# Source Drafting

> Use when drafting documents, reports, posts, presentations, or review comments from source materials, or conducting research across provided documents. Enforces read-first drafting, a source map for every claim, agent verification, and separation of verified from unverified claims.

- Skill: `nicknisi/source-drafting` (Agent Skill)
- Install (CLI): `npx skillmds@latest add nicknisi/source-drafting`
- Raw SKILL.md: https://api.skillmd.com/api/skills/nicknisi/source-drafting/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: nicknisi (https://skillmd.com/u/nicknisi)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/nicknisi/source-drafting

---


# Source Drafting Protocol

When drafting documents or conducting research from source materials:

1. **Read first, write second.** Read all provided source documents fully before drafting. Do not begin writing until all sources are loaded.
2. **Maintain a source map.** Track every factual claim, metric, name, or date back to its source. Present the draft clean (no inline tags), with a "Source Map" appendix listing each claim and its origin (document name, section/heading).
3. **Verify before delivering.** For substantive documents (strategy docs, external-facing reports, review comments, posts, presentations), spawn a verification agent that re-reads each source and checks every claim in the source map. Mark any unverifiable claim as [UNVERIFIED].
4. **Separate verified from unverified.** Present the clean draft with unverified claims removed, plus a separate list of removed claims so I can decide whether to add them back with proper sourcing.
5. **No invention.** Never generate statistics, percentages, quotes, or specific details not found in the sources - even if they seem plausible or "directionally correct."