Apollo Research Synthesis

多Agent任务协调模式：研究阶段并行执行，综合阶段由主Agent自己完成。 避免让子Agent做综合判断——那是Coordinator的工作。 触发词：并行研究、综合、汇总、Coordinator模式、多Agent协作。

Security Scan

This is an instruction-only coordination pattern for multi-agent research/synthesis that is internally consistent and does not request extra credentials or install anything.

Purpose & Capability

Name, description, and SKILL.md all describe a multi-agent research/synthesis coordination pattern; the skill declares no binaries, env vars, or installs that would be out-of-scope.

Instruction Scope

Instructions focus on splitting research (parallel) and synthesis (coordinator), recommending session isolation (sessions_spawn) and fresh validators. This is coherent, but it implicitly assumes the agent has access to project files and an API to spawn isolated sessions—confirm that any runtime permissions you grant to your agent (repo/file access, session-spawn capability) are intentional.

Install Mechanism

No install spec and no code files — lowest-risk instruction-only skill.

Credentials

The skill requires no environment variables, credentials, or config paths; instructions do not reference secrets or unrelated credentials.

Persistence & Privilege

always:false and no persistent installs. The skill allows normal autonomous invocation (disable-model-invocation:false), which is the platform default — acceptable here but be mindful if you allow the agent broad autonomous actions.

Assessment

This skill is a guidance template (no code, no credentials) and appears coherent. Before enabling it, verify: (1) you trust the unknown source—there's no homepage or owner info beyond an ID; (2) what runtime permissions your agent has (access to repositories, files, or the sessions_spawn API), since the pattern assumes access to project files and the ability to spawn isolated sessions; (3) whether you want the agent to be allowed to invoke this skill autonomously—test in a sandbox or with limited permissions first, and review logs of any actions the agent takes when following these coordination instructions.

Apollo Research-Synthesis - 研究/综合分离原则

核心准则

多人并行研究，最后一个人汇总。不要让子Agent去综合——那是Coordinator的工作。

Claude Code Coordinator的工作流：

阶段	谁做	做什么
Research	Workers（并行）	各自研究不同方向
Synthesis	Coordinator（自己）	读取结果，形成实施规格
Implementation	Workers	按规格做改动
Verification	Workers	测试改动是否有效

关键规则

1. 研究阶段可以并行

• 独立的研究任务同时跑
• 各自探索不同文件/方向
• 不要让一个Worker等另一个

2. 综合阶段必须自己来

• ❌ "based on your findings"
• ✅ "Fix null pointer in src/auth/validate.ts:42"
• 综合者必须自己理解所有研究结果，不能转发给另一个Agent去综合

3. 验证必须 fresh eyes

• ❌ Continue（继续用刚写代码的Agent）
• ✅ Spawn fresh（新Agent来验证）
• 新Agent没有之前的假设，能发现问题

4. 失败优先 Continue

• 修正失败 → 继续用同一个Agent（它有错误上下文）
• 不要急着新Spawn

Fork 隔离原则

子Agent有独立对话上下文：

• 危险操作在独立空间跑
• 不会污染主对话
• 用 sessions_spawn 实现隔离

应用检查表

• 研究任务是否真的独立可以并行？
• 综合判断是否应该由主Agent自己完成？
• 验证是否需要新的"眼睛"（fresh Agent）？
• 是否有需要隔离的危险操作？

参考

来源：Claude Code coordinatorMode.ts，512,000行源码研究