mspec-research

name	mspec-research
description	research step of mspec workflow — investigate via subagent and produce research.md
when_to_use	User runs /mspec:research, or workflow auto-continues to research

Procedure

1. Run mspec status --change <change-dir> --json first.
2. Run mspec continue --change <change-dir> --json to fetch the ready-made subagent_prompt.
3. Read requires artifacts (proposal.md, specs/*/spec.md).
4. Invoke the mspec-researcher subagent via the Task tool with subagent_prompt. It will perform web search and codebase analysis.
5. Receive the research.md body (Decisions / Web References / Codebase Findings / Open Choices) and write it to disk.
6. For any Open Choices, ask the user via AskUserQuestion (1 question per call) and update the file.
7. Append the ## Constitution Check (Phase 0 only). 7a. readme.md の ## Artifacts 節の - [ ] research.md を - [x] research.md に更新する。
8. Run mspec validate --change <change-dir>. validate が失敗した場合は - [x] research.md を - [ ] research.md にロールバックする。
9. block: true — stop and ask the user to run /mspec:continue.

Observation (Agent Experience Log)

After the subagent completes (step 4), record the run:

mspec agent-run record research \
  --change <change-name> \
  --bytes <sum-of-input-artifact-bytes> \
  --artifacts <space-separated-artifact-paths>

Example: mspec agent-run record research --change 2026-05-25-my-feature --bytes 4821 --artifacts proposal.md specs/my-cap/spec.md

This appends one JSONL entry to changes/<change>/.agent-runs.jsonl. Do NOT include prompt text or file contents.

Verification (C2)

• mspec validate --change <change> — アーティファクト整合性チェック
• mspec anchor check --change <change> — アンカー解決確認

Learning (C3)

このスキルの実行で発生した学習候補を記録する:

<!-- LEARNING: <パターン説明> | source: <FR-ID> | confidence: low|medium|high -->

mspec learn コマンドが archive 済み changes からこれらを収集してpost-condition候補をproposeする。