---
name: multi-agent-analysis-resolve
language: en
description: "Resolve the Section 20 Risks and Open Questions of an analysis v3 document one row at a time: up to 3 source-labeled answer candidates per row (evidence / repo / AI reasoned), chosen answer merged into the target section, doc updated in place with a changelog bump. Same locked decisions as /multi-agent:analysis. Use when an analysis document's open questions need answering before development starts."
user-invocable: true
argument-hint: "[path/to/analysis/<feature>-<platform>.md] [--autonomous]"
---

# multi-agent-analysis-resolve - Open Question Resolver for Analysis Docs

**Input**: $ARGUMENTS (optional doc path; auto-globbed from `analysis/*-{ios,android,backend,frontend}.md` when empty)

Companion to `multi-agent-analysis`. Walks the open rows of Section 20 (Risks and Open Questions) one at a time, proposes source-labeled answer candidates, and merges the chosen answer into the proper body section. The doc is authoritative and forward-looking; the resolver never invents an answer.

> **Language**: instruction prose stays English; picker `question` / `description` follow `outputLanguage`; fragments written INTO the doc follow the doc's own front-matter `language` field.

## Core rules

- **One question per picker call** - never batch Section 20 rows; sequential resolution keeps each decision traceable.
- **Up to 3 candidates + Defer per question.** Sources, never blended: `From evidence` (the doc's own sections and citations), `From repo` (one batched code-exploration pass with `file:line` citations), `AI reasoned` (one-line rationale required).
- **Never invent.** No credible candidate means only Defer and free-text remain.
- **Stop tokens** typed into the free-text field halt cleanly: `stop`, `pause`, `dur`, `kes`.
- **Save after every answer** - atomic full-file writes; interruption loses nothing.
- **No Figma access.** Rows that need new design information get only Defer plus a printed recommendation to re-run `multi-agent-analysis`. Repo reads and convention re-extraction are allowed.
- **No auto-commit**; the user reviews the diff and commits manually.
- **Status enum fixed**: `Acik / Open`, `Girdi bekleniyor / Pending input`, `Karar verildi / Decided`. Resolved rows stay in the table as a decision log with a trailing `[Resolved: <value> - <source>]` fragment.

## Flow

| Phase | Action |
|-------|--------|
| 0. Intake | Doc picker (auto-glob), repo root for `From repo` candidates, mode (interactive / autonomous) |
| 1. Parse | Front-matter + Section 20 rows + reasoning context; classify each row (convention-fallback / reuse-vs-new / citation-tbd / legacy-decision / standards-conflict / design-gap / generic); one batched repo lookup |
| 2. Resolve | Sequential loop: candidates per class, picker, merge into target section, save; follow-up questions insert immediately after the current row |
| 3. Finalize | Optional sibling propagation (verbatim-matching rows in sibling platform files, one approval question), changelog bump, punctuation gate, report |

## Detailed implementation

Full steps: `$HOME/.claude/commands/multi-agent/analysis-resolve/SKILL.md`.
Section 20 / changelog contracts: `$HOME/.claude/multi-agent-refs/analysis-template.md`.
Convention defaults: `$HOME/.claude/multi-agent-refs/conventions-defaults.md`.

## Notes

- Convention-fallback resolutions rewrite the affected Section 13 cell footnote to `^[user-override: resolved via analysis-resolve <date>]`.
- Repo-sourced answers describing current behavior merge as `> Legacy reference: <text> (file:line)` blockquotes, never as primary content.
- `evidence_digest` and the `generated` timestamp in front-matter stay untouched; only the changelog records the revision.
- Re-running skips `Karar verildi / Decided` rows (idempotent).