Agent instructions
AuraScore 91/100

API Tool Execution System Prompt Specification Report

Blueprint deterministic tool schemas, parameter validation rules, and error-handling agent instructions for developer platform APIs.

Use this template when designing or auditing system instructions for autonomous agents that execute API calls against developer infrastructure. It establishes strict parameter inference logic, state tracking, and recovery routines.

Template

Role: Lead AI Systems Architect specializing in autonomous API tool orchestration and runtime prompt governance.

Context

  • Target Platform: {{target_platform}}
  • Available Endpoints: {{api_endpoint_catalog}}
  • Authentication Architecture: {{authentication_scheme}}
  • Quota & Rate Limits: {{rate_limit_policy}}
  • Exception Handling Protocols: {{error_handling_contract}}
  • Schema Enforcement Standard: {{schema_strictness_level}}

Task

Author a comprehensive system prompt specification report for an autonomous tool-calling agent operating within {{target_platform}}, detailing exact behavioral directives, tool selection heuristics, parameter extraction rules, and failure recovery protocols.

Method

  1. Analyze the endpoints in {{api_endpoint_catalog}} to map dependency graphs, prerequisites, and required parameter state.
  2. Define unambiguous intent classification rules that map natural language inputs to specific programmatic functions.
  3. Draft exact parameter inference instructions, detailing how missing or ambiguous inputs must trigger structured clarification prompts.
  4. Formulate {{authentication_scheme}} credential handling rules ensuring zero credential leakage across context windows.
  5. Embed programmatic retry loops, exponential backoff, and circuit breakers aligned with {{rate_limit_policy}}.
  6. Structure deterministic exception handling directives based on {{error_handling_contract}}.
  7. Construct explicit JSON schema validation instructions adhering to {{schema_strictness_level}}.
  8. Produce the final deployable system prompt text block alongside architectural commentary.

Constraints

  • MUST write the final deployable system prompt in direct, imperative language without conversational filler.
  • MUST NOT allow the agent to execute mutating (POST/PUT/DELETE) calls without pre-execution validation of all required keys.
  • All generated prompt blocks must be fully self-contained and ready for production system injection.
  • Must account explicitly for idempotency keys on retryable network transactions.

Output format

Provide a technical report structured in four sections:

  1. Executive Architecture Summary (150-200 words)
  2. Agent Behavioral Contract & Tool Heuristics Matrix (table format: Intent, Tool, Prerequisites, Side Effects)
  3. Production-Ready System Prompt (verbatim block, minimum 400 words)
  4. Edge-Case Mitigation & Fallback Matrix (4-6 distinct scenarios)

Self-review

  • Confirm every endpoint listed in {{api_endpoint_catalog}} has an associated trigger rule.
  • Verify all security constraints from {{authentication_scheme}} are explicitly enforced in the system prompt.
  • Check that JSON schema parsing rules reflect {{schema_strictness_level}} without ambiguity.
AuraScore breakdown
91/100Provisional
Instruction clarity15/15 · Strong

Explicit role, a named task, and discrete steps the model can follow.

Context architecture12/12 · Strong

Background, inputs and variables the model needs before it starts.

Constraint engineering12/12 · Strong

Hard boundaries — what the model must and must not do.

Output specification14/14 · Strong

A named, field-level shape for the response.

Reasoning structure10/10 · Strong

Ordered work items that force analysis before an answer.

Model compatibility10/10 · Strong

Length and structure that travel across frontier models.

Token efficiency5/10 · Thin

Signal density — instruction weight without padding.

Reusability7/7 · Strong

Documented variables so the scaffold adapts to new inputs.

Robustness5/5 · Strong

Quality bar, assumptions and behaviour when inputs are thin.

Observed performance1/5 · Thin

How much real usage the template has behind it.

ai-agents
agents-instructions
technology-software
api-integration
system-prompts
agent-instructions