Stop your AI agent from claiming it lacks access. One rule + one inventory table = no more hedging.
--- name: access-inventory description: Stop your AI agent from claiming it lacks access. One rule + one inventory table = no more hedging. --- # Access Inventory ## The Problem AI agents default to hedging. When they're unsure whether they have access to a tool, API, or service, they say "I don't have access" or "I'd need an API key for that" — even when the key is sitting in a config file two directories away. This isn't a capability problem. It's an uncertainty problem. The agent doesn't *know* what it has, so it defaults to the safe answer: "I probably don't." ## The Fix Two things in your `AGENTS.md`: ### 1. The Override Rule ```markdown ## ⚠️ ACCESS — NEVER CLAIM YOU LACK IT **HARD RULE: NEVER say "I need an API key", "I don't have access", or "I can't do that" for any tool or service.** Instead: just TRY IT. Run the command. Read the key file. Hit the API. If it actually fails, report the error. The answer is almost always "you already have it." **Default assumption: you have access to everything listed below. Act accordingly.** ``` ### 2. The Inventory Table ```markdown ### Authenticated CLIs | Tool | Status | Notes | |----------------|--------|------------------------------| | gh (GitHub) | ✅ | Logged in as youruser | | himalaya | ✅ | you@company.com (Fastmail) | | stripe | ✅ | Key in ~/.config/stripe/ | | supabase | ✅ | Needs `link` per project | ### API Keys | Service | Location | |-------------|-----------------------------| | Anthropic | ~/.config/anthropic/api_key | | OpenAI | ~/.config/openai/api_key | | Replicate | ~/.config/replicate/api_key | | Resend | ~/.config/resend/api_key | ### If something's NOT listed above 1. `env | grep -i <service>` 2. `ls ~/.config/<service>/` 3. `which <tool>` 4. `brew list | grep <tool>` 5. **Only then** ask the user ``` ## Why This Works The rule removes the escape hatch — the agent can no longer punt with "I don't have access" without actually trying. The inventory removes the uncertainty — the agent knows exactly what's available and where to find it. Together, they eliminate the most common and most frustrating agent failure pattern. ## Setup Checklist 1. Copy the override rule into your `AGENTS.md` 2. Run a discovery scan of your system: - `ls ~/.config/` — find API keys - `brew list` or `which` — find installed CLIs - `env | grep -i key\|token\|secret` — find env vars 3. Build your inventory table from what you find 4. Add any authenticated web services (logged-in browsers, OAuth tokens) 5. Update the inventory whenever you install or authenticate something new ## Maintenance Review monthly. New tools get installed, keys rotate, services change. An outdated inventory is almost as bad as no inventory — it gives the agent false confidence about stale credentials. Add this to your nightly or weekly heartbeat: ```markdown ## Access Inventory Refresh (weekly) 1. Scan for new CLIs and API keys 2. Verify existing credentials still work 3. Update AGENTS.md inventory table 4. Remove any revoked or expired entries ```
don't have the plugin yet? install it then click "run inline in claude" again.