implementation-plan

Produce a decision-complete, repository-grounded implementation plan with scope, architecture, file-level changes, data/API impacts, tests, migration, rollout, rollback, risks, and acceptance criteria

SKILL.md

GitHub

Implementation Plan

Discover

Read project instructions, relevant tree/files, manifests, tests, configs, data contracts, recent history, and dirty-worktree state. Trace the current behavior and identify consumers. Verify external/library behavior from installed or current official sources when it can drift.

Ask only for a genuinely blocking choice that cannot be discovered or safely inferred. Otherwise state assumptions and continue.

Plan content

# Goal and measurable acceptance criteria
## Current behavior and evidence
## Scope in / out
## Assumptions and constraints
## Proposed design and alternatives rejected
## File-by-file changes
## API, schema, data, config, dependency, and security impacts
## Implementation sequence and dependencies
## Test and validation matrix
## Migration, compatibility, rollout, observability, and rollback
## Risks, edge cases, and unresolved decisions

Every step must identify the outcome, likely files/symbols, prerequisites, behavior preserved, validation, and rollback. Separate read-only diagnosis, local edits, and external/deployment actions; planning does not authorize publishing or production changes.

Quality gate

  • paths and current behavior were verified;
  • acceptance criteria are observable;
  • data/API compatibility and failure modes are covered;
  • security/privacy/authorization boundaries are explicit;
  • tests include normal, boundary, regression, and failure cases;
  • migration is reversible and operational signals are defined;
  • no placeholder "decide later" remains on the critical path.

Invoke through natural language or the active host's documented skill mechanism. Do not assume an external command file or slash command exists.