agent-platform-tool-permission-matrixlisted

# Govern agent tool access with permission matrices > Role-scoped tool grants, parameter-level policies, and audit trails making agent autonomy insurable. **Track:** 🤖 AI Engineering · **Domain:** Agent Platform Engineering · **Level:** advanced · **~40 min** **Who this is for:** AI Engineers, ML Engineers, LLM App Developers, Agent Builders ## When to Use This Skill Role-scoped tool grants, parameter-level policies, and audit trails making agent autonomy insurable. Use it whenever a matching task appears in conversation — the agent loads these instructions on demand. ## Steps 1. Inventory every tool × action class; classify read/write/irreversible 2. Bind permissions to agent ROLES, not individual deployments ad hoc 3. Enforce parameter constraints server-side (amount caps, allowlisted targets) 4. Require dual approval workflows for irreversible classes in prod 5. Log grant changes separately from usage; both auditable immutably 6. Red-team the matrix quarterly: can role combos escalate? ## Common Pitfalls - Wildcard grants copied between environments 'temporarily' - Parameter validation living only in prompts ## Commands **Install with skills CLI** ```bash npx skills add aniruddhaadak80/skills --skill agent-platform-tool-permission-matrix ``` **Install globally** ```bash npx skills add aniruddhaadak80/skills --skill agent-platform-tool-permission-matrix -g ``` --- Part of [aniruddhaadak80/skills](https://github.com/aniruddhaadak80/skills) · Browse all at http