# Literature Synthesis

> Synthesize multiple sources into coherent narratives with explicit evidence tracking and confidence ceilings.

- Skill: `majiayu000/literature-synthesis` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add majiayu000/literature-synthesis`
- Raw SKILL.md: https://api.skillmd.com/api/skills/majiayu000/literature-synthesis/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Research & Search
- Author: majiayu000 (https://skillmd.com/u/majiayu000)
- Updated: 2026-09-09
- Page: https://skillmd.com/skills/majiayu000/literature-synthesis

---




## STANDARD OPERATING PROCEDURE

### Purpose
- Combine findings across papers/articles into concise, evidence-backed syntheses.
- Expose constraints and assumptions clearly to avoid overgeneralization.
- Maintain structure-first artifacts for reuse and traceability.

### Trigger Conditions
- **Positive:** literature reviews, state-of-the-art summaries, related work sections.
- **Negative:** single-source summaries (use `academic-reading-workflow`), or prompt-only edits (`prompt-architect`).

### Guardrails
- Constraint buckets: HARD (scope, dates, domains), SOFT (style, depth), INFERRED (coverage, exclusions) with sources.
- Two-pass refinement: synthesis structure → epistemic audit with evidence and ceilings.
- Cite sources inline with links/pages; avoid claim aggregation without evidence.

### Inputs
- Research question and intended audience.
- Source list with metadata and access.
- Formatting needs (bullets, narrative, tables) and timebox.

### Workflow
1. **Scope & Constraints**: Define inclusion/exclusion criteria and outputs; confirm INFERRED assumptions.
2. **Source Mapping**: Group sources by theme and strength; note gaps or conflicts.
3. **Draft Synthesis**: Build outline, then fill with claims tied to sources and confidence ceilings.
4. **Epistemic Pass**: Check for overclaims, contradictions, and missing evidence; rebalance coverage.
5. **Deliver & Store**: Provide synthesized narrative plus evidence table; update references/examples.

### Validation & Quality Gates
- Every claim tied to at least one source with confidence ceiling.
- Coverage matches constraints and inclusion criteria.
- Conflicts highlighted with proposed resolutions or follow-ups.

### Response Template
```
**Scope & Constraints**
- HARD / SOFT / INFERRED.

**Key Themes**
- Theme → evidence → confidence ceiling.

**Conflicts / Gaps**
- ...

**Next Steps**
- ...

Confidence: 0.81 (ceiling: research 0.85) - based on cross-source synthesis and evidence tracking.
```

Confidence: 0.81 (ceiling: research 0.85) - reflects completed synthesis with evidence-linked claims.