Your notes become agents.
Describe what you want. Syner figures out the rest.
Docs Β· Get Started Β· Philosophy
Syner turns your markdown notes into an operating system for AI agents.
You write notes the way you always have β no schemas, no databases, no enforced structure. Syner reads them for context, routes your intent to focused skills, and delivers concrete output: PRs, reports, documents, decisions.
You supervise. Agents execute.
You (intent)
β
Syner (orchestrator)
β
Skills (31 focused capabilities)
β
Your notes (context)
β
Output (PRs, reports, documents)
| Problem | Syner's answer |
|---|---|
| AI agents don't know your context | Your notes are the context β no setup required |
| One monolithic agent fails at everything | 31 skills that each do one thing well |
| Chat disappears after the session | Agents produce PRs, issues, documents β artifacts that persist |
| You have to babysit every step | Background agents work while you do other things |
| Every tool wants you to change your workflow | Syner reads your existing notes as-is |
| Skill | What it does |
|---|---|
/find-ideas |
Surface ideas buried across your notes |
/find-links |
Connect two domains you've been circling |
/grow-note |
Graduate a thought into a real document |
/track-idea |
See how an idea evolved over time |
/load-all |
Load your full context in one shot |
| Skill | What it does |
|---|---|
/create-syner-app |
Scaffold a new app |
/create-syner-skill |
Create a new skill |
/create-syner-agent |
Define a new agent |
/syner-skill-reviewer |
Audit skill quality before shipping |
/workflow-reviewer |
Review GitHub Actions before running |
| Skill | What it does |
|---|---|
/whats-next |
Figure out what to work on today |
/syner-daily-briefing |
Daily status with opinionated recommendations |
/backlog-hygiene |
Clean stale issues and duplicates |
/syner-gh-auth |
Authenticate GitHub via App tokens |
/syner-fix-symlinks |
Repair broken skill/agent links |
Syner delegates to specialized agents. Each owns a domain.
| Agent | Role | What it owns |
|---|---|---|
| syner | Orchestrator | Routes intent, loads context, coordinates |
| vaults | Context Engineer | Reads notes, synthesizes, returns structured context |
| dev | Ecosystem Builder | Creates skills, agents, apps, workflows |
| bot | Integration Bridge | Slack, GitHub, webhooks β delivery to the outside world |
| design | Design Lead | UI/UX review, accessibility, spatial computing |
Plus 14 lead agents for packages and infrastructure β each with their own specialist teams totaling 70+ domain experts.
Four apps, one ecosystem.
| App | URL | Purpose |
|---|---|---|
| Vaults | syner.md | Browse your thinking β vault dashboard |
| Dev | syner.dev | Docs, skill catalog, specs |
| Bot | syner.bot | Slack & GitHub integrations β execution platform |
| Design | syner.design | Components, tokens, slide generation |
synerops/syner
βββ apps/ # 4 Next.js applications
β βββ vaults/ # syner.md
β βββ dev/ # syner.dev
β βββ bot/ # syner.bot
β βββ design/ # syner.design
βββ packages/ # 8 shared packages
β βββ sdk/ # Core runtime (skill resolution, context loading)
β βββ osprotocol/ # Protocol types & validators
β βββ github/ # GitHub App auth & actions
β βββ slack/ # Slack integration & streaming
β βββ vercel/ # AI SDK tools & sandbox execution
β βββ ops/ # Friction tracking & pattern analysis
β βββ ui/ # Shared components & design tokens
β βββ create-syner-agent/ # Agent scaffolding
βββ skills/ # 31 symlinked skills
βββ agents/ # 19 agent definitions
βββ .syner/ # Your vaults (local, gitignored)
| Layer | Technology |
|---|---|
| Runtime | Bun |
| Framework | Next.js 16 |
| Monorepo | Turborepo |
| AI | AI SDK Β· Claude Code |
| Execution | Vercel Sandbox |
| Docs | Fumadocs |
| Design | shadcn/ui Β· OKLCH tokens Β· Geist |
# Clone and install
git clone https://github.com/synerops/syner.git
cd syner
bun install
# Start development
bun run dev
# Talk to the orchestrator
# (requires Claude Code)
/syner- Bun v1.3+
- Claude Code for agent interactions
- GitHub App credentials for
/syner-gh-auth(setup guide)
Software is built on layers of abstraction. Each generation loses understanding of the layer below but gains power from the layer above. The next layer is intent.
Three principles:
- Skills, not monoliths β Each skill does one thing well. An agent that "does everything" does nothing reliably.
- Markdown as primitive β Notes, configs, outputs β all markdown. Humans read it, machines parse it, LLMs understand it natively.
- Notes as context β Skills read notes for understanding, not extraction. Like a colleague reading your docs before helping.
Syner is built by agents, for agents β supervised by humans.
# Create a skill
/create-syner-skill
# Create an agent
/create-syner-agent
# Review before shipping
/syner-skill-reviewerEvery PR goes through code review. Every skill gets audited. Every agent gets tested.
Built with background agents. Supervised by humans.