Execute Work Plan

VerifiedCaution

Execute a work plan by making changes, running validation, and committing. Follows detailed instructions and handles feedback.

Sby Skills Guide Bot
DevelopmentAdvanced
7/24/2026
Claude Code
#plan-execution#git-commit#ci-validation#autonomous-agent#code-changes

Recommended for

Automated summary

Executes an already-written work plan in a target repository: applies the changes step by step, validates with `make ci`, and commits each step.

Strengths

  • Consumes `o/do/feedback.md` first (review comments or notes from a previously timed-out attempt) before continuing the plan
  • Granular per-step commits with explicit file staging, plus an explicit ban on destructive git commands
  • Defers to the repo's `AGENTS.md` for conventions and build targets, and checks `git status` before every staging
  • Reads the initial CI output (`o/build/log.txt` and its exit code) before touching any code

Limitations

  • Assumes a specific orchestration layout (`o/repo/`, `o/plan/plan.md`, `o/do/`) that a normal checkout does not have
  • Does no planning or design: with no pre-existing plan there is nothing for it to execute
  • The fixed 300s CI timeout and assumed `make` targets may not fit every repository
When to use it

Use it when a plan already exists and needs to be turned into committed, CI-validated code inside the sandbox.

When not to use it

Don't use it to explore a problem or design a solution — that belongs to the planning phase, not execution.

Security analysis

Caution
Quality score85/100

This is a legitimate coding-agent workflow and it explicitly forbids destructive git commands. However, it executes repo build scripts and trusts repo-provided instructions, so it involves meaningful execution risk even though there is no exfiltration, obfuscation, or clearly malicious behavior.

Findings
  • •Runs repository-controlled build/test/lint targets such as `make ci`, which can execute arbitrary code from the target repo.
  • •Follows repository-provided instructions in `AGENTS.md`, the plan, feedback, and work-item JSON, creating a prompt-injection surface if those inputs are untrusted.
  • •Performs filesystem edits and git commits in the sandbox; legitimate for the task but still a powerful operation.
How this analysis is made

Examples

Implement a planned feature
Execute the work plan in o/plan/plan.md against o/repo/. Work through each step, commit per step with descriptive messages, then run make ci and fix any failures.
Address PR review feedback
The work item is a "pr" type: read o/do/feedback.md first, fix the review comments listed there, then continue with any remaining plan steps and re-run make ci.
Resume a timed-out attempt
A previous do attempt timed out. Read the timeout notes and uncommitted changes recorded in o/do/feedback.md, resolve those issues first, then finish the remaining plan steps and amend the commits.

name: do description: Execute a work plan. Make changes, run validation, commit.

Do

You are executing a work item. Follow the plan.

Environment

  • The target repository is at o/repo/. Make all changes there.
  • The feature branch is already checked out.
  • The work item JSON follows this prompt after ---.
  • If type is "pr", you are addressing review feedback on an existing PR.
  • If type is "issue", you are implementing new work.
  • Build dependencies are pre-fetched under o/repo/o/. You can run make ci (or individual targets like make test, make check-types, make lint) inside the sandbox. Read o/build/log.txt for the initial CI output and o/build/log.txt.exit for the exit code.

Setup

Read o/repo/AGENTS.md if it exists. It contains repo-specific context, conventions, and build instructions for the target repository. Follow its guidance.

Read o/plan/plan.md for the full plan.

Read o/do/feedback.md — if non-empty, it contains notes from a previous attempt. This may be review feedback from a check phase, or timeout notes from a timed-out do attempt (with errors encountered, key observations, and uncommitted changes). Address those issues first, then continue with any remaining plan steps.

Instructions

  1. Read the plan and every file you intend to modify before editing.
  2. If feedback.md is non-empty, fix those issues first.
  3. Maintain a running list of files you modify.
  4. For each remaining step in the plan: a. Make the changes for that step. b. Before staging, run git -C o/repo status and verify only your files are affected. c. Stage the specific files changed (not git add -A). d. Commit with a descriptive message for that step.
  5. Run format and lint checks from o/repo/AGENTS.md (e.g. make check-format, make format). Fix any issues and amend the last commit.
  6. Run cd o/repo && make ci to validate all changes. Use a 300s bash timeout ("timeout": 300000) — some repos take over 2 minutes for a full CI run. Fix any failures and amend.
  7. If validation requires fixes, stage and commit them.

