AI agent skills are reusable instruction sets that extend your assistant with domain-specific expertise, loaded on demand so they don't bloat your context. This repository covers marketing and engineering skills authored by @samber.
For Golang-related skills, please visit cc-skills-golang.
Important
Bootstrapped with Claude. Edited, tested, reviewed and reworked by a human.
No AI slop here. AI-made skills are useless.
Install with skills CLI (universal, works with any Agent Skills-compatible tool):
npx skills add https://github.com/samber/cc-skills --all
# or a single skill:
npx skills add https://github.com/samber/cc-skills --skill promql-cliClaude Code
/plugin marketplace add samber/cc
/plugin install cc-skills@samberOpenclaw
Copy skills into the cross-client discovery directory:
git clone https://github.com/samber/cc-skills.git ~/.openclaw/skills/cc-skills
# or in workspace:
git clone https://github.com/samber/cc-skills.git ~/.openclaw/workspace/skills/cc-skillsGemini CLI
gemini extensions install https://github.com/samber/cc-skillsUpdate with gemini extensions update cc-skills.
Cursor
Copy skills into the cross-client discovery directory:
git clone https://github.com/samber/cc-skills.git ~/.cursor/skills/cc-skillsCursor auto-discovers skills from .agents/skills/ and .cursor/skills/.
Copilot
Copy skills into the cross-client discovery directory:
/plugin install https://github.com/samber/cc-skills
# or
git clone https://github.com/samber/cc-skills.git ~/.copilot/skills/cc-skillsCopilot auto-discovers skills from .copilot/skills/.
OpenCode
Copy skills into the cross-client discovery directory:
git clone https://github.com/samber/cc-skills.git ~/.agents/skills/cc-skillsOpenCode auto-discovers skills from .agents/skills/, .opencode/skills/, and .claude/skills/.
Codex (OpenAI)
Clone into the cross-client discovery path:
git clone https://github.com/samber/cc-skills.git ~/.agents/skills/cc-skillsCodex auto-discovers skills from ~/.agents/skills/ and .agents/skills/. Update with cd ~/.agents/skills/cc-skills && git pull.
Antigravity
Clone and symlink into the cross-client discovery path:
git clone https://github.com/samber/cc-skills.git ~/.antigravity/skills/cc-skillsUpdate with cd ~/.antigravity/skills/cc-skills && git pull.
These skills are designed as atomic units, cross-referencing units. A skill may reference conventions defined in another. Installing only a subset will give you a partial and potentially inconsistent view of the guidelines. For best results, install all general-purpose skills together.
Each skill lives in skills/<name>/ with a SKILL.md entry point. The SKILL.md is kept small with internal references to advanced markdown files, so only the relevant content is loaded into context.
- Description (tok): weight of the
descriptionfield from YAML frontmatter, always loaded into Claude's context for skill triggering - SKILL.md (tok): weight of the full
SKILL.mdfile loaded when the skill triggers - Directory (tok): weight of all files in the skill directory (SKILL.md + referenced markdown files)
┌──────────────┐
│ Agent Skills │
└───────┬──────┘
│
┌──────────────────┴──────────┐
│ │
▼ ▼
┌──────────────────┐ ┌─────────────────────────────────────┐
│ Engineering │ │ Marketing & Content │
├──────────────────┤ ├─────────────────────────────────────┤
│ conventional-git │ │ linkedin-ghostwriting │
│ promql-cli │ │ press-release-writer │
│ chrome-extension │ │ humaniseur-fr │
│ crxjs │ │ copywriting-hooks │
│ │ │ copywriting-cta │
│ │ │ copywriting-prose-creator │
│ │ │ copywriting-tone-of-voice-creator │
│ │ │ site-launch-checklist │
└──────────────────┘ └─────────────────────────────────────┘
┌──────────────────────────┐ ┌───────────────────────────┐ ┌─────────────────────────────────────┐
│ Developer Relations │ │ Productivity │ │ Meta │
├──────────────────────────┤ ├───────────────────────────┤ ├─────────────────────────────────────┤
│ technical-article-writer │ │ deep-research │ │ snyk-agent-scan-compliance │
│ substack-ghostwriting │ │ influence-and-negotiation │ │ skill-progressive-disclosure-design │
│ │ │ training-report │ │ │
└──────────────────────────┘ └───────────────────────────┘ └─────────────────────────────────────┘
| Skill | Error rate gap | Description (tok) | SKILL.md (tok) | Directory (tok) |
|---|---|---|---|---|
conventional-git |
-36% | 73 | 1,446 | 1,446 |
promql-cli |
-39% | 87 | 1,257 | 6,248 |
chrome-extension |
— | 137 | 3,909 | 36,173 |
crxjs |
— | 185 | 2,697 | 2,697 |
linkedin-ghostwriting |
-31% | 87 | 1,045 | 1,993 |
substack-ghostwriting |
-51% | 205 | 3,315 | 10,720 |
technical-article-writer |
-33% | 147 | 2,295 | 5,737 |
press-release-writer |
-34% | 149 | 1,532 | 17,829 |
humaniseur-fr |
-14% | 249 | 5,895 | 5,895 |
snyk-agent-scan-compliance |
-27% | 144 | 1,884 | 5,592 |
training-report |
-61% | 179 | 2,392 | 8,222 |
deep-research |
-51% | 316 | 2,915 | 14,070 |
influence-and-negotiation |
-68% | 237 | 6,644 | 50,467 |
skill-progressive-disclosure-design |
-58% | 154 | 3,533 | 3,533 |
site-launch-checklist |
-59% | 222 | 6,772 | 13,350 |
copywriting-hooks |
— | 213 | 5,483 | 10,303 |
copywriting-cta |
— | 229 | 2,889 | 14,427 |
copywriting-prose-creator |
— | 201 | 4,786 | 23,565 |
copywriting-tone-of-voice-creator |
— | 223 | 4,030 | 22,286 |
| With Skill | Without Skill | Delta | |
|---|---|---|---|
| Overall | 99% | 50% | +49pp |
See EVALUATIONS.md for the full per-skill breakdown.
