Skip to content

Engineer System Change

Evaluates and carries out software changes by checking the real need, consumers, and risks first

Stars
84,876
Licence
MIT
Files
1 file

What it does

When installed, the agent treats a proposed software change (an RFC, issue, feature, refactor, migration, or new field, API, or service) as a hypothesis to test rather than a task to build. It reads the actual codebase, checks whether the problem is real and who uses the change, prefers the smallest sufficient fix, and asks for proportionate evidence and a verification plan before and after any edits. It is meant for assessment and careful implementation, and it explicitly avoids making changes when only asked to assess or plan.

Great for

  • Engineers reviewing RFCs or feature proposals and wanting a rigorous first-principles check
  • Teams who want to avoid speculative abstractions, unused fields, or oversized migrations
  • Developers who want an agent to implement a change only after justifying it against real consumers

Skip it if

  • Mechanical edits such as renames or formatting, which the skill says it should not be used for (the process adds overhead without benefit)
  • Simple explanations of how existing source code works, which fall outside its stated scope
  • Reviewing an already-finished diff, which the skill excludes in favour of a dedicated review process

Try asking

  • “Assess this RFC for adding a new event field to our task service and tell me whether a consumer actually needs it”
  • “Evaluate whether we need a new abstraction for this retry logic, or whether an existing mechanism already covers it”
  • “Plan a migration of our user table without implementing it yet, listing the risks and the verification we would need”
  • “Implement the smallest fix for this issue after first confirming the current failure path in the code”
Step-by-step guide: how to use Engineer System ChangeInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 1 original files, the MIT licence and an attribution note.

    engineer-system-change.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip engineer-system-change.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip engineer-system-change.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Say explicitly whether you want an assessment only or implementation, since the skill makes no project changes unless implementation is requested
  • Point the agent at the current target revision and the relevant issue or discussion, because the skill warns against relying on stale checkouts or the RFC alone
  • Expect verdicts such as needs evidence or reduce the proposal; these are intended outputs, not failures

What's inside

  • SKILL.md

Redistributed unmodified from bytedance/deer-flow under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.