Forbidden

Do not use destructive git commands: git reset --hard, git checkout ., git clean -fd, git stash, git commit --no-verify.

Output

Write o/do/do.md:

# Do: <title>

## Changes
<list of files changed>

## Commit
<SHA or "none">

## Status
<success|partial|failed>

## Notes
<issues encountered>

Follow the plan. Do not add unrequested changes.

From the file to a first result

The download contains exactly the content displayed on this page. This copy may differ from the latest version in the source repository.

View the source and supporting files

This download contains SKILL.md only. Any scripts, references, templates and dependencies are not included. Get the required files from the source, preserving their directory structure, before continuing.

1. Download the file

Download SKILL.md

From the root of a test project, save the file as .skill-downloads/execute-work-plan-mryiqhx7/SKILL.md, or use the command below. This folder is for inspecting the skill before installation.

Command for macOS/Linux or WSL: requires a POSIX shell and curl. On Windows without WSL, use the download button and create the folder manually.

mkdir -p '.skill-downloads/execute-work-plan-mryiqhx7' && curl --fail --show-error --location 'https://skills-guides.com/api/skills/execute-work-plan-mryiqhx7/download' --output '.skill-downloads/execute-work-plan-mryiqhx7/SKILL.md.part' && mv '.skill-downloads/execute-work-plan-mryiqhx7/SKILL.md.part' '.skill-downloads/execute-work-plan-mryiqhx7/SKILL.md'

2. Check prerequisites

Read SKILL.md before using it. If it references scripts or external services, inspect them and prepare the required access in your test environment.

Original excerpt: Setup

Read o/repo/AGENTS.md if it exists. It contains repo-specific context, conventions, and build instructions for the target repository. Follow its guidance.

Read o/plan/plan.md for the full plan.

Read o/do/feedback.md — if non-empty, it contains notes from a previous attempt. This may be review feedback from a check phase, or timeout notes from a timed-out do attempt (with errors encountered, key observations, and uncommitted changes). Address those issues first, then continue with any remaining plan steps.

The platforms on this page are catalog indications. The paths below follow tool documentation; they do not prove this skill has been run in those tools. No minimum tool version is confirmed for this listing.

3. Load the skill

After inspection, copy the complete folder into your project using one of the commands below, then open a new tool session. The command stops if a folder with the same name already exists: compare the files before replacing it.

Claude Code
test -f '.skill-downloads/execute-work-plan-mryiqhx7/SKILL.md' && test -s '.skill-downloads/execute-work-plan-mryiqhx7/SKILL.md' && mkdir -p '.claude/skills' && test ! -e '.claude/skills/do' && cp -R '.skill-downloads/execute-work-plan-mryiqhx7' '.claude/skills/do'

OpenSkills

OpenSkills must already be installed. The local install/read procedure is checked with OpenSkills 1.5.0, which requires Node.js ≥ 20.6.0. Catalog slugs are not resolved automatically.

Official OpenSkills installation
openskills install './.skill-downloads/execute-work-plan-mryiqhx7'

openskills read 'execute-work-plan-mryiqhx7'

Reading should display the installed file content and its base directory. It does not run the skill or install its dependencies.

For an agent that reads AGENTS.md, then sync the index. This command changes AGENTS.md: review the diff before using it.

openskills sync

Manual method or another tool: attach SKILL.md to a conversation, or open it in an agent that can read files. Include any required supporting resources. No universal folder is recognized by every tool.

4. Try a first task

Use a short, synthetic input: an API endpoint, a small document or an interface brief. In a new conversation, attach the file or give the agent its path, then adapt this request:

Read .skill-downloads/execute-work-plan-mryiqhx7/SKILL.md and apply its instructions to this example: [describe a test input]. Identify any missing files, dependencies or access. Produce the result requested by the skill without inventing data or running unreviewed scripts.

Expected result: output matching the skill instructions, or a precise list of missing prerequisites. A read confirmation alone does not validate a first use. If an external resource is unavailable, that path remains untested.

Related skills