structured-autonomy-implement

github/awesome-copilot

Executes a given implementation plan exactly, step by step, with no scope creep.

What is structured-autonomy-implement?

A strict implementation-agent prompt that executes a pre-written implementation plan step by step, without deviation. It enforces plan-only changes, inline progress tracking via checkboxes, build/test verification, and stopping at designated STOP points to hand control back to the user.

Requires an explicit implementation plan as input; refuses to proceed without one.
Follows the plan exactly in order, starting from the next unchecked step.
Restricts all code changes to only what is specified in the plan.
Checks off completed items in the plan document using markdown checkbox syntax.
Runs build or test commands specified in the plan to verify completed work.
Stops at STOP instructions in the plan and returns control to the user.

How to install structured-autonomy-implement

npx skills add https://github.com/github/awesome-copilot --skill structured-autonomy-implement

Prerequisites

A pre-written implementation plan document with checkbox-style steps and STOP markers to feed to the agent

Claude Code

Cursor

Windsurf

Cline

How to use structured-autonomy-implement

1.Prepare a written implementation plan with discrete steps and checkbox items, including STOP markers where control should return to the user.
2.Pass the implementation plan as input to the agent when invoking the skill.
3.Let the agent pick up at the next unchecked step in the plan and implement only what that step specifies.
4.Allow the agent to check off completed items inline in the plan document as it progresses.
5.Let the agent run the build/test commands specified in the plan to verify the work.
6.Resume the agent after a STOP point once you've reviewed progress, to continue with subsequent steps.

Use cases

Good for

Executing a multi-step implementation plan produced by a separate planning phase, without the agent improvising.
Working through a long feature build incrementally, stopping at checkpoints for user review.
Ensuring an AI coding agent doesn't introduce unplanned changes during implementation.
Resuming implementation work across sessions by tracking checked-off steps in the plan document.
Validating each implementation step against build/test commands defined in the plan.

Who it's for

Developers using AI coding agents who want controlled, predictable execution of a pre-approved plan.
Teams that separate planning and implementation phases in their AI-assisted development workflow.
Engineers who need checkpoint-based review during long or multi-session implementations.

structured-autonomy-implement FAQ

What happens if I don't provide an implementation plan?

The agent will respond with "Implementation plan is required." and will not proceed.

Will the agent make improvements or changes beyond the plan?

No. It is restricted to implementing only what is explicitly specified in the plan, with no additional code changes.

Does the agent track progress through the plan?

Yes, it updates the plan document inline by checking off completed items using standard markdown syntax.

Does the agent verify its work?

Yes, it runs the build or test commands specified in the plan to check its work.

When does the agent stop working?

It stops when it reaches STOP instructions defined in the plan and returns control to the user.

Full instructions (SKILL.md)

Source of truth, from github/awesome-copilot.

name: structured-autonomy-implement description: 'Structured Autonomy Implementation Prompt'

You are an implementation agent responsible for carrying out the implementation plan without deviating from it.

Only make the changes explicitly specified in the plan. If the user has not passed the plan as an input, respond with: "Implementation plan is required."

Follow the workflow below to ensure accurate and focused implementation.

<workflow> - Follow the plan exactly as it is written, picking up with the next unchecked step in the implementation plan document. You MUST NOT skip any steps. - Implement ONLY what is specified in the implementation plan. DO NOT WRITE ANY CODE OUTSIDE OF WHAT IS SPECIFIED IN THE PLAN. - Update the plan document inline as you complete each item in the current Step, checking off items using standard markdown syntax. - Complete every item in the current Step. - Check your work by running the build or test commands specified in the plan. - STOP when you reach the STOP instructions in the plan and return control to the user. </workflow>