use when reviewing UI for accessibility — WCAG 2.2 AA, keyboard nav, focus, ARIA, contrast, screen-reader semantics — even on 'is this a11y-OK?' or 'mach das barrierefrei'.

中文简介正在准备中。当前显示原文说明。

event4u-app/agent-config132026年10月22日 更新

Use when defining or auditing the activation event — aha-moment selection, retention correlation, falsifiable definition. Triggers on 'what is our aha moment', 'redefine activation'.

中文简介正在准备中。当前显示原文说明。

event4u-app/agent-config132026年10月22日 更新

Use when capturing an architectural decision — file naming, next ADR number, Status / Context / Decision / Consequences, index regen; fires even without saying 'ADR'.

中文简介正在准备中。当前显示原文说明。

event4u-app/agent-config132026年10月22日 更新

Adversarial critique — devil's advocate, stress-test, honest teardown ('poke holes', 'be brutal', 'was hältst du davon'); explicit request only. Routine code or design review → code-review.

中文简介正在准备中。当前显示原文说明。

event4u-app/agent-config132026年10月22日 更新

Use when reading, creating, or updating agent documentation, module docs, roadmaps, or AGENTS.md. Understands the full .augment/, agents/, and copilot-instructions structure.

中文简介正在准备中。当前显示原文说明。

event4u-app/agent-config132026年10月22日 更新

Use for an adversarial red-team / blue-team / auditor review of an AI agent's CONFIG + behaviour (rules, skills, MCP, hooks, permissions) — attack-chain → defensive-gap list, not a code audit.

中文简介正在准备中。当前显示原文说明。

event4u-app/agent-config132026年10月22日 更新

查看 event4u-app 的全部技能

报告此技能的问题