Install
$ agentstack add skill-butterbase-ai-butterbase-skills-journey ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
Butterbase Guided Journey
End-to-end orchestrator. Walks the user from idea to deployed Butterbase app, and (in hackathon mode) on through submission. Each stage writes a markdown artifact under docs/butterbase/ in the user's project. The orchestrator reads docs/butterbase/00-state.md to know the cursor and dispatches the matching journey-* stage skill.
When to use
Invoke automatically when the user signals end-to-end intent: "I want to build…", "let's build an app", "ship this", "help me build a hackathon project". Invoke explicitly when the user runs /butterbase-skills:journey.
If the user wants to do a single stage only (e.g., just design a schema), defer to the matching standalone skill (schema-design) or per-stage command (/butterbase-skills:journey-schema) instead of starting the full journey.
Toolchain
The journey assumes two npm packages alongside the MCP tools:
| Package | Role | When to install | |---|---|---| | @butterbase/sdk | TypeScript client for the deployed app — auth.signIn, db.from(...).select(), storage.upload, realtime subscriptions, function invocation. Works both in the browser (frontend) and in Node (functions, scripts, server-side). | Install in any frontend or any Node service that talks to the deployed app. Auto-added by frontend scaffolds. | | @butterbase/cli | Local-dev CLI — project scaffolding, log tailing, function invocation from the shell, API-key generation, schema diff preview without the dashboard. | Install globally (npm i -g @butterbase/cli) once per dev machine. Used during journey-preflight. |
The journey will prompt for both in plan (which SDK surfaces does the app need?) and preflight (is the CLI installed?). The MCP tools and these packages are complementary: MCP is for agentic / orchestration flows; SDK + CLI are for ordinary application code and the human dev loop.
Procedure
- Detect state. Check whether
docs/butterbase/00-state.mdexists in the working directory.
- If absent: this is a fresh journey. Ask the user
"Is this a hackathon submission? (yes/no)"to sethackathon_mode. Createdocs/butterbase/and write a starter00-state.md(template below). Ifhackathon_mode: false, the starter template'ssubmitrow must be written as- [ ] submit (n/a — not a hackathon)rather than plain- [ ] submit. Then proceed to stageidea. - If present: read the front-matter and the stage checklist. Identify the first unchecked, non-skipped stage. That is the next stage.
- Confirm with user. Print a one-line summary of where we are:
"Resuming journey at for app_id ."Ask:"Continue from ? (yes / jump to other stage / redo previous)".
- Dispatch. Invoke the matching skill via the Skill tool. Do not do the stage's work inline — delegate.
| Stage | Skill | |---|---| | idea | butterbase-skills:journey-idea | | plan | butterbase-skills:journey-plan | | preflight | butterbase-skills:journey-preflight | | docs | butterbase-skills:journey-docs | | schema | butterbase-skills:journey-schema | | rls | butterbase-skills:journey-rls | | auth | butterbase-skills:journey-auth | | storage | butterbase-skills:journey-storage | | functions | butterbase-skills:journey-functions | | ai | butterbase-skills:journey-ai | | rag | butterbase-skills:journey-rag | | realtime | butterbase-skills:journey-realtime | | durable | butterbase-skills:journey-durable | | agents | butterbase-skills:journey-agents | | frontend | butterbase-skills:journey-frontend | | deploy | butterbase-skills:journey-deploy | | substrate | butterbase-skills:journey-substrate (optional) | | submit | butterbase-skills:journey-submit (hackathon_mode only) | | templates | butterbase-skills:journey-templates (optional) |
Docs gate. Stage docs runs once, right after preflight, to prime butterbase_docs for every capability in the plan. Subsequent build stages start with the relevant docs cached at docs/butterbase/03b-docs-cache.md. If the user changes the plan mid-build, re-run /butterbase-skills:journey-docs before the affected stage.
- After the stage skill returns, re-read
00-state.mdand ask the user whether to advance to the next unchecked stage. Stage selection rules:
- Build-stage order is: schema → rls → auth → storage → functions → ai → rag → realtime → durable → agents → frontend.
- If
hackathon_mode: trueand all build stages are done, the next stage isdeploythensubstratethensubmitthentemplates. - If
hackathon_mode: false, the journey ends atdeployunless the user opts intotemplates— skipsubstrateandsubmitentirely (treat them as(n/a)). - The
substratestage is always optional. In hackathon mode, default to skipping it (mark as(n/a — optional, can add post-submission)); in non-hackathon mode, skip it entirely as it is not part of the core journey. - The
templatesstage is always optional. It runs afterdeploy(non-hackathon) or aftersubmit(hackathon). Default is to skip unless01-idea.mdor02-plan.mdflagspublish_as_template: true. - Loop until the cursor reaches
DONE(every stage checked or annotatedn/a).
Starter 00-state.md template
When initialising a fresh journey, write this to docs/butterbase/00-state.md (ask the user "Is this a hackathon submission? (yes/no)" first to set hackathon_mode):
---
app_id: null
api_base: null
hackathon_mode:
hackathon_deadline: null
frontend_stack: null
current_stage: idea
last_updated:
---
# Journey state
## Stages
- [ ] idea
- [ ] plan
- [ ] preflight
- [ ] docs
- [ ] schema
- [ ] rls
- [ ] auth
- [ ] storage
- [ ] functions
- [ ] ai
- [ ] rag
- [ ] realtime
- [ ] durable
- [ ] agents
- [ ] frontend
- [ ] deploy
- [ ] substrate (optional)
- [ ] submit
- [ ] templates (optional)
## Notes
- Journey initialised .
If hackathon_mode: false, write the submit row as - [ ] submit (n/a — not a hackathon) instead of the plain unchecked form, so the orchestrator and any direct stage invocation will skip it.
Outputs
- Creates
docs/butterbase/and00-state.mdon first run. - Updates
current_stageandlast_updatedin00-state.mdwhenever it dispatches.
Anti-patterns
- ❌ Doing a stage's work inline instead of delegating to its skill. The orchestrator only routes.
- ❌ Skipping the "Continue from ?" confirmation — users may want to jump or redo.
- ❌ Letting
00-state.mddrift out of sync. Always re-read after each stage returns.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: butterbase-ai
- Source: butterbase-ai/butterbase-skills
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.