Feedback Observer — Development agent for Claude Code
用户给出修正或反馈后,由主 Agent 派发.
How to install Feedback Observer
Installs to ~/.claude/agents/zylimit-cc-base-feedback-observer.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/zylimit/cc-base/HEAD/.claude/agents/feedback-observer.md -o ~/.claude/agents/zylimit-cc-base-feedback-observer.md Restart Claude Code, or start a new session, for it to be picked up.
What Feedback Observer does
name: feedback-observer description: 用户给出修正或反馈后,由主 Agent 派发。使用 feedback-writer skill 分析并记录 feedback。 skills: feedback-writer model: sonnet color: blue disallowedTools: Bash, Task maxTurns: 25
[角色] 你是一名观察员,专门分析用户的反馈和修正,将有价值的信号记录为结构化 feedback。
你不替用户总结——你基于主 Agent 提供的上下文,判断有没有值得记录的信号。
没有信号就说没有,不强行制造 feedback。
[任务] 收到主 Agent 派发后,使用 feedback-writer skill: 1. 分析传入的上下文,识别是否有 feedback 信号(观察维度 1-5) 2. 有信号 → 写入 feedback 文件 + 更新索引 3. 无信号 → 返回"无新 feedback"
[Non-goals]
Alternatives in Development
- Brag Spotter — Proactively scans for achievements and wins that aren't in the brag doc yet 4.6k ★
- Omd Critic — Reads the full run output, user feedback, and persona ABANDONs, then writes a root-cause critique of omd-maste 465 ★
- General PM — Use this agent proactively when you need comprehensive product management oversight for software development i 248 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
PR Comment Poster
Stage 4 of the PR review pipeline. Takes finished comments from pr-feedback-writer and posts them to the pull
PR Feedback Writer
Stage 3 of the PR review pipeline. Turns verified findings from pr-code-analyzer into final, ready-to-post Git
Skill Health Observer
Analyze skill outcomes, score per-skill health, write SKILL_HEALTH.md. Haiku-powered, ~$0.02/run.
Act 5 Observer
Read-only narrative steward for Act V (slides 19-21) — the Observer agent demo that proves durable tool-call r
Runtime Observer.Agent
Use when Architrave needs runtime/product truth: health, logs, versions, deployed digests, app launch, or drif
Runtime Observer
Interprets imported runtime evidence (traces, metrics, profiles, topology) against the static graph and report
Related Skills
Content Research Writer
Assists in writing high-quality content by conducting research, adding citations, improving hooks, iterating o
Task Observer
A meta-skill that builds and improves all your skills, including itself.
Observer
Local-first control plane for AI coding agents — launch browser terminals, run and remotely control sessions,