Skip to content

adisinghstudent/ara.so

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

29 Commits
 
 
 
 
 
 

Repository files navigation

Openclaw-config

by ara.so — 1.5K+ installs on skills.sh

Agent skill for managing, debugging, and operating OpenClaw — the open-source AI agent runtime with 30+ LLM providers and 14 messaging channels.

Install it and your coding agent instantly knows how to diagnose issues, search sessions, edit config, and troubleshoot every channel.

Install

npx skills add adisinghstudent/ara.so

Works with Claude Code, Cursor, Codex, Windsurf, Cline, GitHub Copilot, and 30+ other agents.

Repository Structure

.
├── README.md
└── skills/
    └── openclaw-config/
        └── SKILL.md          # 850 lines of operational knowledge

What's in the Skill

Diagnostics

  • Quick health check — one-shot command that checks gateway, config JSON, channels, plugins, credentials, cron, recent errors, and memory DB
  • Known error patterns — 12 documented errors with exact meaning and fix

File Map

Complete reference for everything in ~/.openclaw/:

~/.openclaw/
├── openclaw.json                    # Main config — channels, auth, gateway, plugins
├── agents/main/
│   ├── agent/auth-profiles.json     # LLM auth tokens
│   └── sessions/
│       ├── sessions.json            # Session index
│       └── *.jsonl                  # Session transcripts
├── workspace/                       # Agent workspace
│   ├── SOUL.md                      # Personality and tone
│   ├── IDENTITY.md                  # Name and branding
│   ├── USER.md                      # Owner context
│   ├── AGENTS.md                    # Operating rules
│   ├── BOOT.md                      # Startup instructions
│   ├── HEARTBEAT.md                 # Periodic task checklist
│   ├── MEMORY.md                    # Long-term curated memory
│   ├── TOOLS.md                     # Contacts, SSH hosts
│   ├── memory/                      # Daily logs
│   └── skills/                      # Workspace-level skills
├── memory/main.sqlite               # Vector memory DB (Gemini embeddings)
├── logs/
│   ├── gateway.log                  # Runtime events
│   └── gateway.err.log              # Errors
├── cron/
│   ├── jobs.json                    # Job definitions
│   └── runs/                        # Per-job run logs
├── credentials/
│   ├── whatsapp/default/            # Baileys session (~1400 files)
│   ├── telegram/{bot}/token.txt     # Bot tokens
│   └── bird/cookies.json            # X/Twitter auth
├── extensions/{name}/               # Custom plugins (TypeScript)
├── browser/openclaw/user-data/      # Chromium profile
└── tools/signal-cli/                # Signal CLI binary

Channel Troubleshooting

Deep troubleshooting playbooks for each channel:

Channel Covered Issues
WhatsApp No reply to messages, 408 timeouts, cross-context blocks, session lookup, allowlist/group policy, lane congestion, full disconnection, credential wipe
Telegram Config validation errors (botToken vs token), polling timeouts, offset stuck, bot "forgetting" (compaction), correct config template
Signal RPC failures, signal-cli process health, rate limiting, wrong target format, profile name spam, daemon restart
Cron Job status overview, failure details, run log parsing, common failure causes, next scheduled times, disabling broken jobs

Memory System

Three-layer memory architecture:

  1. Context window — in-session, pruned by compaction
  2. Workspace filesMEMORY.md + daily logs in memory/
  3. Vector DB — SQLite + Gemini embeddings with FTS5 search

Includes commands for inspecting each layer, checking embedding rate limits, and rebuilding the index.

Session Search

  • Find conversations by person, channel, or date
  • Search message content across all sessions
  • Read specific session transcripts with formatted output
  • Understand the JSONL format (session, message, compaction, model_change events)

Config Editing

Safe edit patterns using jq:

  • Switch channel policies (open, allowlist, pairing, disabled)
  • Enable autopilot mode
  • Change LLM model
  • Set concurrency limits
  • Enable/disable plugins
  • Backup and restore

Security Modes

Mode Behavior Risk
open + allowFrom: ["*"] Anyone can message, bot responds to all HIGH
allowlist + allowFrom: ["+1..."] Only listed numbers LOW
pairing Unknown senders get approval code LOW
disabled Channel off NONE

Extending OpenClaw

  • Skills — markdown knowledge packs via ClawdHub or npx skills add
  • Extensions — custom TypeScript channel plugins
  • Cron jobs — scheduled autonomous tasks
  • Multi-agent — spawn Codex/Claude Code/Pi as background workers
  • Cross-channel — receive on WhatsApp, notify on Signal
  • Canvas — push HTML/dashboards to connected devices
  • Voice calls — Twilio/Telnyx/Plivo integration

Example Usage

Ask your agent any of these:

Why is my WhatsApp channel not connecting?
Show me the last 10 sessions on Telegram
Search all sessions for "deployment"
Switch Signal to allowlist mode for just my number
Which cron jobs are failing and why?
How do I add a new Telegram bot?

The skill gives the agent the exact commands, files, and fixes — no guessing.

Links

License

MIT

About

OpenClaw operations skill for AI agents — 1.5K installs on skills.sh

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors