- Home
- Skills
- Desplega Ai
- Ai Toolbox
- Implementing
implementing_skill
- TypeScript
13
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill desplega-ai/ai-toolbox --skill implementing- SKILL.md7.1 KB
Overview
This skill executes an approved technical plan phase by phase with verification checkpoints. It coordinates branch/worktree setup, commit strategy, and automated/manual verification while respecting the user's autonomy level. The goal is reliable, auditable implementation that updates the plan and stops for decisions when required.
How this skill works
The skill reads the full plan, detects completed items, and creates a phased todo list. It asks targeted questions (using the AskUserQuestion mechanism) to gather preferences for file review, branch/worktree choice, and commit strategy, then implements each phase, runs automated checks, and updates the plan. When mismatches or manual verification are required, it pauses and requests user direction according to the selected autonomy mode.
When to use it
- When invoking /implement-plan to execute an existing plan
- When another skill requests this required sub-skill for implementation
- When you want an agent to carry out an approved technical plan with checkpoints
- When you need structured commits and verification per phase
- When you want clear handling of mismatches between plan and codebase
Best practices
- Establish file-review and commit preferences before starting and store them for the session
- Default to Critical autonomy unless explicitly set to Autopilot or Verbose
- Read the entire plan and relevant files fully before making changes
- Commit after each phase for complex plans to keep history clear, or choose single commit for small changes
- Stop and present clear, actionable mismatch summaries when the repo diverges from the plan
Example use cases
- Implement a multi-step feature from a documented plan, committing after each phase and running linters/tests
- Apply a migration plan across repositories with a wts worktree to isolate changes
- Execute fixes across packages: run automated checks, update plan checkboxes, and pause for manual QA
- Carry out a refactor with file-review enabled for inline code feedback during implementation
- Resume work on a partially completed plan by picking up from the first unchecked item
FAQ
Use Critical for cautious progress (default). Choose Autopilot to run without pauses, or Verbose to confirm before each major change.
What happens if the plan and repository disagree?
If not Autopilot, the skill stops and presents a concise issue summary with options to adapt, proceed, or stop; Autopilot will make a documented judgment and continue.