better-agent-skills

Manage OpenClaw Agent's built-in memory features — enable/disable/configure Dreaming (Light→REM→Deep auto memory consolidation) and Active Memory (proactive memory injection with three style presets: conservative / balanced / aggressive). Auto-detects K8s ConfigMap mirrors, creates *.bak.<timestamp> backups before write, supports --dry-run preview, --style preset for active-memory, --half-life / --max-age / --timezone for dreaming, and triggers gateway restart automatically (opt-out via --no-restart). Triggers: "enable dreaming", "tune memory half-life", "configure agent memory", "enable active memory", "switch memory style", "conservative/balanced/aggressive memory mode".

管理 OpenClaw 容器启动时自动执行的脚本钩子。gateway 重启、pod 重启、pod 重建均不丢失（数据持久化在 workspace/.init-hooks/）。 支持 inline 命令、本地脚本（shell/python）、CDN URL（zip/tar.gz 自动下载解压）三种钩子类型。 用于：新增/查看/修改/禁用/删除启动钩子、立即测试运行、查看执行日志。 触发词：init-hooks、启动钩子、startup hook、重启初始化、pod重启执行、开机自启。 安装触发词：帮我安装 init-hooks、初始化 init-hooks、设置 init-hooks、执行 init-hooks 安装。 配置变更触发词：修改 openclaw.json、改 gateway 配置、改配置文件 → 执行后自动调用 sync_openclaw_config.py。

Team-wide P2P mesh for OpenClaw agents running on different containers/pods. Each agent's gateway listens on its own pod IP:18789 over WebSocket; the mesh CLI lets you ping, send-and-await-reply, broadcast, and discover the whole team. No broker, no Supabase, no central server — just direct WS calls between teammates' agents. Includes auto-detect of "this machine's identity" (USER.md / sso.json / env var), secure token storage (separate chmod 600 file, not committed to git), message size limits (4KB warn / 8KB block), --dry-run preview for both send and broadcast, and an optional IM fallback hook when an agent is unreachable. Triggers: "message my teammate's agent", "ping bob's agent", "broadcast to the team", "agent mesh", "team agent communication".

One-click subagent timeout configurator for OpenClaw. Configures the three related fields (agents.defaults.timeoutSeconds + subagents.runTimeoutSeconds + acp.runtime.ttlMinutes) atomically with built-in TTL constraint validation (ttlMinutes * 60 ≥ runTimeoutSeconds). Three preset profiles — quick (60s/1h/1h, fast feedback), normal (3min/2h/2h, everyday dev), patient (5min/3h/3h, long-running tasks) — plus --custom for arbitrary values. Auto-backup before write, optional gateway auto-restart, --dry-run preview. Triggers: "set subagent timeout", "subagent 超时", "配置 timeout", "调整超时", "subagent timeout config", "openclaw timeout", "tool call timeout".

Extract a standardized cross-tool user profile from any AI agent's long-term memory (USER.md, MEMORY.md, AGENTS.md, CLAUDE.md, .cursorrules, etc.) so you can re-use it in another AI tool without retraining. Output is a portable Markdown document with five categories — instructions, identity, profession, projects, preferences. Works with Claude Code, OpenClaw, Codex CLI, Cursor, Cline, Continue, Aider, or any LLM agent. 适用场景:生成我的画像/导出 profile/ 拷贝到其他 AI 工具/跨工具同步偏好/换 agent 不丢记忆。

Manage your OpenClaw Agent's avatar. Send an image/URL directly, or describe a style and let Freepik vector search pick candidates for you. Auto-saves to the workspace avatars/ directory and updates IDENTITY.md. 当用户说「换头像」「找一张头像」「更新头像」「给我换个头像」「change avatar」时使用。 Requires a free Freepik API Key for auto-search mode (direct image/URL mode works without it). OpenClaw-only — relies on `openclaw agents set-identity`, `IDENTITY.md`, and `TOOLS.md` conventions.

