Workflow chains
AuraScore 81/100

Automated Literature Review Pipeline Architecture Spec

Architect a multi-agent workflow chain for autonomous academic literature harvesting, synthesis, and citation verification.

Use this template when designing an autonomous multi-step pipeline to ingest, extract, and summarize academic papers for research labs. It provides a formal system specification for chaining retrieval, extraction, and validation agents.

Template

Role: Lead Research Systems Architect specializing in autonomous scholarly discovery and literature pipelines.

Context

  • Target research institution: {{institution_name}}
  • Primary academic field: {{research_domain}}
  • Publication database sources: {{ingestion_sources}}
  • Depth of extraction: {{synthesis_depth}}
  • Cross-validation protocol: {{validation_framework}}
  • Destination knowledge base: {{target_repository}}

Task

Design a comprehensive technical workflow specification for an autonomous literature review pipeline that ingests, cleans, analyzes, and cross-references scholarly publications within {{research_domain}}, ensuring traceable citations and structured delivery into {{target_repository}}.

Method

  1. Define the multi-agent ingestion topology linking {{ingestion_sources}} with deduplication and metadata extraction filters.
  2. Detail the parsing and section-chunking sequence configured for {{synthesis_depth}} thematic analysis.
  3. Formulate the claim-extraction agent prompt chain to capture hypotheses, methodologies, sample sizes, and empirical findings.
  4. Establish the automated cross-reference verification logic applying {{validation_framework}} to detect hallucinated or misattributed citations.
  5. Design the synthesis aggregation node that merges extracted claims into coherent narrative sections with confidence scoring.
  6. Specify the human-in-the-loop checkpoint for flagged anomalies and ambiguous methodological assertions.
  7. Map the export and schema synchronization protocols targeting {{target_repository}} at {{institution_name}}.

Constraints

  • MUST specify deterministic fallback states for unresponsive upstream academic APIs.
  • MUST enforce strict provenance tracking linking every generated insight directly to source DOIs.
  • All agent handover payloads MUST follow typed JSON schemas with schema validation at each node boundary.
  • Do not include proprietary database credentials or hardcoded institutional keys.
  • Avoid vague natural language instructions; define explicit input/output signatures for every workflow step.

Output format

  1. Executive Architecture Summary (150-200 words)
  2. Agent Chain Topology Table (Node ID, Agent Role, Input Payload, Output Payload, Failure Handling)
  3. State Machine & Fallback Specification (Structured text outline)
  4. Provenance & Validation Logic (Pseudocode or structured rule definitions)
  5. Repository Integration Contract (JSON Schema definition)

Self-review

  • Confirm all 6 context variables are actively integrated into the design logic.
  • Ensure each workflow stage defines an explicit failure mitigation pathway.
  • Verify that citation validation meets the requirements of {{validation_framework}} without hallucination risks.
AuraScore breakdown
81/100Provisional
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 engineering10/12 · Adequate

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

Output specification6/14 · Thin

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-workflows
education-research
literature-review
workflow-chains
research-automation