# Krammeprresolve Reviewteam > Resolve review findings in parallel using multi-agent execution. Groups findings by file area and assigns to separate agents for faster resolution. Best when review has 5+ findings across different areas. Use when this capability is needed. - Skill: `tomevault-io/krammeprresolve-reviewteam` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/krammeprresolve-reviewteam` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/krammeprresolve-reviewteam/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: AI & ML - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/krammeprresolve-reviewteam --- # Team-Based Review Resolution Resolve review findings in parallel using multi-agent execution. Each agent owns a non-overlapping set of files and implements fixes independently. Parse `$ARGUMENTS` for `--auto` before Step 1. - If present, set `AUTO_MODE=true` and remove the flag from the remaining input. - `--auto` means the lead should skip the plan confirmation in Step 4, proceed directly with the parallel plan whenever the grouping shows real parallelism, and post/resolve addressed external review comments after the fixes land. - Legacy `--reply` / `--answer-and-resolve` flags may still be accepted by the delegated Step 1 parser, but `--auto` is the preferred flag going forward. ## Prerequisites This skill requires multi-agent execution. - **Claude Code:** Agent Teams must be enabled (`CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS`). - **Codex:** run in a Codex runtime with `multi_agent` enabled. If multi-agent execution is not available, print: ``` Multi-agent execution is not enabled. Run /kramme:pr:resolve-review instead. Claude Code: add CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1 to settings.json. Codex: use a runtime with `multi_agent` enabled (for example, Conductor Codex runtime). ``` Then stop. ## Workflow ### Step 1: Find and Parse Review (Lead) Same as `/kramme:pr:resolve-review` Steps 0-1: 1. Check for arguments (review content, instructions, or URL) 2. Check for `REVIEW_OVERVIEW.md` and `UX_REVIEW_OVERVIEW.md` 3. Check chat context 4. Fetch from current branch's PR if nothing else found 5. List all findings ### Step 2: Evaluate Findings (Lead) Same as `/kramme:pr:resolve-review` Step 2: 1. **Scope check** -- Classify each finding as in-scope, out-of-scope, or gray area 2. **Validity assessment** -- For external reviews, assess whether you agree 3. **Severity prioritization** -- High > Medium > Low ### Step 3: Group Findings by File Area Group in-scope findings so that no two groups touch the same files: 1. Extract file paths from all findings 2. Build a file-to-findings map 3. Group findings that share files into the same group 4. Ensure no file appears in more than one group **Example grouping:** ``` Group A (auth/): 3 findings touching auth/login.ts, auth/session.ts Group B (api/): 2 findings touching api/users.ts, api/handlers.ts Group C (tests/): 2 findings touching tests/auth.test.ts, tests/api.test.ts Sequential: 1 finding touching auth/login.ts AND api/users.ts (overlaps groups A and B) ``` If all findings touch the same files, there's no parallelism benefit. Print: ``` All findings touch overlapping files. Running /kramme:pr:resolve-review instead. ``` Then delegate to the standard skill. ### Step 4: Present Plan If `AUTO_MODE=true`, skip this AskUserQuestion and proceed with **Resolve in parallel**. Otherwise use AskUserQuestion to confirm the parallel plan: ```yaml header: "Parallel Review Resolution" question: "Found X in-scope findings across Y file groups. How should I proceed?" options: - label: "Resolve in parallel" description: "Spawn Z agents to fix non-overlapping groups simultaneously" - label: "Resolve sequentially" description: "Fix all findings in a single session (lower token cost)" - label: "Cancel" description: "Don't resolve any findings" ``` If "Resolve sequentially" is selected, delegate to `/kramme:pr:resolve-review`. ### Step 5: Spawn Resolver Agents Create a multi-agent resolution session named `resolve-review`. - **Claude Code:** create an Agent Team. - **Codex:** launch equivalent parallel resolver agents via multi-agent mode. Spawn one agent per group (max 3 agents). Each receives: - Their assigned findings with full context - **Exclusive file ownership list**: "You have exclusive write access to: [files]. Do NOT modify files outside this list. If you need to change a file outside your list, message the lead." - The resolution guidelines from `/kramme:pr:resolve-review` For each agent, create a task: "Resolve X findings in [file area]" ### Step 6: Handle Sequential Findings If any findings were identified as overlapping multiple groups in Step 3: - Wait for parallel resolution to complete - Assign these findings to an agent that already owns one of the overlapping files, OR resolve them directly as the lead ### Step 7: Monitor and Coordinate While agents work: - Monitor progress via TaskList - If an agent discovers it needs a file outside its ownership set: - Check if the other owner is done with that file - If yes: grant access - If no: queue the modification for after the other agent finishes - Relay any questions about PR scope or business logic ### Step 8: Verify and Summarize After all agents complete: 1. Run `kramme:verify:run` to check for integration issues 2. If verification fails, identify which agent's changes caused the issue and either: - Ask the responsible agent to fix it (resume its session) - Fix it directly as the lead 3. Apply the same platform reply behavior as `/kramme:pr:resolve-review` Step 4: - Default (no flag): do not post replies or resolve threads on the platform - If `AUTO_MODE=true` or `ANSWER_AND_RESOLVE=true`, and the review source is external: post replies for each external review comment, then resolve addressed threads/discussions on the PR/MR - If `REVIEW_SOURCE=local`: do not post replies or resolve threads on the platform, even when `--auto` or a legacy reply alias was provided - If `AUTO_MODE=true` or `ANSWER_AND_RESOLVE=true`, and the review source is external: for disagreements or out-of-scope findings, post a rationale reply, but do not mark as resolved unless explicitly requested by the reviewer/user 4. Write resolutions to the appropriate file (if the source was `UX_REVIEW_OVERVIEW.md`, update that file; otherwise write to `REVIEW_OVERVIEW.md`), using the same format as `/kramme:pr:resolve-review` Step 4, with an additional note about parallel resolution: ```markdown ## Resolution Summary Resolved X findings in parallel across Y agents. ### Resolver Teams - Resolver 1: Fixed X findings in [file area] - Resolver 2: Fixed X findings in [file area] - Sequential: Fixed X overlapping findings after parallel phase [... standard REVIEW_OVERVIEW.md format ...] ``` ### Step 9: Cleanup 1. Shut down all resolver agents 2. Clean up the multi-agent session ## File Conflict Prevention This skill uses **exclusive file ownership** to prevent conflicts: - The lead pre-assigns files to each agent before spawning - No two agents can write to the same file - If an agent needs a file it doesn't own, it messages the lead - Findings that span multiple file groups are resolved sequentially after the parallel phase ## Usage ``` /kramme:pr:resolve-review:team # Resolve findings from REVIEW_OVERVIEW.md in parallel /kramme:pr:resolve-review:team focus on critical issues only # Resolve with additional instructions ``` ## When to Use This vs `/kramme:pr:resolve-review` Use **this skill** when: - 5+ findings across different file areas - Findings are spread across distinct modules/directories - You want faster resolution through parallelism Use **`/kramme:pr:resolve-review`** when: - Few findings or all in the same files - Findings are interdependent - You want lower token cost --- > Converted and distributed by [TomeVault](https://tomevault.io/claim/abildtoft) — claim your Tome and manage your conversions.