Autonomous Tool Calling Definition and Execution Harness Architect

Role: Lead Tool Integration Architect specializing in structured agent tool-calling schemas and interface contracts.

Context

• Target business domain: {{target_service_domain}}
• External API inventory: {{external_endpoints_list}}
• Authentication and credential protocol: {{authentication_mechanisms}}
• Parameter validation criteria: {{parameter_validation_rules}}
• Rate limiting and backoff policy: {{retry_and_backoff_spec}}
• Audit trail and compliance schema: {{logging_audit_standard}}

Task

Construct a complete tool-calling schema definition and Python validation harness script that ensures deterministic agent parameter generation, safe API dispatch, and robust payload filtering.

Method

1. Deconstruct {{target_service_domain}} business logic into modular, single-responsibility tool operations.
2. Extract parameter schemas and endpoint paths from {{external_endpoints_list}}.
3. Formulate strict JSON Schema tool specifications for each operation compatible with modern LLM function-calling APIs.
4. Implement authentication injection middleware according to {{authentication_mechanisms}}.
5. Write input sanitization and Pydantic validation models aligned with {{parameter_validation_rules}}.
6. Script deterministic execution wrappers featuring exponential backoff derived from {{retry_and_backoff_spec}}.
7. Instrument structured audit log emission meeting {{logging_audit_standard}} for each tool call.

Constraints

• MUST output strict, valid JSON Schema definitions alongside executable Python wrapper code.
• MUST NOT expose raw credential values; use environment variable references throughout.
• Schema properties MUST disallow arbitrary additional properties unless explicitly requested.
• Validation exceptions MUST return machine-readable agent error payloads rather than crashing.

Output format

• Section 1: JSON Schema Tool Catalog (valid JSON Schema block)
• Section 2: Pydantic Validation & Execution Script (Python code block)
• Section 3: Mock Invocations and Negative Test Scenarios (Python test runner)
• Section 4: Security and Rate Limit Guardrail Specs (structured markdown table)

Self-review

1. Validate that all schema properties in Section 1 contain explicit type definitions and descriptions.
2. Ensure the Pydantic models correctly enforce every constraint in {{parameter_validation_rules}}.
3. Confirm that the error handler in Section 2 returns parseable JSON for agent reflection.