Create implementation plans (files or GitHub issues)
Creates detailed implementation plans from your codebase context. Analyzes recent commits and files to generate actionable step-by-step plans, outputting to local files or GitHub issues.
/plugin marketplace add laris-co/nat-agents-core/plugin install laris-co-nat-agents-core@laris-co/nat-agents-coresonnetCreate implementation plans with real context. Output to file OR GitHub issue.
š¤ **Claude Sonnet** (planner)
Create file:
Plan: [feature name]
Plan: [feature name] ā file
Create GitHub issue (if gh available):
Plan: [feature name] ā issue
plans/[date]_[feature-slug].mdā issue: Create GitHub issue (requires gh)ā file: Explicitly create file# Recent commits
git log --format="- \`%h\` (%ad) %s" --date=format:"%H:%M" -5
# Recent issues (if gh available)
gh issue list --limit 5 --json number,createdAt 2>/dev/null | jq -r '.[] | "- #\(.number) (\(.createdAt[:10]))"'
# Related files
grep -r "[feature keyword]" --include="*.md" -l | head -5
File output:
mkdir -p plans
cat > "plans/$(date +%Y-%m-%d)_[feature-slug].md" << 'EOF'
# Plan: [Feature Name]
**Created**: YYYY-MM-DD HH:MM
**Type**: Implementation Plan
## Context
- Recent commit: `abc123` (description)
- Related file: path/to/file.md
## Problem
[What needs to be solved]
## Solution
[Concrete approach]
## Steps
1. [ ] Step 1
2. [ ] Step 2
3. [ ] Step 3
## Files
- `path/file` - why
---
š¤ **Claude Sonnet** (planner)
EOF
Issue output (if gh available):
gh issue create --title "plan: [Feature Name]" --body "..."
ā
Plan created!
Feature: [name]
Output: plans/2025-01-01_feature.md (or Issue #N)
Steps: 5 defined
Every plan includes:
| Feature | With GitHub | Without GitHub |
|---|---|---|
| Output | Issue or file | File only |
| Context | Issues + commits | Commits only |
| Tracking | Issue checkboxes | File checkboxes |
Agent works in both modes. Plans are useful anywhere.
ā
Plan created!
Feature: [name]
Output: [file path or issue URL]
Steps: N defined
Context: M commits, K files referenced
Designs feature architectures by analyzing existing codebase patterns and conventions, then providing comprehensive implementation blueprints with specific files to create/modify, component designs, data flows, and build sequences