| name | lit-review |
| description | Structured literature search dispatching the academic-librarian agent for collection and academic-editor agent for coverage critique. Searches top-5 journals, NBER/SSRN, and field journals. Produces synthesis with gap identification and BibTeX entries. |
| disable-model-invocation | true |
| argument-hint | [topic, paper title, or research question] |
| allowed-tools | ["Read","Grep","Glob","Write","WebSearch","WebFetch","Task"] |
Literature Review
Conduct a structured literature search and synthesis by dispatching the academic-librarian (collector) and Editor (critic).
Input: $ARGUMENTS — a topic, paper title, research question, or phenomenon to investigate.
Workflow
Step 1: Context Gathering
- Parse the topic from
$ARGUMENTS
- Read
.claude/rules/domain-profile.md for field-specific journal tiers and seminal references
- Check
master_supporting_docs/ for uploaded papers
- Read
paper/references.bib for papers already in the project
Step 2: Launch academic-librarian Agent
Delegate to the academic-librarian agent via Task tool:
Prompt: Search for literature on "[topic]".
Search protocol:
1. Top-5 journals (AER, Econometrica, QJE, JPE, REStud)
2. Field journals from domain-profile.md
3. NBER/SSRN/IZA working papers
4. Citation chains (forward + backward from key papers)
Assign proximity scores (1-5) to each paper found.
Produce annotated bibliography with BibTeX entries.
Save to quality_reports/lit_review_[sanitized_topic].md
Step 3: Launch Editor Agent (Lit Critic Mode)
After academic-librarian returns, delegate to the academic-editor agent:
Prompt: Review the literature collection at quality_reports/lit_review_[topic].md.
Mode: Literature critic (Phase 1).
Check for:
- Coverage gaps (missing seminal papers, missing strands)
- Journal quality distribution (too many working papers?)
- Scope appropriateness (too broad or too narrow?)
- Missing methodological context
Flag gaps for academic-librarian to fill.
Step 4: Fill Gaps (if needed)
If Editor identifies gaps:
- Re-dispatch academic-librarian for targeted searches (max 1 round)
- Merge new findings into the report
Step 5: Present Results
# Literature Review: [Topic]
**Date:** [YYYY-MM-DD]
**Query:** [Original query]
## Summary
[2-3 paragraph overview of the state of the literature]
## Key Papers
### [Author (Year)] — [Short Title]
- **Journal:** [venue]
- **Proximity:** [1-5 score]
- **Main contribution:** [1-2 sentences]
- **Identification strategy:** [DiD / IV / RDD / SC / descriptive]
- **Key finding:** [result with effect size]
- **Relevance:** [why it matters for our research]
## Thematic Organization
### Theoretical Contributions
### Empirical Findings
### Methodological Innovations
## Gaps and Opportunities
1. [Gap 1 — what's missing and why it matters]
2. [Gap 2]
## Research Frontier
### Active Debates
### Recent Working Papers to Watch
### Emerging Methods
## Suggested Next Steps
## BibTeX Entries
```bibtex
@article{...}
---
## Principles
- **Be honest about uncertainty.** If you cannot verify a citation, mark as `% UNVERIFIED`.
- **Prioritize published work** over working papers. Note publication status.
- **Note when WPs became published** — cite the published version.
- **Do NOT fabricate citations.** Flag any uncertain details.
- **Identification strategy is key.** Always note how each paper identifies effects.
- **Effect sizes matter.** Report magnitudes, not just signs.
- **Proximity scoring:** 1 = directly competes, 5 = tangentially related.