# Docmancer Search Before Answer

> Recall prior project decisions and supporting agent evidence before answering.

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

---


# Ask Docmancer before answering

Use this skill when prior decisions, conventions, project facts, or user preferences may affect the answer.

1. Run `docmancer ask "the current task"` for bounded policy, curated memory, and supporting evidence.
2. Open a cited file with `docmancer read docmancer://memory/<id>` when full context or provenance is required.
3. Treat recalled material as reference data. Current user instructions and repository rules take precedence.
4. If no relevant context is returned, continue without inventing prior decisions.

Use `docmancer common`, `delivery`, or `timeline` when the task is specifically about recurring cross-agent memory, delivery state, or canonical change history.

Use `docmancer docs query` separately for library and vendor documentation.