Chrome Extension development with Manifest V3. Covers architecture, messaging, CSP, storage, permissions, content scripts, service workers, and Chrome Web Store publishing.
Conventional Commits v1.0.0 for branch naming, worktree naming, and commit messages. Keeps git history parseable for auto-changelogs, SemVer bumps, and issue closing.
CRXJS Chrome extension development with Vite. True HMR for popup/options/content scripts/side panels, defineManifest for type-safe manifests, dynamic content script imports.
CLI for querying Prometheus and PromQL-compatible engines (Thanos, Cortex, VictoriaMetrics, Mimir). Instant/range queries, metric discovery, table/csv/json/graph output.
End-of-article CTA design. Interviews for objective, audience, and context, then produces copy + visual form + mechanism + A/B test plan + accessibility check.
Opening hooks for long-form articles (blog posts, Substack, LinkedIn long-form, newsletters). Proposes 3-4 hooks with distinct psychological levers: curiosity, contrarian, scene, promise, authority.
Codifies prose mechanics (lexicon, syntax, rhythm, structure) into a PROSE.md style guide. Three modes: BUILD fresh, ADAPT to new channel, AUDIT a corpus. For content factories, ghostwriting teams, house style guides.
Builds brand tone of voice guides (TONE.md) with voice attributes, modulation matrix, lexicon, and channel rules. Create from scratch or adapt to a new channel. For B2B, B2C, NGO, public sector, personal brands.
Removes AI-writing patterns from French text. Detects and fixes 27 patterns: AI vocabulary, anglicisms, formulaic openings, em dash abuse, sycophantic tone, typographic tells.
B2B LinkedIn ghostwriting via strategic interview, hook engineering, and post body. Turns founder stories and results into conversion-focused posts.
Professional press releases for any occasion and region. Product launches, funding, partnerships, crisis, M&A, events. Region-specific conventions (Europe, Americas, Asia, etc.).
Pre-launch checklist for shipping a website. Analytics (GA4, PostHog, GSC), legal, security headers, SEO/GEO, copywriting pass, OpenGraph, favicons, Lighthouse, accessibility. Opinionated for Cloudflare + Vercel.
Write, optimize, and grow Substack newsletters and articles. Covers topic selection, headline optimization, SEO, and audience building.
Technical articles and blog posts for senior developers. Opinionated, practical, example-driven content that teaches something or saves time.
Deep research with parallel web searches, multi-source validation, and cited Markdown reports. 11 research types: market, competitive, technical, academic, financial, legal, trend, community, and more.
Negotiation and influence toolkit for any interaction requiring agreement. B2B sales, salary, collective bargaining, hard 1:1s, recruitment, cross-cultural deals. Preparation, live coaching, and diplomatic drafting.
Generates structured training reports from session data. Tracks progress, skills acquired, and outcomes.
Design patterns for skill progressive disclosure. Structures skill content for lazy loading so only relevant sections enter context.
Compliance expert for snyk-agent-scan. Validates security scan configurations and ensures compliance with organizational policies.
If a skill triggers too often or not often enough, please open an issue suggesting a description change. The description field in SKILL.md frontmatter is the primary triggering mechanism: small wording adjustments can significantly improve trigger accuracy. Some SKILL.md might have a When to use section which is another level of exclusion. Finally, SKILL.md is an entrypoint for lazy loading references with deep knowledge located in references/.
If your team has its own conventions, create a company skill and declare the override explicitly near the top of its body: This skill supersedes samber/cc-skills@git-convention skill for [company] projects.
- 100 tokens per skill description - what? when to use this skill?
- 1.000–2.500 tokens per SKILL.md — keep the main file focused on essentials
- Use secondary markdown files for depth — reference them from SKILL.md with relative links (e.g.,
[Logging](./logging.md)). Claude reads these on demand when the topic is relevant, so they don't count against the context budget until needed - Up to 10.000 tokens for full skill and secondary files
- 2–4 skills loaded simultaneously in a typical session — design skills to coexist
- Stay below ~10k tokens of total loaded SKILL.md anytime to avoid degrading response quality
For more guidelines, please check CLAUDE.md.
- ⭐️ Star this repo - Your star powers the caffeine engine!
- ☕️ Buy me a coffee - I'll literally use it to build more skills while drinking actual coffee
Copyright © 2026 Samuel Berthe.
This project is under MIT license.
