implement-spec

Implements a specification as a single PR using concurrent subagents.

10|3|Updated Jul 18, 2019
One-click install
npx skills add https://github.com/tanqimin/MyFavsORM --skill implement-spec-tanqimin
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: implement-spec
Source: https://github.com/tanqimin/MyFavsORM/tree/main/.agents/skills/implement-spec
Command: npx skills add https://github.com/tanqimin/MyFavsORM --skill implement-spec-tanqimin

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Coordinating the implementation of a large specification across many interdependent tickets is slow and error-prone when done sequentially, and parallel work often causes merge conflicts and duplicated context. ## Core Features & Use Cases - Task Graph Execution: Treats tickets as a dependency graph with blocking relationships, always working on the frontier of ready tickets. - Concurrent Subagents: Runs implementer subagents in parallel, each in its own git worktree and branch, then merges results with a merger subagent. - Context Pointer Communication: Keeps subagent communication sparse by referencing the spec, tickets, research notes, and prior commits instead of duplicating information. - Use Case: Given a spec issue with ten linked tickets, the Skill creates a draft PR, fans out implementer subagents as tickets unblock, runs a code review pass, and delivers one review-ready PR. ## Quick Start Ask the agent to implement the spec from a given issue and its associated tickets as a single pull request.

Frequently Asked Questions about implement-spec

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I implement a spec with multiple tickets in one PR?

Read the spec and tickets to understand the dependency graph, create a branch and draft PR, then run implementer subagents in parallel on the frontier of unblocked tickets. Merge each completed branch into the PR branch and finish with a code review pass.

How do I run coding subagents in parallel without conflicts?

Give each implementer subagent its own git worktree and branch so concurrent work never touches the same files. A separate merger subagent then integrates each finished branch into the shared PR branch.

What is a ticket task graph with a frontier?

Tickets have blocking relationships rather than a linear order, so at any time only a subset of tickets is ready to start. That ready subset is the frontier, and completing tickets can unblock new ones.

How should subagents share context during implementation?

Communicate through context pointers to the spec, tickets, research notes, and previous commits rather than duplicating information. Exploration notes should be saved as markdown files outside the repo so all subagents can read them.

When should I not use parallel subagent implementation?

Avoid it for small specs with one or two trivial tickets, where worktree and merge overhead exceeds the benefit. It is designed for larger specs whose tickets have meaningful parallelism.