better-agent-skills

Skill suite for getting more out of your AI agent (OpenClaw, Hermes, Claude Code, Codex CLI, Cursor, Cline, and more).

A growing, security-conscious collection of portable agent skills — from zero-dependency environment diagnostics to cross-tool memory transfer and multi-agent coordination. All authored by Evan Song (@Songhonglei).

🧩 Skills in this suite

Categorization follows the Better Agent canonical taxonomy — Agent Memory & Boot / Agent Identity & Profile / Workspace & Session / Cross-container Coordination.

🧠 Agent Memory & Boot

Skill	Latest	What it does
`init‑hooks`	v1.0.0	Run scripts automatically on every container/host boot — survives gateway restarts, container restarts and rebuilds. Registry-driven, idempotent, with inline / local-script / remote-CDN hook types.
`subagent‑timeout‑config`	v1.0.0	One-click subagent timeout configurator for OpenClaw with preset profiles.
`claw‑memory‑manager`	v1.1.0	One-command management of OpenClaw built-in memory features. Configure Dreaming (Light→REM→Deep auto consolidation) with tunable half-life, max-age, IANA timezone.

🪪 Agent Identity & Profile

Skill	Latest	What it does
`copy‑my‑profile`	v1.0.0	Portable Markdown profile for AI tools — like vCard for AI assistants. Extract once from Tool A, paste into Tool B. Supports 10+ tools.
`agent‑avatar‑manager`	v1.0.1	Manage your OpenClaw Agent's avatar. Send an image/URL, or describe a style and let Freepik vector search pick.
`rename‑session`	v1.0.0	Rename or auto-generate a friendly session label. Random label generator (zh/en with locale auto-detect), multi-agent auto-detection.

🗃️ Workspace & Session

Skill	Latest	What it does
`workspace‑git‑setup`	v1.0.3	One-command Git tracking with a security-focused `.gitignore` (credentials / TLS / SSH keys auto-excluded), large-file guard, and `--audit` mode.
`session‑recovery`	v1.0.1	Recover lost agent session content and file changes from on-disk conversation logs. Streaming and OOM-safe on 700MB+ daily logs.
`token‑slim`	v1.0.0	Guided token optimization for agent workspaces — identifies bloat, recommends slimming, supports beast mode.
`hello‑env`	v1.0.2	Zero-dependency Bash environment health check for Linux, macOS, containers, K8s pods.
`collective‑memory`	v1.0.0	Broadcast a single memory note to multiple AI agent workspaces in one shot — upserts into MEMORY.md / AGENTS.md / TOOLS.md.

🌐 Cross-container Coordination

Skill	Latest	What it does
`agent‑team‑mesh`	v1.0.0	Team-wide P2P mesh for OpenClaw agents running on different containers/pods. Each gateway listens on its own pod for direct WebSocket calls.

📦 Install

Three ways to install, pick what fits your agent:

Option 1 — ClawHub (recommended for OpenClaw users)

clawhub install <skill-name>
# e.g.
clawhub install hello-env
clawhub install claw-memory-manager

Option 2 — skills.sh CLI (works for Claude Code, Cursor, Cline, Continue, OpenClaw, and 30+ other agents)

npx -y skills add Songhonglei/better-agent-skills -s <skill-name>
# e.g.
npx -y skills add Songhonglei/better-agent-skills -s copy-my-profile

Option 3 — Direct from GitHub

# Clone the whole suite
git clone https://github.com/Songhonglei/better-agent-skills.git
cp -r better-agent-skills/skills/<skill-name> /path/to/your/agent/skills/

# Or just one skill via curl + tar
curl -sL https://github.com/Songhonglei/better-agent-skills/archive/refs/heads/main.tar.gz | \
  tar -xzf - --strip-components=2 better-agent-skills-main/skills/<skill-name>

🎯 Use cases

Scenario	Try
New machine setup, want quick env diagnostic	`hello‑env`
Worried about losing workspace history	`workspace‑git‑setup`
Switching from Tool A to Tool B, don't want to retrain	`copy‑my‑profile`
Agent runs out of context too fast	`token‑slim`
Want agent to remember things better across days	`claw‑memory‑manager`
Lost conversation history after crash	`session‑recovery`
Multiple agents need to share knowledge	`collective‑memory`
Multiple agents need to call each other	`agent‑team‑mesh`
Subagent keeps timing out	`subagent‑timeout‑config`
Want to give your agent a personalized look	`agent‑avatar‑manager`
Sessions all have terrible default names	`rename‑session`
Init logic keeps getting lost on container rebuild	`init‑hooks`

🛡️ Quality standards

Every skill in this suite follows these standards:

Zero secret leakage — automated scan before publish
Cross-platform — Linux + macOS + (where applicable) Windows-via-WSL/Git-Bash
Locale-aware — language defaults from $LC_ALL / $LC_MESSAGES / $LANG, not hardcoded
Safe defaults — auto-backup before write, --dry-run for high-risk ops, validate before sync
Portable frontmatter — name + description only, ≤1024 bytes (works on strict YAML parsers like Qoder)
Real attribution — **Author** / **Repository** / **License** in every SKILL.md body
Single dependency budget — pure stdlib (Python) or pure Bash where possible

📚 Related ecosystems

OpenClaw — open-source agent framework (most skills target this primarily)
ClawHub — public skill marketplace
skills.sh — multi-agent skill installer (npx skills)
Anthropic Skills spec — the underlying SKILL.md format

🤝 Contributing

Want to fork a skill, file an issue, or suggest improvements? Open an issue or PR on this repo. For new skills, start by reading any existing skill's SKILL.md to see the conventions used.

👤 Author

Evan Song — github.com/Songhonglei

Built and maintained as part of daily AI-agent productivity work; skills are released here as they stabilize.

Better Agent Skills

README