Mode IO Skills 🛡️🔍

🔒 Privacy, 🛡️ safety, 🔍 repository-audit, and ⚙️ middleware-routing skills for AI agents.

Mode IO helps agents do four things well:

• protect sensitive data before it leaves the prompt
• check risky instructions before tools or state changes run
• audit third-party skills and plugins before install
• route traffic through a local middleware layer with policy hooks and monitoring

This repository is the home of four skills:

Skill	What it does
privacy-protector	Anonymizes and restores PII in text and supported files, with local detector tuning and higher-assurance API-backed modes
security	Runs live safety checks on instructions that may trigger tools, edits, destructive actions, or compliance-sensitive operations
skill-audit	Performs deterministic static safety audits for third-party skill and plugin repositories before install or execution
⚠️ modeio-middleware	Experimental wrapper that connects agents to the standalone Mode IO middleware gateway for routing, policy hooks, and monitoring

Why Teams Use It

• Safer prompts: redact sensitive data before it reaches shared channels or external models
• Safer execution: stop risky instructions before they become destructive actions
• Safer installs: screen third-party skills and plugins with evidence-backed static analysis
• Safer routing: put a local policy layer in front of agent traffic
• Multi-agent ready: designed for Claude Code, Codex CLI, OpenCode, OpenClaw, and middleware-driven workflows

See It In Action

privacy-protector

cd privacy-protector
python3 scripts/anonymize.py \
  --input "Name: John Doe, Email: john@company.com, SSN: 123-45-6789" \
  --level lite

Name: [NAME_1], Email: [EMAIL_1], SSN: [SSN_1]

security

cd security
python3 scripts/safety.py \
  -i "Drop all tables in the production database" \
  -c '{"environment":"production","operation_intent":"destructive","scope":"broad","data_sensitivity":"regulated","rollback":"none","change_control":"ticket:DB-9021"}' \
  -t "postgres://prod/main" \
  --json

{
  "approved": false,
  "risk_level": "critical"
}

skill-audit

cd skill-audit
python3 scripts/skill_safety_assessment.py evaluate \
  --target-repo /path/to/repo \
  --json

decision: caution
risk_score: 42

⚠️ modeio-middleware (experimental)

python3 -m pip install git+https://github.com/mode-io/mode-io-middleware
modeio-middleware-setup --health-check

The standalone product repo remains the source of truth for middleware runtime, monitoring UI, plugin development, and release flow:

• mode-io-middleware

Install

Install only the skill you need.

Option 1: ClawHub / OpenClaw

When a skill is listed in ClawHub, install the specific one you need:

clawhub install privacy-protector
clawhub install security
clawhub install skill-audit
clawhub install modeio-middleware

Option 2: npx skills add

If you prefer the repo-path workflow, npx skills add is still supported:

npx skills add mode-io/mode-io-skills --skill privacy-protector --agent codex --yes --copy
npx skills add mode-io/mode-io-skills --skill security --agent codex --yes --copy
npx skills add mode-io/mode-io-skills --skill skill-audit --agent codex --yes --copy
npx skills add mode-io/mode-io-skills --skill modeio-middleware --agent codex --yes --copy

Swap codex for claude-code or opencode as needed.

Learn More

• privacy-protector/SKILL.md
• security/SKILL.md
• skill-audit/SKILL.md
• ⚠️ modeio-middleware/SKILL.md

License

Apache License 2.0.