Broadcast a single memory note to multiple AI-agent workspaces in one shot, upserting into MEMORY.md / AGENTS.md / TOOLS.md / USER.md across all targets. Pure file-ops, zero network, zero LLM. Works on OpenClaw, Claude Code, Cursor, and any agent runtime that stores per-project memory as Markdown. Triggers: "broadcast memory", "tell every agent", "remember this everywhere", "sync this to all agents", "make all my agents know", "集体记忆", "广播这条记忆", "你们全都记住", "所有agent记住", "大家都记一下", "让他们都记住".

Rename or auto-generate a friendly label for an OpenClaw-style session by editing sessions.json directly. Supports random labels (zh/en with locale auto-detect), multi-agent auto-detection, listing sessions, retry with verification, and XDG-compliant history storage. 适用场景:在 OpenClaw / 任何兼容的 session 管理体系中给会话起名/改名/换花名/重命名 session/给会话加 label。

Recover lost agent session content and file changes from on-disk conversation logs. Streaming and OOM-safe on 700MB+ daily JSONL. Two commands: search.py for keyword search across recent sessions (with hit snippets, file-op listing, JSON for agent consumption); extract.py for pulling full write/edit content from a single session by ID prefix, with optional replay-rebuild for pure-edit sequences and safe restore-to-disk (refuses silent overwrites without --yes). Multi-agent aware via --agent main|all|a,b. Configurable data root via --root flag or SESSION_RECOVERY_ROOT env var (default ~/.openclaw/agents/). Trigger when user wants to find lost session content, recover files written by an agent, locate which session modified a file, search session history by keyword, or rebuild a file from an edit replay. Also triggers on 找回会话, 会话被覆盖, 历史会话搜索, 文件被删了, session 丢了, 找回某个文件, 重放编辑.

Guided token optimization for AI agent workspaces. Triggers on phrases like "save tokens", "optimize tokens", "context window too large", "memory files too big", "scan my workspace", "trim workspace", "find token waste", "再扫一下", "帮我省 Token", "优化 Token 使用", "上下文太长了", "内存文件太大", "还有哪里可以优化". Also handles brutal-mode toggle: "enable brutal mode", "disable brutal mode", "开启狂暴模式", "关闭狂暴模式". On first use walks the user step-by-step through workspace cleanup; supports on-demand re-scans at any time. Works on OpenClaw, Claude Code, and any agent runtime with a writable working directory.

Zero-dependency Bash environment health check for Linux, macOS, containers, and K8s pods. Reports OS, current user, Node.js, Python3, basic tools (git/curl/jq/make/docker) with paths and versions, network (hostname, IP, inferred subnet), container/K8s detection with a configurable env-var probe, workdir and git status, optional config-file check, and PVC remount detection (device-number plus watch-dir count snapshot diff to catch silent K8s PersistentVolume reattach to a fresh disk). Three-layer config — CLI flags override env vars override defaults. Trigger when user wants to check the environment, verify a dev box is healthy, detect container or K8s context, look up local IP or subnet, see Node and Python versions, list installed basic tools, or watch for silent PVC rotation in K8s pods. Also triggers on 检查我的环境, 环境自检, hello-env, 查看系统信息, 环境有没有问题, IP 是多少, 在什么环境, PVC 换卷了吗, 基础工具检查, node 版本, 当前用户是谁.

One-command Git tracking setup for any working directory, with a security-focused .gitignore (credentials, TLS/SSH private keys, tokens, runtime caches & PIDs, node_modules, Python caches, build artifacts, editor temp files), a large-file guard (>10MB), and a read-only --audit mode that detects tracked secrets, untracked files, and core.autocrlf misconfiguration. Three modes: (1) init — git init + safe .gitignore + first commit; (2) --audit — read-only health check (no mutations); (3) --dry-run — preview every change without applying. Reads Git identity from GIT_AUTHOR_NAME/GIT_AUTHOR_EMAIL env vars > existing git config > interactive prompt; falls back to the email prefix when only the email is provided. Workspace path resolution: CLI arg > WORKSPACE_DIR env var > current directory. Default branch is main; sets core.autocrlf=input to avoid CRLF false diffs. Trigger this skill when the user says any of: 'add git to my project', 'init git', 'git init', 'track my changes', 'version history', 'back up my workspa