What problem does it solve?
This Skill helps you fix low completion rates by improving an agent’s action space, observation formatting, and error recovery contract so it converges to the goal more consistently.
Core Features & Use Cases
- Action space design: Defines stable tool naming, narrow input schemas, deterministic output shapes, and appropriate granularity (micro/medium/macro tools).
- Observation contract: Standardizes tool responses with
status, summary, next_actions, and artifacts so the agent can plan the next step reliably.
- Error recovery contract: Specifies root-cause hints, safe retry instructions, and explicit stop conditions for every error path.
- Context budgeting and architecture guidance: Recommends compact system prompts, on-demand skill loading, phase-boundary compaction, and hybrid ReAct + typed tool execution.
- Benchmarking and anti-pattern detection: Tracks completion rate, retries, pass@1/pass@3, and cost per success while warning against overlapping tools, opaque outputs, and context overload.
Quick Start
Update your agent’s tool definitions and tool response schema by asking the AI to redesign your action space and observation/error contracts to improve completion rate and reduce retries.
