implement-dev

Executes an approved implementation plan, evidences acceptance criteria, and writes a completion report.

Updated Apr 28, 2026
One-click install
npx skills add https://github.com/byron1st/personal-harness --skill implement-dev-byron1st
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: implement-dev
Source: https://github.com/byron1st/personal-harness/tree/main/skills/implement-dev
Command: npx skills add https://github.com/byron1st/personal-harness --skill implement-dev-byron1st

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Turning an approved plan into working code often loses track of which TODOs shipped, which acceptance criteria were proven, and where the implementation diverged from the plan. This Skill executes a saved plan file end to end and produces an auditable implementation report under docs/agents/dev. ## Core Features & Use Cases - Plan-driven execution: Works the plan's ## TODOs in order, flipping each checkbox as it completes so the run stays resumable, and reads linked research before each TODO. - Deviation handling: Resolves detail-level choices itself, escalates direction-level conflicts as blocked, and routes design-bearing forks to a plan consultant via needs-design-decision. - Evidence-backed reporting: Verifies with lint/format/test/build commands, evidences every Acceptance Contract row, and writes a structured report with TODO fulfillment, red flags, open questions, and plan divergence. - Use Case: After approving a plan from plan-dev, hand its path to this Skill to implement the TODOs, run the test suite, and receive a report linking each change to its acceptance criterion. ## Quick Start Use the implement-dev skill to execute the plan at docs/agents/20260622153045_PROJ-42_PLAN_add-login.md and produce the implementation report.

Frequently Asked Questions about implement-dev

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

FAQPage Schema
How do I implement a saved plan file with an AI agent?▼

Pass the plan file path to the implement-dev skill, which works the plan's TODOs in order, checks off each completed item, runs verification commands, and writes a completion report under docs/agents/dev. It can run standalone or inside a dev-loop.

What happens when implementation conflicts with the plan?▼

Detail-level deviations like helper names or edge cases are resolved by the executor and recorded on the TODO's deviation line. Direction-level conflicts that contradict the plan's goals or key decisions stop execution and return a blocked status with the decision the user must make.

Can implement-dev run without a dev-loop orchestrator?▼

Yes, it runs standalone where the current session is the executor and asks the user directly when a decision is needed. In loop mode it cannot ask the user and instead returns a stage status for the loop to handle gates and retries.

How are acceptance criteria verified during implementation?▼

Every row in the plan's Acceptance Contract must be evidenced with the proof its Evidence column names, recorded on the report's AC lines and in the executor return. An unevidenced acceptance criterion blocks a pass status.

When does implement-dev stop retrying failed verification?▼

It stops after three failed attempts on the same error and returns failed in loop mode or asks the user standalone. The skill never retries under a different model.