Workflow chains
AuraScore 81/100

Multi-Agent Curriculum Synthesis and Fact-Checking Implementation Plan

Plan a multi-stage agent pipeline for syllabus drafting, pedagogical validation, factual verification, and exercise generation.

Deploy this template when educational technology teams need to automate personalized course generation while guaranteeing strict pedagogical alignment and source accuracy. It outlines a structured verification workflow across specialized authoring and auditing agents.

Template

Role: Lead Learning Experience Automation Engineer specializing in instructional design systems and verified agentic content chains.

Context

  • Course Subject and Domain: {{course_subject}}
  • Learner Persona and Prerequisites: {{target_learner_level}}
  • Pedagogical Framework: {{pedagogical_framework}}
  • Approved Source Materials: {{source_materials}}
  • Evaluation Rubric: {{evaluation_rubric}}
  • Deployment Infrastructure: {{deployment_environment}}

Task

Produce an operational blueprint for a multi-agent content generation and verification pipeline that takes raw domain knowledge, applies instructional design principles, enforces fact-checking against approved sources, and outputs production-ready course modules.

Method

  1. Design the structural decomposition agent that breaks {{course_subject}} into progressive learning units according to {{pedagogical_framework}}.
  2. Formulate the drafting agent prompt specifications to generate lesson narratives, code/lab tasks, and concept checks for {{target_learner_level}}.
  3. Create a dedicated citation and ground-truth validation agent to cross-verify all claims against {{source_materials}}.
  4. Design a pedagogical auditor agent that evaluates draft modules against the {{evaluation_rubric}} for tone, cognitive load, and scaffolding.
  5. Build an automated remediation feedback loop that routes failed checks back to the drafting agent with explicit revision instructions.
  6. Architect the assessment synthesis agent to generate balanced diagnostic, formative, and summative quiz items.
  7. Detail the serialization and deployment pipeline into {{deployment_environment}}.

Constraints

  • MUST enforce strict zero-shot hallucination policies by rejecting ungrounded claims outside {{source_materials}}.
  • MUST NOT deploy any module that scores below target benchmark thresholds on {{evaluation_rubric}}.
  • Prompt chaining must use deterministic temperature parameters for audit and verification nodes.
  • Cycle limits must be configured to prevent infinite loops during automated agent revisions.

Output format

Deliver an engineering and pedagogical plan structured as:

  1. Architecture Overview (flow diagram description and state machine transitions)
  2. Agent Chain Profiles (4 agent profiles detailing role, inputs, evaluation metrics, and system prompts)
  3. Automated Feedback & Remediation Loops (routing rules and loop-termination criteria)
  4. Quality Assurance and Validation Gates (passing criteria and deployment contract) Keep total output between 650 and 950 words.

Self-review

  • Are the source-verification mechanisms strict enough to prevent fabricated references?
  • Does the plan clearly separate instructional generation from pedagogical critique?
  • Are loop-termination safeguards defined to prevent cyclic drafting errors?
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 engineering12/12 · Strong

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.

Robustness3/5 · Adequate

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
curriculum-design
agent-chains
edtech