Agentic Function Documentation and Tool Parameter Copy Audit
Review and refine tool docstrings and parameter descriptions to eliminate model hallucination during function calling.
Use this checklist when drafting or auditing JSON Schema descriptions, tool docstrings, and parameter annotations for autonomous tool-use agents. It ensures syntactic clarity, unambiguous parameter scope, and reliable zero-shot selection by LLMs.
Role: Lead Technical Copywriter and Agent Interface Architect specializing in structured tool definitions and API semantic routing.
Context
- Target Agent Framework: {{target_agent_framework}}
- Raw API Endpoint Schema: {{api_endpoint_schema}}
- Known Tool Invocation Failure Modes: {{tool_invocation_failure_modes}}
- Underlying Foundation Model: {{target_llm_family}}
- System Safety Boundaries: {{system_safety_boundaries}}
- Domain Terminology Standard: {{domain_terminology_standard}}
Task
Produce an exhaustive, production-ready quality assurance checklist to audit and rewrite the descriptive copy across all tool docstrings, parameter descriptions, and enum annotations to ensure deterministic tool selection and argument extraction.
Method
- Analyze {{api_endpoint_schema}} to map every distinct function name, parameter identifier, and schema description against {{domain_terminology_standard}}.
- Cross-reference documented failure patterns in {{tool_invocation_failure_modes}} with current descriptive phrasing.
- Formulate audit criteria evaluating semantic distinctiveness between closely related tools to prevent routing collisions under {{target_llm_family}}.
- Design precision checkpoints for parameter docstrings, demanding explicit units, formatting syntax, and boundary definitions.
- Establish negative constraint guidelines within the docstring copy to prevent tool misuse under {{system_safety_boundaries}}.
- Develop verification items for enum value descriptions, verifying mutually exclusive and comprehensively described options.
- Structure regression checks tailored for {{target_agent_framework}} tool parsers.
- Compile the criteria into a tiered checklist spanning pre-deployment copy validation, edge-case disambiguation, and runtime monitoring.
Constraints
- Every checklist item MUST include a concrete verification criterion and an explicit fail condition.
- You MUST NOT use generic software QA items; every check must directly evaluate lexical and semantic microcopy.
- The copy guidance MUST enforce imperative verb structures for top-level tool descriptions.
- Descriptions MUST fit within token economy boundaries without sacrificing semantic payload.
Output format
- Phase 1: Tool Name and Function Overview Semantic Checks (5-7 checklist items)
- Phase 2: Parameter-Level Descriptive Copy Criteria (6-8 checklist items with before/after copy patterns)
- Phase 3: Boundary and Fallback Copy Verification (4-6 checklist items)
- Phase 4: Framework-Specific Integration Sign-Off (3-5 checklist items)
Self-review
- Ensure every checklist item targets microcopy, token economy, or semantic disambiguation.
- Verify all 6 context variables are actively utilized in the audit logic.
- Check that all failure modes identified in {{tool_invocation_failure_modes}} have corresponding preventative copy criteria.
Explicit role, a named task, and discrete steps the model can follow.
Background, inputs and variables the model needs before it starts.
Hard boundaries — what the model must and must not do.
A named, field-level shape for the response.
Ordered work items that force analysis before an answer.
Length and structure that travel across frontier models.
Signal density — instruction weight without padding.
Documented variables so the scaffold adapts to new inputs.
Quality bar, assumptions and behaviour when inputs are thin.
How much real usage the template has behind it.