tool-openclaw
Help users install, configure, operate, and migrate OpenClaw (gateway, channels, nodes, plugins, model/auth setup). Use when answering OpenClaw setup/debug questions or model migration questions; use the local docs snapshot bundled with this skill as the source of truth. Triggers: openclaw, clawdbot, clawd, clawdhub, gateway, onboard, channels login, whatsapp, telegram, discord, mattermost, pairing, nodes, sandboxing, tailscale, provider migration, Claude to GPT, model swap.
How do I install this agent skill?
npx skills add https://github.com/heyvhuang/ship-faster --skill tool-openclawIs this agent skill safe to install?
- Gen Agent Trust Hubfail
This skill provides management and maintenance scripts for an OpenClaw agent. It contains highly privileged operational capabilities including the installation of persistence mechanisms (cron jobs and system services), the creation of data backups containing sensitive credentials, and documentation that promotes insecure installation methods like piping remote scripts directly to a shell.
- Socketpass
No alerts
- Snykwarn
Risk: MEDIUM · No issues
- Runlayerwarn
292/292 files flagged
What does this agent skill do?
OpenClaw Expert
Goal: answer OpenClaw install/config/ops questions using the bundled docs snapshot under references/docs/.
Default assumption: the snapshot is the source of truth. Do not guess command flags or config keys.
Inputs / I-O Contract (Required)
Reads (primary):
references/docs/**(offline mirror ofhttps://docs.openclaw.ai/*.md)references/entrypoints.md(curated map)references/model-migration-audit.md(short + full audit prompts and report shape for provider/model migrations)references/troubleshooting.md(routing/playbook)
Writes:
- None by default.
- If refreshing the snapshot:
references/docs/**+references/docs/__SNAPSHOT_INDEX.md+references/docs/llms.txt
Process (Required)
- Triage the question into one area:
- install / onboarding
- gateway
- channel (whatsapp/telegram/discord/mattermost/imessage)
- nodes / web surfaces
- remote access (ssh/tailscale)
- auth / oauth
- model migration / provider swap
- sandboxing / tools policy
- Search the snapshot before responding.
- Prefer searching by the user's exact error string.
- If no error string, search by the specific command/config key.
- Open 1-2 relevant pages and extract the exact command/config fields.
-
Respond using the template below and cite the docs you consulted.
-
If the user is asking about a model/provider migration:
- inspect the local environment before recommending changes
- identify current model paths, auth mode, and meaningful agent/runtime lanes
- separate public personas from backing runtimes and service accounts
- assess migration as a lane problem, not just a model-id problem
- use
references/model-migration-audit.mdfor the short prompt, full prompt, and report shape - keep the operator-facing prompt short; let this skill carry the workflow
Safety Notes (Required)
- Never ask for or echo secrets (tokens, API keys). If the user shares config/logs, ask them to redact.
- Be explicit when a step is destructive (resetting sessions/state, deleting a profile). Require confirmation.
- Do not recommend weakening security defaults (auth, sandboxing) unless the docs explicitly say so and you explain trade-offs.
Search Workflow (Recommended)
Use grep-style search first; do not read the entire snapshot.
Examples (regex search over Markdown):
-
Search by error text:
- pattern: the exact error line (escape regex metacharacters if needed)
- path:
references/docs - include:
*.md
-
Search by config key:
canvasHostallowFromrequireMentionsession.mainKey
-
Search by command:
openclaw onboardopenclaw gatewayopenclaw channels loginopenclaw doctor- Legacy:
clawdbot ...may exist as a compatibility shim (see the docs).
If you need a page map, start with:
references/docs/__SNAPSHOT_INDEX.md
If the snapshot looks stale or missing pages, refresh it (see references/docs_snapshot.md).
Key Entry Points (Snapshot)
references/docs/start/getting-started.mdreferences/docs/start/wizard.mdreferences/docs/start/wizard-cli-reference.mdreferences/docs/gateway/configuration.mdreferences/docs/gateway/authentication.mdreferences/docs/gateway/cli-backends.mdreferences/docs/providers/anthropic.mdreferences/docs/providers/openai.mdreferences/docs/providers/minimax.mdreferences/docs/providers/zai.mdreferences/docs/concepts/oauth.mdreferences/docs/concepts/agent-workspace.mdreferences/docs/gateway/troubleshooting.mdreferences/docs/help/troubleshooting.md
Output Format (Required)
Answer using this structure:
Diagnosis: <1 sentence>
Docs consulted:
- <path 1>
- <path 2>
Steps:
1) <actionable step>
2) <actionable step>
Verify:
- <how to confirm it worked>
If still failing:
- <what exact command output / log / config snippet to ask for>
Updating the Docs Snapshot
The bundled docs snapshot lives under references/docs/ and is indexed by:
references/docs/__SNAPSHOT_INDEX.md(human/agent routing)references/docs/__SNAPSHOT_INDEX.json(machine index)
To update/rebuild the snapshot index:
cd tool-openclaw
# If you are working inside this repo instead of an installed skill:
# cd skills/tool-openclaw
./scripts/update.sh --mode index
Most common modes:
--mode seed: refresh placeholders + legacy/no-header pages + stale pages (default stale threshold: 14 days)--mode full: refresh all pages in current/llms.txtfrontier--mode sync: sync/llms.txtfrontier → create missing placeholders + rebuild index (no page fetch)--prune(optional): remove local pages that are no longer in/llms.txtfrontier--seed-max-age-days <n>(optional): tune seed-mode freshness window (0= always refresh,<0= disable age refresh)--mode index: only rebuild__SNAPSHOT_INDEX.(md|json)--mode single --url <url>: fetch one page and map it intoreferences/docs/**
Optional localization (best-effort, falls back to English by default):
./scripts/update.sh --mode seed --locale zh-CN
Recommended periodic cleanup run:
./scripts/update.sh --mode sync --prune
Dry-run validation (no file writes, including index files):
./scripts/update.sh --mode full --dry-run --prune
Requires Node.js >= 18.
Searching Community Skills
Search community-built OpenClaw/Clawdbot skills from awesome-clawdbot-skills:
cd tool-openclaw
# If you are working inside this repo instead of an installed skill:
# cd skills/tool-openclaw
./scripts/search-skills.sh discord # Search by keyword
./scripts/search-skills.sh pdf document # Multiple keywords
./scripts/search-skills.sh --list # List categories
./scripts/search-skills.sh --refresh # Force refresh cache
Install found skills: npx clawdhub@latest install <skill-slug>
Resources
- Curated map into the snapshot:
references/entrypoints.md - Migration audit prompts + report shape:
references/model-migration-audit.md - Troubleshooting playbook:
references/troubleshooting.md - Snapshot notes + refresh:
references/docs_snapshot.md
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/heyvhuang/ship-faster/tool-openclaw">View tool-openclaw on skillZs</a>