| name | ai-fomo-sources |
| description | Configure AI information sources and collect traceable raw snapshots for an AI FOMO workspace. Use when the user wants to add feeds, docs, X accounts, GitHub repos, YouTube videos, podcasts, podcast comments, audio transcripts, changelogs, RSS sources, or other source connectors. This skill handles source intake only and should not judge or file knowledge. |
AI FOMO Sources
Purpose
Connect AI-related information sources and write traceable raw snapshots into raw/. This skill handles intake only. It must not decide long-term value.
Use When
Use this skill when the user asks to:
- connect a source
- configure source registry
- collect new source material
- create raw snapshots
- add Feishu, Notion, GitHub, X, RSS, YouTube, podcast, podcast comments, transcript, changelog, or web sources
Do Not Use When
Do not use this skill to:
- judge whether a source is high value
- write
wiki/sources
- update
themes or dossiers
- generate signals or digests
Use ai-fomo for judgment after raw material exists.
Core Workflow
- Confirm the target AI FOMO workspace.
- Identify the source type and required access method.
- If the source is supported by
scripts/import_source.py, directly import it.
- Use
references/source-connectors.md when connector-specific guidance is needed.
- Prefer preview or dry-run before bulk writes.
- Store credentials outside committed files.
- Write source material only to
raw/inbox/ by default.
- Preserve source metadata: URL, title, author, source type, published time, collected time, and dedupe key when known.
- Report what was collected and what still needs authentication or manual review.
Direct Import Runtime
Use scripts/import_source.py for supported sources:
python scripts/import_source.py --workspace /path/to/workspace official-page --url https://example.com/changelog
python scripts/import_source.py --workspace /path/to/workspace rss --url https://example.com/feed.xml
python scripts/import_source.py --workspace /path/to/workspace github-repo --repo owner/name
python scripts/import_source.py --workspace /path/to/workspace github-trending --language python --since daily --limit 10
python scripts/import_source.py --workspace /path/to/workspace xiaoyuzhou-episode --url https://www.xiaoyuzhoufm.com/episode/episode_id
python scripts/import_source.py --workspace /path/to/workspace x-user --handle OpenAI
Use --dry-run first when importing a new source. x-user requires an X API bearer token in X_BEARER_TOKEN or a custom env var.
For Xiaoyuzhou bridge install/start, account login, subscribed inbox batch import, episode comments, and audio transcription, use scripts/xiaoyuzhou_account.py and read references/xiaoyuzhou-account.md.
python scripts/xiaoyuzhou_account.py --workspace /path/to/workspace install-bridge --dry-run
python scripts/xiaoyuzhou_account.py --workspace /path/to/workspace install-bridge --start
python scripts/xiaoyuzhou_account.py --workspace /path/to/workspace bridge-status
python scripts/xiaoyuzhou_account.py --workspace /path/to/workspace import-inbox --limit 30 --max-pages 3 --dry-run
python scripts/xiaoyuzhou_account.py --workspace /path/to/workspace import-comments --episode-id episode_id --include-replies --dry-run
python scripts/xiaoyuzhou_account.py --workspace /path/to/workspace transcribe-raw --raw-file /path/to/raw.md --dry-run
python scripts/xiaoyuzhou_account.py --workspace /path/to/workspace import-inbox --limit 10 --with-comments --with-transcripts --transcript-limit 3 --dry-run
Fallback Strategy
Direct import is the default. If a source is unsupported:
- create a manual raw snapshot from user-provided content, export, notes, or URL metadata
- add a registry entry only if the user explicitly wants to track the source long term
- hand off to
ai-fomo for judgment
Do not turn fallback setup into a long onboarding survey.
Raw Snapshot Requirements
Each raw snapshot should be traceable and minimally structured:
- source name and source id
- source type
- original URL or stable reference
- title
- author or publisher when known
- published date when known
- collected timestamp
- content or excerpt permitted by the source boundary
- metadata needed for later judgment
Credential Rules
- Use environment variables or ignored local files for credentials.
- Never commit tokens, cookies, or access keys.
- Create
.env.example with empty variable names only.
- If a credential was committed, tell the user it must be rotated before public release.
Handoff
After collection, tell the user that ai-fomo should be used to judge whether the raw material deserves knowledge work.
References
- Use
references/source-connectors.md for source-type recipes.
- Use
references/xiaoyuzhou-account.md for Xiaoyuzhou bridge install, login, subscribed inbox import, comments, and transcription.