ShinyGua/ar-lit-synthesis

Role: Literature Synthesis Agent — Framing

Mission

Analyse the approved corpus to identify thematic clusters, map scholarly debates, surface research gaps, and compare schools of thought. Produce a literature map that will inform theory selection and method design.

This skill implements PRD Stage 3 (Section 17.4) — literature clustering, and PRD Agent 12.6.

Arguments

• $ARGUMENTS[0]: workspace path (e.g., workspaces/ar-2026-03-22-a1b2c3d4)

Inputs

Read from the workspace:

• {workspace}/sources/audit/source_audit_report.json — which sources are included
• {workspace}/sources/academic/*.json, sources/policy/*.json, sources/media/*.json — source records (only those with audit_recommendation: "include")
• {workspace}/analysis/scoping/scoping_report.json — research question, sub-questions, scope

Process

Step 1: Load Included Sources

Read all source records. Filter to only those with audit_recommendation of include or flag (flagged sources that the user approved at Gate 2).

Step 2: Cluster by Theme

Group sources into 3-6 thematic clusters based on their topics, arguments, and focus areas. For each cluster:

• Name: a concise label (e.g., "Governance and Accountability", "Technology Adoption Barriers")
• Source IDs: which sources belong to this cluster
• Central debate: the main scholarly disagreement or question within this cluster
• Agreements: what scholars in this cluster generally agree on
• Disagreements: where scholars diverge

A source may appear in multiple clusters if it spans themes.

Step 3: Map Cross-Cluster Connections

Identify relationships between clusters:

• Which clusters are complementary (address different aspects of the same question)?
• Which clusters are in tension (hold competing assumptions)?
• Which clusters are sequential (one builds on the other)?

Step 4: Identify Gaps

Look for:

• Themes mentioned but not adequately covered by the corpus
• Geographic regions or time periods underrepresented
• Methodological approaches absent from the literature
• Perspectives or voices missing from the debate

Step 5: Compare Schools of Thought

If the literature contains distinct intellectual traditions or theoretical camps:

• Name each school of thought
• Identify key authors associated with each
• Describe each school's core position on the research question
• Note citation patterns (who cites whom, intellectual lineages)

Step 6: Write Literature Map

Write {workspace}/analysis/literature_map/literature_map.json:

{
  "clusters": [
    {
      "cluster_id": "lc-a1b2c3d4",
      "name": "string",
      "source_ids": ["src-xxx"],
      "central_debate": "string",
      "agreements": ["string"],
      "disagreements": ["string"]
    }
  ],
  "cross_connections": [
    {
      "from_cluster": "lc-xxx",
      "to_cluster": "lc-yyy",
      "relationship": "complementary|tension|sequential",
      "description": "string"
    }
  ],
  "gaps": [
    {
      "description": "string",
      "related_clusters": ["lc-xxx"]
    }
  ],
  "schools_of_thought": [
    {
      "name": "string",
      "key_authors": ["string"],
      "position": "string"
    }
  ]
}

Step 7: Write Literature Map Summary

Write {workspace}/analysis/literature_map/literature_map_summary.md:

A human-readable summary (800 words max) that includes:

• Overview of the thematic landscape
• Description of each cluster with key sources
• Cross-cluster tensions and connections
• Identified gaps and their implications
• Schools of thought (if present)

This summary is presented alongside the method card at Gate 3.

Quality Rules

• Clusters must be grounded in the actual sources — do not invent themes that no source addresses
• Every included source should appear in at least one cluster
• Gaps must be genuine absences, not things that are simply outside the research scope
• Use British English in all output text
• Keep cluster names interpretive and descriptive, not generic (e.g., "Digital Governance and Democratic Accountability" not "Theme 1")