Implementation
Skill tranviet0710/agent-skills/.github/skills/implementation
Implement a feature based on a plan file. Use when: implementing a plan, execute a plan, build from plan, start working on plan, implement PLAN.md. Reads the plan, asks clarifying questions, creates TASKS.md with subtasks in the same folder, then implements each task and marks it done.From its SKILL.md
npx -y skills add tranviet0710/agent-skills --skill implementationAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
3.1 KB, 708 tokens by cl100k_base, as published. Nobody here has run it
Implement From Plan
When to Use
Invoke this skill when the user says:
- "implement based on the plan"
- "execute the plan"
- "build from PLAN.md"
- "start working on [feature] plan"
- "implement [folder]/PLAN.md"
Procedure
Step 1 — Locate & Review the Plan
- Find the
PLAN.md(or any*PLAN*.md) in the referenced folder or the currently open file. - Read the entire plan carefully.
- Note: the plan's folder is the working directory for this skill — all output files go there.
Step 2 — Ask Clarifying Questions
Before doing anything else:
- List every ambiguity, assumption, or missing detail you found while reading the plan.
- Present them as a numbered question list to the user.
- If the plan is fully clear with zero ambiguities, explicitly state: "The plan is clear — no questions. Proceeding to create TASKS.md."
- Wait for the user's answers before continuing.
Step 3 — Create TASKS.md
Create a TASKS.md file in the same folder as the plan file, structured as follows:
# Tasks
> Auto-generated from [PLAN.md](./PLAN.md) on <date>
## Task List
- [ ] **Task 1 — <short title>**
- Description: <what needs to be done and why>
- Subtasks:
- [ ] <subtask a>
- [ ] <subtask b>
- [ ] **Task 2 — <short title>**
- Description: ...
- Subtasks:
- [ ] ...
...
Rules for task decomposition:
- Each task maps to one logical, independently testable unit of work.
- Subtasks are concrete, atomic steps (a single file edit, a single function, etc.).
- Include enough description so anyone can pick up the task cold.
- Do not include tasks that are already done in the plan.
Step 4 — Implement Tasks One by One
For each task in TASKS.md:
- Mark the task as in-progress by updating the checkbox to
[-](or add_(in progress)_inline) inTASKS.md. - Implement the task — write code, create files, make edits as needed.
- Validate the change compiles / passes linting if applicable.
- Mark the task as done by changing
[ ]→[x]for the task and all its completed subtasks inTASKS.md. - Move to the next task.
Step 5 — Final Summary
After all tasks are marked [x]:
- Add a
## Summarysection at the bottom ofTASKS.mdlisting what was built. - Report to the user that implementation is complete.
Important Rules
- Never skip Step 2. Always surface questions before writing code.
- Always write
TASKS.mdbefore writing any implementation code. - Keep
TASKS.mdup to date in real time — update it before and after each task. - If a new task is discovered mid-implementation, add it to
TASKS.mdbefore starting it. - Do not mark a task
[x]unless it is truly complete (code written, no errors).
What ships with it: 1 file
2.4 KB alongside SKILL.md
- README.md2.4 KB