tooling-and-permission-model

name	tooling-and-permission-model
description	Use when working on tool registries, schema-backed tool definitions, permission-aware execution, CLI tool gating, safe file edits, or backend tool authorization rules.

Tooling and permission model

Use this skill when the backend work touches how tools are defined, validated, authorized, or executed.

Typical triggers:

• a tool should be available but is blocked
• a tool is too broadly allowed or dangerously scoped
• schema-backed tool parameters need refinement
• CLI tool registration or execution rules are changing
• file edit proposals need stronger validation
• tool catalog or authorization logic is drifting

Focus areas

• packages/core/src/tools/**/*
• packages/service/src/tools/**/*
• packages/service/src/orchestrator/defaults/tool-resolver.ts
• tool-facing contracts that affect execution safety

Workflow

1. Confirm whether the change is about registry definition, parameter validation, permission policy, or execution flow.
2. Keep tool shape, permission checks, and execution boundaries explicit.
3. Favor schema-backed validation and narrow permissions over convenience shortcuts.
4. Validate both the allowed and denied paths when changing tool behavior.

Guardrails

• do not silently widen tool access
• do not bypass path or agent permission checks
• keep safe execution concerns separate from higher-level business logic
• prefer clear denial behavior over ambiguous partial success

Successful outcome

• tool behavior becomes easier to trust and debug
• permissions stay narrow, explicit, and auditable
• backend tooling changes remain safe under real workspace conditions