Skip to content

uniffy-io/uniffy / .agents/skills/execute

execute

Verified

Execute or resume an implementation plan from `.agents/local/plans/`, keeping its companion backlog and validation evidence current. Use when the user asks to implement a saved plan or continue planned multi-phase work.

Execute: Implement from Plan

Use the plan path supplied with the request. If no path is supplied, proceed only when one plan is the unambiguous subject of the conversation; otherwise ask which plan to execute.

1. Read and understand

  • Read the ENTIRE plan. Read the always-on rules and every path-scoped rule in .agents/rules/ that matches the files in scope; the plan may name additional mandatory rule files.
  • Check .agents/local/plans/backlogs/ for an existing backlog. If one exists, resume from the first [ ] or [~] item instead of starting over.
  • If the plan is multi-phase and no backlog exists, create one (convention in AGENTS.md "Plans, Backlogs and Reviews").
  • Validate the plan against the current codebase before writing code: file paths, symbol names, and patterns it references may have drifted since it was written. Fix the plan's assumptions, not the codebase, when they disagree on facts.

2. Execute tasks in order

For each task:

  • Read the referenced files before modifying them.
  • Follow the plan's specification; stay consistent with the surrounding code and the auto-loaded rules.
  • Verify as you go: correct imports, correct types, no drive-by refactors outside the task.

If something in the plan is wrong or missing, document the deviation in the backlog and continue; do not silently diverge.

3. Tests

Create the test files the plan specifies, following the existing test organization (unit tests for operations, integration tests for handlers). Cover the edge cases the plan lists.

4. Validate

./manage.py proto            # if .proto files changed
./manage.py lint -s backend  # if Python changed
./manage.py lint -s ui       # if frontend changed
./manage.py test             # backend tests
./manage.py test -s ui       # frontend tests

Fix and re-run until green. Do not skip validation steps.

5. Update the backlog

  • Mark completed tasks [x], in-progress [~] with a note, blocked [!] with the reason.
  • Add a dated session note (absolute date) summarizing what changed.
  • Commit the backlog update alongside the work.

6. Report

Summarize: tasks completed, files created/modified, tests added with results, validation output, deviations from the plan. State plainly whether the work is ready for the commit skill.

reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 469kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 321kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 321kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 321kUpdated Claude CodeCodexVerifiedInstall