A continuity system for LLM–harness collaboration. Builds technical specifications through structured persona conversations — and carries session state across compactions, sessions, and time.
claude --plugin-dir ~/workspace/storytime
Describe a problem. Storytime surveys your codebase, assembles a team of domain-expert personas, and runs a structured conversation that produces a plan — grounded in your actual code, with citations, decisions, and visual aids. Underneath the spec workflow is a consolidation loop that preserves continuity across compactions and sessions.
Phases collapse when empty. Not every run uses every gear. See a full walkthrough →
/storytime "our public API has no rate limiting and we're getting abuse"
Storytime surveys your code, finds the Express middleware chain has no
throttling layer. Assembles @owner [anchor], @systems [lattice],
@critic [forge], @operator [tide]. They investigate, and produce:
specs/.storytime/sessions/rate-limiting/001/
├── survey.md Codebase context + fingerprint
├── team.md Persona definitions
├── icebreaker.md Status quo discussion
├── breakout-algo.md Token bucket vs sliding window
├── breakout-store.md Redis vs in-memory for counters
└── plan.md ASCII slide deck + implementation steps/storytime-remember stages
remembrance before /compact.anchor, lattice,
kestrel). Default core: OWNER, OPERATOR, CRITIC ×2. One driver per
leg — supporters stay silent unless useful AND non-distortive.
@role is a lens directive, not just a skill trigger./compact and loaded as the first action post-compact. Carries session
state across context boundaries. Written at pauses and pre-compact.Callout-> / Callout<- sigil lines —
forward authoritative, reverse is lint-cached._thread.md is the continuity ledger —
decisions, consolidation log, open questions. Park mid-session and resume
with full context.| Skill | Description |
|---|---|
| /storytime | Full workflow: survey, team, icebreaker, breakouts, plan |
| /storytime-survey | Standalone codebase survey with artifact inventory |
| /storytime-breakout | Focused investigation with one driver + silent supporters |
| /storytime-converge | Merge breakout results into a unified plan |
| /storytime-buildout | Implement the plan with persona-driven pair programming |
| /storytime-retro | Retrospective: plan vs what was built + evaluation scorecard |
| Skill | Description |
|---|---|
| /storytime-cohort | Hire, fire, bench, promote, evolve personas |
| /storytime-qa | Formal queries to personas about past decisions (context-loaded) |
| /storytime-echo | One-shot spawning-pool voice test (ephemeral, no state) |
| /storytime-pr-qa | Team handles PR review comments |
| Skill | Description |
|---|---|
| /storytime-remember | Stage remembrance for /compact (wakeup document + prompt) |
| /storytime-lint | Mechanical + reasoning-tier validation against process rules |
| /storytime-bootstrap | Initialize .storytime/ in a repo |
| /storytime-consolidate | Organize, archive, roll up documents |
| /storytime-absorb | Team reads and interprets existing docs |
| /storytime-export | Convert to ADRs, issues, Kiro specs |
| /storytime-status | Dashboard: cohort, threads, decisions, tutorial progress |
| /storytime-undo | Revert: phase, episode, thread, or file |
"We need rate limiting on the API"
"Should we split this service?"
"Webhooks are silently failing"
"The auth middleware needs a rewrite"
"How did this system get here?"
"Did the plan match what we built?"