synthesys-lab/agentize/external-synthesize

Claude Code and agent skill by synthesys-lab/agentize

Trust tier 2
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Synthesize implementation plan(s) from 5-agent debate using external AI - exposes disagreements as developer decisions

Install

npx vskill@latest install synthesys-lab/agentize/external-synthesize

SKILL.md

External Synthesize Skill

This skill synthesizes implementation plan(s) and exposes disagreements from a multi-agent debate with dual proposers using external AI review.

Modes

  1. Standard mode: Determines consensus/disagreement, generates options for disagreements
  2. Resolve mode: Same 5 reports but with "User Resolution" section appended, produces unified plan

Key Features

  • No Automatic Dropping: AI cannot drop ideas from proposals; all contested ideas become Disagreement sections
  • Mandatory Developer Arbitration: Disagreements require explicit developer selection via --resolve mode
  • Flexible Options: Minimum 2, recommended 3, no upper limit per disagreement
  • Source Attribution: Each option must cite its source (Bold, Paranoia, Hybrid, etc.)
  • AI Recommendations: Advisory only; developer must select
  • Collapsible Code Drafts: <details> tags for implementation details

Inputs

This skill requires 5 agent report file paths, with optional 6th and 7th arguments:

  • Report 1: Bold proposer report (.tmp/issue-{N}-bold.md)
  • Report 2: Paranoia proposer report (.tmp/issue-{N}-paranoia.md)
  • Report 3: Critique report (.tmp/issue-{N}-critique.md)
  • Report 4: Proposal reducer report (.tmp/issue-{N}-proposal-reducer.md)
  • Report 5: Code reducer report (.tmp/issue-{N}-code-reducer.md)
  • Arg 6 (optional): Previous consensus file (.tmp/issue-{N}-consensus.md)
  • Arg 7 (optional): History file (.tmp/issue-{N}-history.md)

Context ordering rationale: The combined report is assembled as: agent reports (1-5) -> previous consensus (6) -> history (7). This ensures the AI sees the history table's last row (current task) as the final context, leveraging LLM recency bias to prioritize the current request.

For resolve/refine modes: Pass both 6th and 7th arguments:

  1. consensus.md as arg 6 (previous plan being modified)
  2. history.md as arg 7 (operation history with current task in last row)

Outputs

Files created:

  • .tmp/issue-{N}-debate.md - Combined 5-agent debate report (7 parts if resolve mode)
  • .tmp/issue-{N}-consensus.md - Final plan(s)
  • .tmp/issue-{N}-history.md - Accumulated selection and refine history

For resolve/refine mode, the debate file includes additional sections:

  • Part 6: Previous Consensus Plan (from consensus.md)
  • Part 7: Selection & Refine History (from history file, last row = current task)

Output format:

SectionDescription
Agent Perspectives Summary5-agent position table
Consensus Status3-condition evaluation with CONSENSUS/DISAGREEMENT verdict
Goal / Codebase AnalysisProblem statement and file changes
Implementation StepsAgreed changes with code drafts
Disagreement N (if any)Per-disagreement options with A/B/C choices
Disagreement SummarySummary table and suggested combination
Validation (resolve mode)Selection history and compatibility check

See `external-synthesize-prompt.md` for complete output format specification.

Implementation Workflow

Step 1: Invoke External Synthesize Script
bash
.claude-plugin/skills/external-synthesize/scripts/external-synthesize.sh \
.tmp/issue-42-bold.md \
.tmp/issue-42-paranoia.md \
.tmp/issue-42-critique.md \
.tmp/issue-42-proposal-reducer.md \
.tmp/issue-42-code-reducer.md

Script automatically:

  1. Validates all 5 report files exist
  2. Extracts issue number from first report filename
  3. Combines all 5 reports into debate report
  4. Invokes external AI (Codex or Claude Opus)
  5. Determines if consensus or multiple options
  6. Saves result to consensus file

Timeout: 30 minutes (same as external-consensus)

Error Handling

Error MessageCauseSolution
Report file not found: {path}Missing agent reportEnsure all 5 reports were generated
External review failed with exit code {N}API/network issueCheck credentials and retry

Usage Examples

ModeArgumentsExample
Standard5 reports.tmp/issue-{N}-bold.md ... .tmp/issue-{N}-code-reducer.md
Resolve5 reports + consensus + historyAdd .tmp/issue-{N}-consensus.md .tmp/issue-{N}-history.md
RefineSame as resolveSame issue-{N} prefix, history records refine operation
bash
# Standard mode
.claude-plugin/skills/external-synthesize/scripts/external-synthesize.sh \
.tmp/issue-42-bold.md \
.tmp/issue-42-paranoia.md \
.tmp/issue-42-critique.md \
.tmp/issue-42-proposal-reducer.md \
.tmp/issue-42-code-reducer.md

Output (stdout, last line):

.tmp/issue-42-consensus.md

Security

  • Security scanPassed, 100 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

Repository stars
55
Repository forks
10
7-day trend score
20

Stars and forks are for the whole GitHub repository, not this skill alone. Refreshed 21 Mar 2026, 7 months ago.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/synthesys-lab/agentize/external-synthesize/badge)](https://verified-skill.com/skills/synthesys-lab/agentize/external-synthesize)

Something wrong with this skill or its page? Tell us.