Structured Autonomy Bundle of 3 skills
- Use it when:
- You want a feature researched and planned, and every step's code written out, before an agent applies any of it.
- It does:
- Plans the feature as the commits of one pull request, writes each step's complete code into a document, then applies it one step at a time.
- You get:
- plans/{feature-name}/plan.md, plans/{feature-name}/implementation.md and the code, with a stop after each step for you to test and commit.
- Needs:
- Subagents for the generate step's research; the plan step does its own research when it can't start one.
- Written for:
- GitHub Copilot: the prompts name its tools, as in #tool:runSubagent, and its CLI plugin installs the three as slash commands.
TL;DR
Structured Autonomy splits a feature into three passes, one skill each. structured-autonomy-plan writes no code: it researches the codebase, docs and libraries, in a subagent when it can, and breaks the request into the commits of one pull request on its own branch, a single commit for a simple feature or several testable ones for a complex one. It saves plans/{feature-name}/plan.md with the branch, the goal and each step's files, change and test, marks open questions [NEEDS CLARIFICATION], asks them and waits for your feedback. structured-autonomy-generate reads that plan, has a subagent research the project's stack, conventions, patterns and library docs once, and writes plans/{feature-name}/implementation.md: each step as checkboxes with complete code to paste into named files, a verification checklist and a STOP & COMMIT point, with no decisions left open. structured-autonomy-implement then follows that file exactly: it picks up the next unchecked step, writes only the code the plan specifies, checks items off, runs the plan's build or test commands and stops at the STOP marker, so you can test and commit before calling it for the next step. The plugin that ships the three sums them up as "premium planning, thrifty implementation".
Summarizes the repo at 797337f.
Any agent, with the skills CLI
npx skills add https://github.com/github/awesome-copilot --skill structured-autonomy-plan --skill structured-autonomy-generate --skill structured-autonomy-implement GitHub Copilot CLI plugin, run as /structured-autonomy:structured-autonomy-plan and so on; an older CLI first needs copilot plugin marketplace add github/awesome-copilot
copilot plugin install structured-autonomy@awesome-copilotSource on GitHubAll skills from github/awesome-copilot
- Total Downloads
- 8,800 of structured-autonomy-plan, its most installed skill, on skills.sh, Nov 9, 2026
- Repo stars
- 40,943 github/awesome-copilot
- First seen
- Feb 25, 2026 on skills.sh, its first skill
Its skills, in the order they're used
The ones marked “Only in the bundle” need the rest of it. The others also work alone.
-
structured-autonomy-plan
8.8K total downloadsResearches a feature and saves a plan of one pull request split into testable commits, then pauses for your feedback before any code.
-
Turns a structured-autonomy plan.md into an implementation.md with complete, copy-paste code for every step, checkboxes, verification checks and commit stops.
Needs the plans/{feature-name}/plan.md that structured-autonomy-plan saves, and takes the feature name, branch, steps and each step's files from it.
-
Carries out an implementation plan exactly as written, one step at a time, checking off items and stopping where the plan says to.
Carries out the implementation.md that structured-autonomy-generate writes, from its next unchecked step to its next STOP & COMMIT point; without a plan passed in, it refuses to start.