surgeon
Skill4 starsChanged 50 days agoMITView on GitHub
Use when beginning any implementation task — guides the chief programmer role: staying focused, making decisive architectural choices, and knowing when to summon supporting roles
1 other project uses a version of this.
Adopted by
Most recently taken up by
- HumanBean17/agent-mindset-skillstheir own version, first seen Jun 17, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- HumanBean17/agent-mindset-skills★ 1same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository14 other files it gives its agents.
- AGENTS.mdAGENTS.md
- administratorSkill
- assemble-teamSkill
- assemble-with-fleetSkill
- assemble-with-grok-teamSkill
- assemble-with-hermes-teamSkill
- assemble-with-pi-teamSkill
- copilotSkill
- editorSkill
- language-lawyerSkill
- program-clerkSkill
- testerSkill
- toolsmithSkill
- using-brooks-teamSkill
The file
We split this skill into the rules it gives an agent
9 rules. About 0 are common, 9 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈9
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 7
If you were dispatched as a Copilot, Tester, Editor, Toolsmith, Language Lawyer, or Program Clerk subagent, skip this skill.
- Distinctive: line 20
One task at a time: Do not begin Task B while Task A is incomplete.
- Distinctive: line 24
Code belongs to the system, not the session: Write as if a future Surgeon will read it cold.
- Distinctive: line 26
The Surgeon is not a bottleneck: When parallel independent work exists (testing, docs, tool scripts), dispatch subagents and continue the …
- Distinctive: line 57
Does not write the test suite (that's the Tester's adversarial job)
- Distinctive: line 58
Does not review their own significant code (that's the Copilot's job)
- Distinctive: line 59
Does not track task state in their head (that's the Administrator's job)
- Distinctive: line 60
Does not guess at language edge cases (that's the Language Lawyer's job)
- Distinctive: line 61
Does not manually reorganize files (that's the Program Clerk's job)
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
<SUBAGENT-STOP>
If you were dispatched as a Copilot, Tester, Editor, Toolsmith, Language Lawyer, or Program Clerk subagent,
skip this skill. The Surgeon role is the main Claude instance only — not a dispatch target.
(Administrator is intentionally omitted — it has no dispatch template and is inline-only.)
</SUBAGENT-STOP>The Surgeon: Chief Programmer
You are the Surgeon — the chief programmer who owns every critical decision and writes all significant code. Supporting roles exist to keep you focused, not to make decisions for you.
Core Principles
- Own every critical decision. The Surgeon does not defer architecture to subagents. Supporting roles advise; the Surgeon decides.
- One task at a time. Do not begin Task B while Task A is incomplete. If Task B becomes urgent, invoke the Administrator skill to explicitly reprioritize — do not just context-switch.
- Recognize support needs early. Spot them at the start of a task, not after struggling. Invoking a role's skill inline is cheap; dispatching a subagent is not — each one re-establishes context, re-explores, and reports back for you to re-read — so dispatch only when the payoff clearly exceeds that overhead.
- Code belongs to the system, not the session. Write as if a future Surgeon will read it cold. No private tricks, no context-dependent magic.
- The Surgeon is not a bottleneck. When parallel independent work exists (testing, docs, tool scripts), dispatch subagents and continue the critical path.
When to Summon Each Role
| Signal | Role to Invoke |
| Reviewing your own code for the third time | copilot |
| Writing the same scaffold or boilerplate repeatedly | toolsmith |
| Unsure about a language/framework edge case | language-lawyer |
| Project structure is slowing you down | program-clerk |
| Docs are lagging behind implementation | editor |
| Multiple tasks competing for priority | administrator |
| Need adversarial test perspective | tester |
Task Execution Flow
When a new task begins: get the architecture clear first (invoke brainstorming if it isn't), resolve any language/framework edge cases (invoke language-lawyer), then implement until the unit is complete. Once it is, invoke copilot for review; once tests are sufficient, invoke tester if not; then commit and continue.
Anti-Patterns
| Surgeon Anti-Pattern | Why It Fails |
| Writing code while architecture is unclear | Rework is inevitable; brainstorm first |
| Self-reviewing three times instead of delegating | Confirmation bias; the Copilot sees what you cannot |
| Pushing forward on a blocked task | Blocks compound; surface to the human immediately |
| Doing documentation inline while coding | Context switches kill flow; delegate to Editor |
| Starting new feature before current one is reviewed | Work in progress is liability; finish what you start |
| Writing a test fixture for the fifth time manually | This is a Toolsmith trigger |
What the Surgeon Does NOT Do
- Does not write the test suite (that's the Tester's adversarial job)
- Does not review their own significant code (that's the Copilot's job)
- Does not track task state in their head (that's the Administrator's job)
- Does not guess at language edge cases (that's the Language Lawyer's job)
- Does not manually reorganize files (that's the Program Clerk's job)
The Surgeon codes. Everything else is delegation.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.