- Home
- Skills
- Eljun
- Workflow Skills
- Implement
implement_skill
0
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 eljun/workflow-skills --skill implement- SKILL.md22.2 KB
Overview
This skill implements tasks defined in docs/task/*.md by reading the task document, following the implementation steps, and updating TASKS.md status. It supports single-task, auto-chained pipelines (test → document → ship), and multi-task parallel execution with safety checks and permission prompts.
How this skill works
You invoke the skill with a task ID or filename and it resolves the task document from TASKS.md or docs/task/*.md, reads the implementation steps, and performs only the file edits and commits required. In auto mode it marks Automation: auto in the task doc and automatically invokes the testing pipeline after implementation. In multi-task mode it runs pre-flight checks, requests approval, then spawns parallel background agents with limited, safe tools.
When to use it
- When a task document exists in docs/task/ and the task is Planned
- When you are ready to start coding a specific task and want traceable commits
- To run a full pipeline from implement through test, document, and ship (use auto)
- To implement several independent tasks in parallel (use -m or --multi)
- When you need enforced pre-flight checks for risky operations before parallel runs
Best practices
- Always rely on the task document as the primary source of truth; read only the referenced files
- Invoke required specialized skills (Vercel or Supabase best-practices) before writing any code
- Use the [task-{ID}] commit prefix for every commit for traceability
- Run pre-flight checks and approve permissions before spawning background agents
- Prefer sequential runs if tasks share file changes or have dependencies to avoid conflicts
Example use cases
- Implement task #1: /implement 1 to start coding and update TASKS.md to In Progress
- Run a full pipeline: /implement auto 001-auth-jwt to implement, then automatically test and ship
- Parallel work: /implement -m 1 2 3 to spawn three agents after approving the pre-flight checklist
- Set automation for a task: run /implement auto 2 to add Automation: auto to the task doc and chain
- Safely handle sensitive operations: pre-flight scan flags migrations, deletions, or secrets for manual approval
FAQ
Auto mode updates the task document to Automation: auto (if invoked) and, after implementation, automatically starts the test → document → ship pipeline.
How does multi-task mode handle conflicts?
Multi-task mode runs file-overlap and dependency checks and warns you. You must approve running in parallel; if conflicts are likely, run tasks sequentially instead.
What specialized skills must be invoked?
If installed and relevant, /vercel-react-best-practices for React/Next.js work and /supabase-postgres-best-practices for database work must be invoked before writing code.