
Mcp For Agents
If you're building an MCP server and watching agents misuse it, this skill is a solid checklist to run against your tool surface. It pushes you toward outcome-shaped tools instead of 1:1 REST mappings, flat parameters with enums and defaults, and tool descriptions that say when to reach for the tool. The part I'd lean on hardest is the error guidance: return failures as result content with isError set to true so the agent can self-correct, and write the message as an instruction. The token advice is worth taking seriously too, since schemas get injected on every request. The source excerpt cuts off before the composable-outputs examples finish, so I can't speak to those details.
- GitHub stars
- 222
- last updated
- Jul 24, 2026
Install
$ npx -y skills add ReinaMacCredy/maestro --skill mcp-for-agents --agent claude-codeInstalls into .claude/skills of the current project.
Files
MCP for agents
Developer-oriented MCP servers often fail agents: 1:1 REST-to-tool mappings that force multi-step orchestration, vague descriptions that cause wrong tool selection, nested parameter objects that invite hallucination, and raw API passthrough that exhausts the context window. Design for the agent's constraints, not the developer's convenience.
Outcomes over operations
The agent decides when to call; the server decides how. Combine backend operations server-side so the agent makes one call, not three.
Bad: Expose get_user_by_email, list_orders, get_order_status separately -- agent chains three calls.
Good: Expose track_latest_order(email) -- server handles the lookup internally, returns what the agent needs.
A tool that maps 1:1 to a REST endpoint is almost always wrong. Ask: "What outcome does the agent want?" and build the tool around that.
Flat, constrained parameters
Agents hallucinate missing keys in nested objects. Flatten parameters to top-level primitives, constrain with enums, and add sensible defaults so the agent makes fewer decisions.
Sponsors
Details
- Repository
- github.com/ReinaMacCredy/maestro
- Categories
- Backend & APIs, AI & Agent Building, Design & UI/UX
- License
- MIT
- First seen
- Aug 4, 2026