By auge2u
Migration system for Gastown. Analyzes projects, extracts Beads, creates Convoys, generates .gt/ structure for multi-agent execution.
Analyze project and generate Gastown semantic memory. Scans codebase, docs, and configs to understand tech stack, constraints, and goals.
Extract work items from project as Gastown Beads. Scans TODOs, issues, PRDs, and backlogs to create actionable work items.
Bundle Beads into Convoys for Gastown assignment. Groups related work items into optimal work packages.
Full Gastown migration - analyze project, extract beads, create convoys. One command to structure any project for Gastown.
Generate a roadmap using native Claude Code orchestration with memory persistence.
Project analyzer and Gastown migration specialist. Scans codebases, extracts work items, and structures projects for Gastown multi-agent execution.
Senior product owner specializing in roadmap scoping from legacy scope, PRDs, and MVP maturity planning.
Native Claude Code orchestrator for autonomous roadmap generation with built-in quality gate validation and loop control.
This skill analyzes existing projects and structures them for [Gastown](https://github.com/steveyegge/gastown) — Steve Yegge's multi-agent workspace manager.
Builds a full-scope product roadmap for an existing project by scanning PRDs in /docs, extracting legacy tasks, and creating a comprehensive plan to move from MVP/early release to the next major stage.
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Give Claude Code a memory before it touches your code.
Lisa is a Claude Code plugin that reads your project before acting on it — extracting tech stack, constraints, prior decisions, and open work into structured files that every future Claude session can instantly use. No more re-explaining what the project is. No more "wait, why was this built this way?" No more lost context.
You know that project you started six months ago and never quite finished? The one with the half-written README, the TODOs in the code, the Slack thread where you worked out the architecture? Open it in Claude Code and run:
/lisa:rescue
Lisa will read your git history like an archaeologist, reconstruct what you were building and why you stopped, understand your tech stack and constraints, generate a fresh roadmap, and create structured work items ready to pick up. In one command, a dormant project becomes a living, organized one — with a memory file that any future Claude session can load in seconds.
Lisa runs a staged pipeline. Each stage produces files that the next stage (and future Claude sessions) can read:
| Stage | Command | What you get |
|---|---|---|
| 0 — Research | /lisa:research | Git archaeology report: original mission, timeline, why it stalled, rescue recommendation |
| 1 — Discover | /lisa:discover | .gt/memory/semantic.json — tech stack, constraints, personas, non-goals |
| 2 — Plan | /lisa:plan | scopecraft/ — vision, phased roadmap, epics + user stories, risks, metrics, open questions |
| 3 — Structure | /lisa:structure | .gt/beads/ + .gt/convoys/ — individual work items and bundled assignments |
| 5 — Reconcile | /lisa:reconcile | Cross-project alignment report — checks multiple repos stay consistent |
Two composite shortcuts:
/lisa:migrate # Stages 1–3: discover + plan + structure (for active projects)
/lisa:rescue # Stages 0–3: full rescue pipeline (for abandoned projects)
/plugin marketplace add auge2u/lisa-helps-ralph-loops
/plugin install lisa@lisa-helps-ralph-loops
That's it. No API keys, no database, no cloud. All outputs are local files.
After /lisa:migrate or /lisa:rescue, your project has:
your-project/
├── .gt/
│ ├── memory/
│ │ └── semantic.json ← everything Claude needs to understand the project
│ ├── beads/
│ │ └── gt-abc12.json ← individual work items with acceptance criteria
│ └── convoys/
│ └── convoy-001.json ← beads grouped for assignment
└── scopecraft/
├── VISION_AND_STAGE_DEFINITION.md
├── ROADMAP.md
├── EPICS_AND_STORIES.md
├── RISKS_AND_DEPENDENCIES.md
├── METRICS_AND_PMF.md
└── OPEN_QUESTIONS.md
The semantic.json is the key file. Point any future Claude session at it and you skip the 20 minutes of re-orientation. It's a structured summary of what the project is, what it's built with, what constraints exist, and what's been decided — all in a format Claude reads natively.
Every stage validates its own output against 31 declarative gates before you move on. If something is missing or wrong, you know before the next agent touches it.
python3 plugins/lisa/hooks/validate.py --stage discover
python3 plugins/lisa/hooks/validate.py --stage all
No PyYAML? It runs in fallback mode (file and JSON checks). Install PyYAML for full pattern validation.
Lisa is the foundation layer of a three-plugin ecosystem:
| Plugin | What it adds | Status |
|---|---|---|
| Lisa (this) | Pipeline, memory, work structure | Alpha v0.3.0 |
| Carlos | Specialist analysis: market fit, tech debt, roadmap depth | Beta v1.2.0 |
| Conductor | Multi-agent orchestration and task tracking | GA v1.0.0 |
Each plugin works completely standalone. Install one, or all three — the ecosystem is additive. Lisa's outputs (.gt/ files and scopecraft/) are the shared language between them.
Claude Code is powerful, but every new session starts cold. You re-explain the project, re-establish context, re-decide things that were already decided. Lisa solves the cold-start problem by giving Claude a structured memory of the project that persists across sessions.
It's especially useful for:
MIT
npx claudepluginhub auge2u/lisa-helps-ralph-loops --plugin lisa-loops-memoryMulti-tenant migration system for Gastown. Features T0/T1/T2 tenant hierarchy, data isolation, tiered features, and customizable branding. Research lost context, discover project facts, plan roadmaps, and structure work items for multi-agent execution.
Product-owner skillset to generate a full-scope roadmap from an existing repo/docs/PRDs and legacy tasks, with an orchestration-friendly loop mode.
Staged migration system for Gastown. Research lost context, discover project facts, plan roadmaps, and structure work items for multi-agent execution.
Project discovery, planning, and quality analysis system with parallel agents
Context-Driven Development plugin that transforms Claude Code into a project management tool with structured workflow: Context → Spec & Plan → Implement
AI Development Environment - Multi-agent orchestration for Claude Code
Flight Levels methodology plugin for Metis. Includes methodology skills, iterative Ralph loops for task execution and initiative decomposition, and Flight Levels agent guidance.
AI-powered cascading development framework with design document system and multi-agent collaboration. Breaks down projects into Features (Mega Plan), Features into Stories (Hybrid Ralph), with auto-generated technical design docs, dependency-driven batch execution, Git Worktree isolation, and support for multiple AI agents (Codex, Amp, Aider, etc.).
Comprehensive UI/UX design plugin for mobile (iOS, Android, React Native) and web applications with design systems, accessibility, and modern patterns