Open infrastructure
Keep growing the library
OpenChat is designed to expand forever — data sources, connectors, APIs, agents, trading bots — in a fixed shape so the library stays searchable and composable.
78
Total
70
Runnable
8
Planned
23
Featured
Four layers (+ pipelines)
- data_sourceData sources → lib/openchat/data-sources/
- connectorConnectors → lib/openchat/connectors/
- agentAgents → lib/openchat/templates/
- trading_botTrading bots → lib/openchat/templates/
- pipelinePipelines → lib/openchat/pipelines/
How to add something
# Scaffold (creates stub + registry entry + changelog line) node scripts/scaffold-plugin.mjs \ --kind connector \ --slug my-api \ --title "My API" \ --category connectors # kinds: data_source | connector | agent | trading_bot | pipeline # Validate registry integrity pnpm library:validate
- Pick kind: data_source | connector | agent | trading_bot | pipeline
- Pick category id from CATEGORIES (not free-text)
- slug: kebab-case, unique across whole registry
- Add open source file under apps/web/src/lib/openchat/{dir}/
- Register manifest in registry/{kind}s.ts (or scaffold CLI)
- Tag with 3–6 searchable tags + apis[]
- maturity: experimental → beta → stable
- If agent: systemAddon + dependsOn connectors
- If data_source: sectors[] + worker connector path
- Document envVars; never commit secrets
- Featured only for flagship / first use-cases
Categories
Data · Finance
Markets, macro, crypto, payments
data-finance
Data · Climate & Earth
Weather, EO, hazards, energy
data-climate
Data · Society & Culture
Trends, books, arts, sports, gov
data-society
Data · Realtime
Live feeds and snapshots
data-realtime
Research
Product, market, competitive agents
research
Content & Social
Posts, video, demos, SEO
content
Documents
Long-form writing agents
documents
Trading
Bots, briefs, market co-pilots
trading
Data & Insights
Lake-grounded analysis agents
data
Media & Video
Voice, video gen, browser record
media
ML · Image
Text/image generation, upscale, rembg (Replicate + HF)
ml-image
ML · Video
T2V, I2V, lip-sync, restyle
ml-video
ML · Audio
TTS, ASR, music, audio classify
ml-audio
ML · Language / NLP
LLM, classify, translate, QA, embeddings, tables
ml-language
ML · Vision
Caption, detect, segment, VQA, document QA
ml-vision
ML · 3D
Image-to-3D and related models
ml-3d
Connectors
API-to-API plugins
connectors
Dev tools
Scaffolding, infra, publish
devtools
Maturity ladder
planned → experimental → beta → stable
Planned items appear on the roadmap tab in /library so growth stays visible without polluting the main browse.
Roadmap (8 planned)
- data_sourceFRED Macro SeriesSt. Louis Fed economic time series (planned deeper package).
- data_sourceSEC EDGAR FilingsCompany filings index for research agents (planned).
- data_sourceNASA FIRMS FiresActive fire detections for risk maps (expand coverage).
- connectorYouTube Data APIUpload/schedule long-form demo cuts (planned).
- connectorSlack Incoming WebhooksPush research briefs to team channels (planned).
- connectorNotion PagesWrite PRDs/memos into Notion (planned).
- agentWeekly Multi-Sector DigestAuto brief across finance, climate, trends (planned).
- agentCompetitor WatchTrack named competitors via trends + markets (planned).
Registry health
✓ Registry validates
Data sources growth loop
- Add worker connector under workers/src/connectors/
- Historics / realtime write into data/lake/
- Seed catalog product if needed
scaffold-plugin --kind data_source- Agents automatically gain leverage via lake tools
Agents do not hardcode every source — they call openchat-data tools. More lake packages = smarter conversations for free.
Self-serve builder (simpler than n8n)
Users add their own API keys at /keys, then compose a linear timeline of actions at /builder— no canvas, no branches. Publish to the library or run in chat as user:<id>.