re-implement

Re-runs Implement and Check agents after PRD requirement changes in Trellis tasks.

Updated Jun 8, 2026
One-click install
npx skills add https://github.com/SilentFlower/flower-trellis --skill re-implement-silentflower
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: re-implement
Source: https://github.com/SilentFlower/flower-trellis/tree/main/enhancements/old/.agents/skills/re-implement
Command: npx skills add https://github.com/SilentFlower/flower-trellis --skill re-implement-silentflower

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When requirements drift mid-implementation, previously generated code no longer matches the updated PRD. This Skill re-synchronizes code with changed requirements by updating the PRD, re-invoking the Implement Agent, and verifying results with the Check Agent, without restarting the whole task workflow. ## Core Features & Use Cases - Change-aware PRD updates: Appends a structured change log (original requirement, new requirement, reason, impact scope) to prd.md before re-implementation. - Incremental re-implementation: Instructs the Implement Agent to preserve correct existing code and only modify parts affected by the requirement change. - Verification loop: Runs the Check Agent afterward to catch inconsistencies between old and new code, regressions, and lint/typecheck failures. - Use Case: During a Trellis task started with /trellis:start, the product owner changes an acceptance criterion. You update the PRD via this Skill, the Implement Agent adjusts only the affected modules, and the Check Agent confirms nothing regressed. ## Quick Start Ask the AI to run the re-implement workflow for the current Trellis task after describing which requirements in the PRD have changed.

Frequently Asked Questions about re-implement

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

FAQPage Schema
How do I re-implement a task after requirements change in Trellis?▼

Run the re-implement workflow on the active task: it updates prd.md with a change record, then calls the Implement Agent to adjust only affected code, and finally the Check Agent verifies consistency, lint, and typecheck.

What is the difference between trellis start and re-implement?▼

The start command runs the full first-time flow of Research, Context, Implement, and Check. Re-implement is a shorter loop for requirement changes: update the PRD, re-run Implement, then Check, preserving correct existing code.

What prerequisites does re-implement require before running?▼

An active task must exist with .trellis/.current-task set, the task directory must contain an updated prd.md, and implement.jsonl must define the code-spec context. Without an active task, start one first with /trellis:start.

When should I add new context configuration during re-implementation?▼

Add context only when the requirement change introduces new modules or specifications. Use task.py add-context to append paths for implement and check phases; otherwise the existing injected context is sufficient.

Does re-implement rewrite all existing code from scratch?▼

No. The Implement Agent is explicitly instructed to preserve correct existing implementations and modify only the parts affected by the PRD change record, reducing regression risk and keeping unrelated logic intact.