From superpowers-plus
Merges external research (Perplexity, web searches, ChatGPT outputs) into existing documents while stripping artifacts, preserving voice, and confirming scope before editing.
How this skill is triggered — by the user, by Claude, or both
Slash command
/superpowers-plus:incorporating-researchThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
> **Guidelines:** See [CLAUDE.md](../../../CLAUDE.md) for writing standards.
Guidelines: See CLAUDE.md for writing standards. Last Updated: 2026-01-26
This skill handles incorporating external research (Perplexity, web searches, ChatGPT outputs, etc.) into existing documents without breaking voice, structure, or adding irrelevant content.
Core principle: Triage first, confirm scope, then edit. Strip artifacts, preserve voice.
Invoke when user says:
Red flag: User says "incorporate" but might mean "review for quality" — clarify if ambiguous.
Ask if unclear:
Confirm:
[path/to/file.md]. Correct?"Separate signal from noise:
| Category | What to Look For | Action |
|---|---|---|
| New content | Relevant facts, data, examples | Extract for incorporation |
| Irrelevant content | Wrong topic, unrelated sections | Strip completely |
| Citation artifacts | [1][2][3], "Sources:", URLs | Strip unless explicitly requested |
| Formatting quirks | Perplexity/ChatGPT styling | Normalize to doc style |
| Hallucinations | Suspicious claims, outdated info | Flag for user verification |
Output triage summary:
Triage Results:
├── New content: [Brief description]
├── Irrelevant: [What's being stripped]
└── Artifacts: [Citation numbers, source sections, etc.]
Read target document first. Determine: new section, inline addition, appendix, or replacement. Flag conflicts/duplicates.
NEVER edit without confirmation. Show: changes to [file], what to strip, concerns. Wait for approval.
Match: header levels, list style, tone, code/link formatting.
Strip: citation numbers [1], source/reference sections, AI metadata ("Based on X sources", "Here's a summary"), unrelated content.
Report: what was added and where, what was stripped, any concerns.
| Red Flag | Example | Action |
|---|---|---|
| Topic mismatch | Kamailio versions in hardware procurement doc | Strip completely, ask user |
| Ambiguous intent | "Incorporate this" (could mean review) | Clarify: "Add to doc or review for quality?" |
| Hallucinations | Outdated versions, wrong dates | Flag: "This claims X, verify?" |
| Contradictions | New content conflicts with existing | Flag before incorporating |
| Duplicate content | Research repeats what's already there | Ask: "This duplicates section Y. Replace or skip?" |
❌ Reviewing instead of incorporating — User says "incorporate" but wants quality review
❌ Breaking document voice — Pasting ChatGPT-style prose into technical doc
❌ Leaving citation artifacts — [1][2][3] scattered throughout
❌ Adding irrelevant content — Including unrelated sections from paste
❌ Editing without confirmation — Making changes before user approves scope
✅ Triage → Map → Confirm → Edit → Summarize
User: "Incorporate this Perplexity output into the hardware doc"
Agent:
I'll incorporate this into `docs/hardware-procurement.md`. Let me triage the input first.
Triage Results:
├── New content: GPU specs for H100 vs A100
├── Irrelevant: Section on Kamailio SIP server versions (wrong topic)
└── Artifacts: Citation numbers [1][2][3], "Sources" section with 5 URLs
I'll make these changes:
1. Add GPU comparison table to "Hardware Options" section
2. Strip Kamailio section (unrelated to hardware)
3. Strip citation numbers and Sources section
Concerns:
- Claims H100 has 80GB VRAM, but some models have 96GB. Verify?
Proceed?
User: "Yes, use 80GB for now"
Agent: [Makes edits]
Incorporated into docs/hardware-procurement.md:
Added:
- GPU comparison table (H100 vs A100) in "Hardware Options" section
Stripped:
- Kamailio SIP server section (unrelated)
- Citation numbers [1]-[5]
- Sources section with URLs
Concerns:
- None (H100 VRAM confirmed as 80GB for standard model)
| Resource | Purpose |
|---|---|
| CLAUDE.md | Writing standards (no AI slop) |
eliminating-ai-slop | Use after incorporating to clean up pasted text |
detecting-ai-slop | Score research quality before incorporating |
grep -n "TODO.*verify" draft.md # find unverified claims before publish
| Failure | Fix |
|---|---|
| Copy-paste without attribution | Cite the source inline |
| Stale sources | Check publication date before incorporating |
| AI-slop injection | Run eliminating-ai-slop after pasting |
npx claudepluginhub bordenet/superpowers-plus --plugin superpowers-plusGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.