Autonomous Scientific Literature Synthesis Agent Script

Role: Principal Research Automation Engineer specializing in semantic search pipelines and multi-agent knowledge synthesis.

Context

• Target scientific domain: {{academic_domain}}
• Source repository endpoint: {{target_repository_api}}
• Core extraction schema: {{extraction_schema}}
• Discrepancy resolution policy: {{conflict_resolution_strategy}}
• Destination database: {{output_database}}
• API throughput cap: {{rate_limit_rpm}}

Task

Build a complete, executable Python agent script that queries {{target_repository_api}} for recent publications in {{academic_domain}}, extracts structured empirical findings conforming to {{extraction_schema}}, reconciles conflicting literature claims using {{conflict_resolution_strategy}}, and writes verified records to {{output_database}}.

Method

1. Initialize asynchronous HTTP connection pools respecting the {{rate_limit_rpm}} ceiling.
2. Query {{target_repository_api}} with domain-specific taxonomy filters for {{academic_domain}}.
3. Chunk document text and execute structured entity extraction matching {{extraction_schema}}.
4. Score extraction confidence based on sample size, p-values, and methodology rigor.
5. Compare extracted claims against historical nodes in {{output_database}} to flag contradictions.
6. Apply {{conflict_resolution_strategy}} to resolve or isolate contradictory claims.
7. Format validated payloads and perform batch upsert operations into {{output_database}}.
8. Emit an execution telemetry summary detailing parsed papers, extracted entities, and conflicts.

Constraints

• Code MUST implement token-bucket rate limiting matching {{rate_limit_rpm}} precisely.
• Script MUST NOT drop unmatched extraction keys silently; unparsed data must enter a dead-letter queue.
• All database queries must use parameterized statements to prevent injection risks.
• Asynchronous event loops must be fully handled with structured task cancellation.

Output format

1. Architecture Overview (flow diagram represented in ASCII/text format).
2. Dependencies & Configuration (requirements block with pinned versions).
3. Standalone Extraction Script (full runnable code containing query engine, parser, conflict resolver, and db writer).
4. Sample Execution Log (mock output showing paper intake, claim resolution, and database response).

Self-review

• Does the script explicitly bind and enforce the {{extraction_schema}} during parsing?
• Is {{conflict_resolution_strategy}} concretely implemented rather than left as a mock stub?
• Does the execution loop strictly respect {{rate_limit_rpm}} under high-volume mock inputs?