| name | vibejam3-e2e-stack |
| description | Project-local end-to-end autonomy stack tuned to vibejam3's PRD-first, guide-production workflow. Selects the minimum skill stack for autonomous execution. |
| argument-hint | Current vibejam3 task, milestone, and whether the work is planning, coding, or full end-to-end |
| user-invocable | true |
| disable-model-invocation | false |
vibejam3 E2E Stack
Use this skill when vibejam3 needs autonomous end-to-end execution that stays aligned with the repo's PRD-first workflow.
This is a project-specialized version of shared-e2e-autonomy-stack. It routes through vibejam3-specific skills instead of generic ones.
Non-bypassable no-stub rule
- Do not write, recommend, preserve, or disguise stubs, placeholder behavior, fake-success paths, mock-complete flows, TODO-backed user paths, or dead UI affordances.
- If real implementation is blocked, stop at the last truthful state, state the blocker plainly, and leave no fake completion behind.
What this skill does
- Chooses the minimum stack for autonomous vibejam3 execution.
- Routes through project-local skills instead of generic shared ones.
- Prevents skill overload while preserving proof and recoverability.
Phase routing for vibejam3
Phase 0 (environment setup)
vibejam3-phase0-environment-inspector
vibejam3-prd-optimizer
Preimplementation
vibejam3-prd-gap-closure
vibejam3-slice-selector
vibejam3-readiness-gate
vibejam3-prd-accelerator
Implementation
e2e-coding
result-driven-tdd
hardening-optimization
Design (when screens or flows need direction)
vibejam3-app-design
Completion and handoff
vibejam3-completion-gate
vibejam3-handoff-pack
Release
vibejam3-quality-coach
vibejam3-release-readiness
Always-on meta skills
Keep these active for autonomous work:
general-autonomy
consistency
tool-usage
Add when the task shape requires:
autonomous-planning for multi-step work
long-task-support for long-running work
traceability for meaningful changes
Selection rules for vibejam3
- Prefer the smallest stack that owns the current milestone phase.
- Do not start coding if the PRD chain is not complete.
- Do not call the project end-to-end complete until validation and hardening passed.
- Do not end a long task without either a completion verdict or a handoff pack.
Hard bans
- No skill pileups without a phase reason.
- No autonomous flow that skips PRD chain on multi-step work.
- No coding path without proof.
- No hardening skipped on guide-production code.
Output required
Return all of these:
- current vibejam3 milestone
- active autonomy stack
- worker skills selected for the current phase
- why this is the minimum sufficient stack
- current phase goal
- next proof step
- what condition advances to the next phase
Completion standard
This skill is complete only when the task has a clear autonomous stack that is small enough to stay sharp and strong enough to carry the work end to end without guesswork.