Prior NEARWEEK work on automated AI x crypto news sourcing and editorial workflow design.
This repo is part of the public work map behind the current SAIBA direction: public ecosystem signals become structured source material, humans review the output, and approved items can move into content or distribution workflows.
Prior work / public archive.
The repo is useful as a technical and product reference. Treat deployment notes, automation scripts, and integration docs as historical context, not as a current production runbook.
- Monitoring public crypto and AI x crypto signals across X, ecosystem accounts, GitHub, and curated source lists.
- Turning noisy public updates into draft newsletters, reports, and social content candidates.
- Routing candidate content through Buffer, Telegram, GitHub, Zapier, and local webhook-style workflows.
- Keeping a manual approval step before public distribution.
- Comparing lightweight automation glue against more durable pipeline architecture.
| Path | Purpose |
|---|---|
src/ |
Express/webhook server and integration glue. |
scripts/ |
Experiments for X/Twitter sourcing, newsletters, Buffer/Zapier-style routing, and media processing. |
configs/ |
Historical config examples for content streams and routing decisions. |
docs/ |
Implementation notes, scaling analysis, and integration write-ups. |
reports/ |
Example/generated output from prior runs. |
| Area | Proof |
|---|---|
| Ecosystem intelligence | Public signals can be turned into reusable source material instead of disappearing in feeds. |
| Content workflow | Automation helps most when it prepares drafts and context, not when it bypasses editorial judgment. |
| Distribution ops | Buffer, Telegram, GitHub, Zapier, and webhooks can form a lightweight content operations loop. |
| Bridge to SAIBA | The same pattern now appears in Agency OS and Link2Post: source signal -> draft -> human approval -> queued distribution -> receipt. |
This repo is not presented as a live autopublishing system.
- Human review is expected before publishing.
- Old performance targets should be read as experiment goals, not guaranteed production metrics.
- API keys, webhook secrets, account IDs, and live channel IDs should not be committed.
- Customer-facing or public social posts should not be sent automatically.
Install dependencies and inspect scripts locally:
npm install
npm run health
npm run testSome scripts expect local environment variables or historical integrations and may need adaptation before reuse.
Use .env.example as the public template. Keep real values in a local secret
store such as 1Password, GitHub Actions secrets, or the deployment platform.