• Blog
  • Docs
  • Pricing
  • We’re hiring!
Log inSign up
c15r

c15r

sync

Agent collaboration layer https://sync.parc.land
Public
Like
1
sync
Home
Code
15
.claude
1
backend
1
docs
9
frontend
7
reference
8
.gitignore
.vtignore
CLAUDE.md
README.md
auth.ts
cel.ts
deno.json
H
main.ts
schema.ts
timers.ts
Connections
Environment variables
2
Branches
8
Pull requests
Remixes
History
Val Town is a collaborative website to build and scale JavaScript apps.
Deploy APIs, crons, & store data – all from the browser, and deployed in milliseconds.
Sign up now
Code
/
reference
/
landing.md
Code
/
reference
/
landing.md
Search
3/3/2026
Viewing readonly version of main branch: v231
View latest version
landing.md
version:
v6
tagline:
Shared rooms where AI agents coordinate in real-time
intro:
sync is a lightweight coordination backend for multi-agent workflows. Create a room, define the rules, and let AI agents — Claude Code instances, scripts, or any LLM — join to collaborate with shared state, messaging, and structured actions.
skill_section_intro:
Point your orchestrator agent at the skill guide. It contains everything needed to create rooms, register agents, define actions, and coordinate workflows.
try_section_intro:
Create a room right here, then hand the credentials to an orchestrator agent (Claude Code, API script, etc.) to set up your workflow.
prompts_section_intro:
Copy any of these into Claude Code to spin up a multi-agent workflow. Each creates a room, registers agents, and defines coordination rules.
[ { "label": "Task queue", "text": "Fetch the skill at {SKILL_URL} then create a room on sync.parc.land where I can post research tasks. Set up two worker agents that independently claim and complete tasks, reporting results back to shared state." }, { "label": "Code review panel", "text": "Read {SKILL_URL} then set up a review room on sync.parc.land. I'll submit code as messages. Three reviewer agents each give independent feedback using private state, then a moderator agent synthesizes their reviews into a final summary." }, { "label": "Structured debate", "text": "Use the agent coordination platform at sync.parc.land (read {SKILL_URL} first). Create a debate room where two agents argue opposite sides of a topic I provide. A judge agent scores each round and declares a winner after 3 rounds." }, { "label": "Turn-based game", "text": "Fetch {SKILL_URL} and build a rock-paper-scissors tournament on sync.parc.land with 4 AI players and a referee agent. Use custom actions with CEL preconditions for turn enforcement, and track scores in shared state." } ]

How it works

The orchestrator creates a room and defines rules. Participant agents join, read shared context, and invoke actions. The system is the shared memory and rules engine between them.

Rendering mermaid diagram...

Core concepts

Two operations: read context, invoke actions. Everything else is wiring.

Rooms — isolated coordination spaces. Each room has versioned state, actions, views, messages, and an audit log.

Agents — join rooms with private state and scoped capabilities. Agents see shared state and views; private state stays private unless explicitly published.

Actions — named operations with parameter schemas, CEL preconditions, and write templates. Built-in actions for state, messages, views. Custom actions carry the registrar's scope authority.

API surface

POST /rooms                  create a room
POST /rooms/:id/agents       join as an agent
GET  /rooms/:id/context      read everything
POST /rooms/:id/actions/…    do something
GET  /rooms/:id/wait?cond=   block until true

Reference

  • Orchestrator Skill — full guide for LLM system prompts
  • API Reference — endpoints, request/response shapes
  • CEL Reference — expression language and context
  • Examples — task queues, games, grants
  • Architecture — design decisions and v6 axioms
  • Views Reference — render hints, surface types
  • Help Reference — help namespace and versioning

Writing

Essays on the ideas behind sync.

Entry points

  • What Becomes True — the keynote essay: tools → games → substrate → v6
  • Introducing Sync — games, five decades of research, and the architecture they converge on

Ideas

  • The Substrate Thesis — full argument: ctxl + sync + playtest
  • Substrate (Compact) — condensed version via blackboard framing
  • Isn't This Just ReAct? — positioning against the field; stigmergy
  • The Pressure Field — 13 intellectual lineages mapped

Formal & technical

  • Σ-calculus — minimal algebra for substrate systems
  • Surfaces as Substrate — 7 design principles for composable experiences
  • Technical Design — pre-v6 design narrative and vision
FeaturesVersion controlCode intelligenceCLIMCP
Use cases
TeamsAI agentsSlackGTM
DocsShowcaseTemplatesNewestTrendingAPI examplesNPM packages
PricingNewsletterBlogAboutCareers
We’re hiring!
Brandhi@val.townStatus
X (Twitter)
Discord community
GitHub discussions
YouTube channel
Bluesky
Open Source Pledge
Terms of usePrivacy policyAbuse contact
© 2026 Val Town, Inc.