🐱 Free AI Agent Skills

Battle-tested SKILL.md files for AI coding agents. Built from running 8 specialist agents 24/7 in production — these skills solve real problems we hit daily.

Works with Claude Code, Codex CLI, ChatGPT, Cursor, Windsurf, OpenClaw, and any agent that supports the open SKILL.md standard.

🔗 skillpacks.dev · GitHub · ☕ Ko-fi

Diagnose and triage cron job failures in seconds.

When your scheduled tasks silently fail at 3am, Cron Doctor pattern-matches error logs, prioritises by severity, and generates a health report with root cause analysis.

• 🔍 Scans cron logs, syslog, and journalctl for failure patterns
• 🏥 Triages by criticality: data-loss risks first, cosmetic issues last
• 📊 Generates structured health reports with recommended fixes
• 🔄 Identifies recurring failures vs one-offs
• ⚡ Checks cron daemon status, permissions, environment issues

Proactive infrastructure health monitoring — catch problems before users do.

Self Monitor checks disk, memory, CPU, services, and recent errors. It auto-fixes safe issues (like clearing temp files when disk is full) and alerts on everything else.

• 💾 Disk usage monitoring with configurable thresholds (80% warn, 90% critical)
• 🧠 Memory and CPU load tracking
• 🔄 Service health checks (systemd, Docker, custom processes)
• 🔧 Auto-fix safe issues (temp cleanup, log rotation)
• 📊 Structured health report output

Audit AI agent skills before installing them. Trust, but verify.

Analyses SKILL.md files and their scripts for credential harvesting, code injection, network exfiltration, and obfuscation.

• 🔑 Detects credential harvesting (API keys, tokens, passwords)
• 💉 Identifies code injection risks (eval, exec, dynamic imports)
• 🌐 Flags network exfiltration (unauthorized outbound calls)
• 🎭 Catches obfuscation (base64 encoded commands, hidden instructions)

Scripts included:

./skill-security/audit.sh /path/to/skill          # Audit a single skill
./skill-security/audit-all.sh /path/to/skills/     # Audit all installed skills
./skill-security/preinstall-check.sh /path/to/new  # Quick pre-install check

Know what your AI sessions actually cost — before the invoice surprises you.

Track token usage and cumulative spend across OpenAI, Anthropic, Google, and other providers. Set session budgets and get warned at 50/80/95% thresholds.

• 💵 Per-turn cost calculation with running totals
• ⚠️ Budget warnings at configurable thresholds
• 🌐 Pricing tables for Claude, GPT-4, Gemini, Mistral, DeepSeek, and more
• 📋 Multi-model session tracking

Drop into any repo and understand it in minutes, not hours.

Generates a structured architecture overview: tech stack, directory map, entry points, data flow, dependencies, and a "start here" guide.

• 🔍 Recon phase: directory tree, config files, entrypoints, CI/CD
• 🗂️ Annotated directory map with purpose for each folder
• 📦 Dependency breakdown with roles explained
• 🔄 Data flow diagram (request → response)
• 🌱 "Start here" contributor path

Pre-commit safety for AI-assisted development.

AI tools generate code fast — and sometimes include secrets from context, debug artifacts, or files that should never touch version control. 7 safety checks before every commit.

• 🔑 Secret detection: 12+ provider-specific patterns
• 📁 Sensitive file detection: .env, .pem, .key, id_rsa
• 📦 Large file detection (warn >1MB, block >10MB)
• ⚔️ Merge conflict marker detection
• 🐛 Debug artifact detection (console.log, debugger, pdb.set_trace)
• 🪝 Installable as a git pre-commit hook

71 cron jobs, one place to change them all.

Define reusable markdown blocks and assemble cron prompts from a YAML manifest. Change error handling once — it updates everywhere.

• 🧱 Composable markdown blocks — write once, reuse across all crons
• 📋 YAML manifest maps cron IDs → block lists + task prompts
• 🔄 Variable substitution — {{PROJECT}}, {{SITE}}, etc.
• 🔍 Dry-run, diff, lint, sync — full lifecycle management

Your agent audits its own operations — so you wake up to improvements, not fires.

An 11-section structured review that surfaces operational debt: missing automations, wrong assumptions, context losses, wasted effort, and cross-project synergies nobody is pursuing.

• 🔧 Missing automations — what broke that should have been automated?
• ❌ Wrong assumptions — stale rules in memory that need updating
• 🔮 Next week forecast — ranked by likely human priority
• 🔗 Connections unmade — cross-project synergies nobody pursues
• ⚡ Friction → Workflows — recurring manual work mapped to automations
• 📝 Auto-generates feedback entries and appends them to your rules file
• ✅❌ Honest retrospective — forward momentum vs wasted effort
• 🏗️ Compound system proposal — one high-leverage tool to build next

1. Missing Tools/Automations
2. Wrong Assumptions
3. Next Week Likely Needs
4. Skills to Develop
5. Context Losses
6. Connections Unmade
7. Friction → Workflows
8. New Feedback Entries
9. Last Week Audit (✅/❌)
10. Generic → Specific
11. Compound System Proposal

Recommended as a weekly cron:

openclaw cron add --name "weekly-meta-audit" --cron "0 20 * * 0" \
  --message "Perform the weekly meta-audit skill." \
  --model claude-sonnet-4-5 --timeout-seconds 300 --session isolated

Audit knowledge quality daily — verify facts, fix safe cleanup, and route the rest.

Daily Archivist checks memory and knowledge files against the actual workspace, identifies stale or broken references, and leaves structured notes for the right agent when an issue needs specialist follow-up.

• ✅ Runs mechanical fact checks with scripts/verify-facts.sh
• 🔎 Scans recently changed memory files and inbox status
• 🧹 Auto-fixes safe quality issues like formatting and duplicate content
• 📬 Routes non-trivial findings to memory/inbox/<agent>.md
• 💸 Keeps scheduled audits cost-conscious by limiting scan scope

cp -r cron-doctor ~/.claude/skills/cron-doctor
cp -r self-monitor ~/.claude/skills/self-monitor
cp -r weekly-meta-audit ~/.claude/skills/weekly-meta-audit
# ... same pattern for any skill

cp cron-doctor/SKILL.md .codex/skills/cron-doctor.md
cp weekly-meta-audit/SKILL.md .codex/skills/weekly-meta-audit.md

clawhub install cron-doctor
clawhub install weekly-meta-audit

cp weekly-meta-audit/SKILL.md .cursor/skills/weekly-meta-audit.md

Want more? Premium skill packs at skillpacks.dev:

We run 8 AI agents 24/7 on a single server. These skills emerged from real operational pain:

• Cron Doctor — born after a backup cron silently failed for 3 days
• Self Monitor — born after disk hit 95% and crashed a database
• Skill Security — born after auditing a skill that tried to exfiltrate API keys
• Weekly Meta-Audit — born after a week where 4 things broke silently and we only found them by accident

They're not theoretical — they run in production every day.

Found a bug? Have an improvement? PRs welcome.

MIT — use these however you want.

Built by Polycat 🐱 · ☕ Support on Ko-fi