ai-tech-rss-fetch — Claude Code skill: what it does, how to install
ai-tech-rss-fetch fetches RSS and Atom feeds, deduplicates entries, and stores metadata in SQLite. Keeps article titles and dates without downloading full body text.
MIT 236 installs ★ 11 updated 2026-10-19
npx skills add tiangong-ai/agent-skills --skill ai-tech-rss-fetch -a claude-codeWhat ai-tech-rss-fetch does
The skill handles RSS and Atom subscriptions via scripts/rss_subscribe.py. Feeds can be added individually by URL or in bulk from OPML files.
Under the hood, it pairs feedparser with sqlite3 to store incoming feeds and entries. Incremental sync relies on layered identity keys—such as GUID, canonical URL, and content hashes—while respecting HTTP 304 Not Modified headers. It captures only metadata; article body scraping and AI summaries are deliberately omitted.
Running the sync script requires feedparser installed in Python. In multi-agent environments, pointing AI_RSS_DB_PATH to an absolute path is required so separate processes target the exact same SQLite file.
What it helps with
- store normalized RSS and Atom feed metadata in SQLite
- import feed collections directly from OPML files
- prevent duplicate articles across syncs using layered keys and hashes
- track incremental feed updates using conditional HTTP requests
Example prompts for the agent
- Import feed subscriptions from assets/hn-popular-blogs-2025.opml into the database
- Run an incremental sync for up to 20 feeds and 100 items per feed
- Add https://example.com/feed.xml to subscriptions and list recent entries
How to install ai-tech-rss-fetch
1Skills CLI (Vercel)
npx skills add tiangong-ai/agent-skills --skill ai-tech-rss-fetch -a claude-code2Manually: put the files in place
~/.claude/skills/ai-tech-rss-fetch/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/ai-tech-rss-fetch/
1Skills CLI (Vercel)
npx skills add tiangong-ai/agent-skills --skill ai-tech-rss-fetch -a codex2Manually: put the files in place
.agents/skills/ai-tech-rss-fetch/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add tiangong-ai/agent-skills --skill ai-tech-rss-fetch -a cursor2Manually: put the files in place
.cursor/skills/ai-tech-rss-fetch/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install tiangong-ai/agent-skills ai-tech-rss-fetch2Manually: put the files in place
.github/skills/ai-tech-rss-fetch/SKILL.md1Skills CLI (Vercel)
npx skills add tiangong-ai/agent-skills --skill ai-tech-rss-fetch -a gemini-cli2Platform CLI
gemini skills install https://github.com/tiangong-ai/agent-skills --path ai-tech-rss-fetchManually: put the files in place
.devin/skills/ai-tech-rss-fetch/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add tiangong-ai/agent-skills --skill ai-tech-rss-fetch -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about ai-tech-rss-fetch
How to install ai-tech-rss-fetch in Claude Code?
Run npx skills add tiangong-ai/agent-skills --skill ai-tech-rss-fetch -a claude-code. The skill definition will be saved to ~/.claude/skills/ai-tech-rss-fetch/SKILL.md.
What does ai-tech-rss-fetch do?
It manages RSS and Atom feeds, tracks publication metadata in an SQLite database, and handles deduplication across repeated runs.
Does ai-tech-rss-fetch extract full article content?
No, it stores only metadata such as URLs, titles, authors, categories, and publication dates without downloading full articles or creating summaries.
Source text
SKILL.md · MIT
---
name: ai-tech-rss-fetch
description: Subscribe to AI and tech RSS feeds and persist normalized metadata into SQLite using mature Python tooling (feedparser + sqlite3). Use when adding feed URLs/OPML sources, running incremental sync with deduplication, and storing entry metadata without full-text extraction or summarization.
---
# AI Tech RSS Fetch
## Core Goal
- Subscribe to RSS/Atom sources.
- Persist feed and entry metadata to SQLite.
- Deduplicate entries with layered identity keys plus content fingerprints.
- Keep only metadata; do not fetch full article bodies and do not summarize.
## Triggering Conditions
- Receive a request to subscribe RSS feeds from URLs or OPML.
- Receive a request to run incremental RSS sync reliably.
- Need stable metadata persistence for downstream processing.
- Need dedupe-safe storage of feed items over repeated runs.
## Workflow
1. Prepare runtime and database.
- Ensure dependency is installed: `python3 -m pip install feedparser`.
- In multi-agent runtimes, pin DB to an absolute path before any command:
```bash
export AI_RSS_DB_PATH="/absolute/path/to/workspace-rss-bot/ai_rss.db"
```
- Initialize SQLite schema once:
```bash
python3 scripts/rss_subscribe.py init-db --db "$AI_RSS_DB_PATH"
```
2. Add feed subscriptions.
- Add one feed URL:
```bash
python3 scripts/rss_subscribe.py add-feed --db "$AI_RSS_DB_PATH" --url "https://example.com/feed.xml"
```
- Import feeds from OPML:
```bash
python3 scripts/rss_subscribe.py import-opml --db "$AI_RSS_DB_PATH" --opml assets/hn-popular-blogs-2025.opml
```
3. Run incremental sync.
- Fetch active feeds and store metadata:
```bash
python3 scripts/rss_subscribe.py sync --db "$AI_RSS_DB_PATH" --max-feeds 20 --max-items-per-feed 100
```
- Optional one-feed sync:
```bash
python3 scripts/rss_subscribe.py sync --db "$AI_RSS_DB_PATH" --feed-url "https://example.com/feed.xml"
```
4. Query persisted metadata.
- List feeds:
```bash
python3 scripts/rss_subscribe.py list-feeds --db "$AI_RSS_DB_PATH" --limit 50
```
- List recent entries:
```bash
python3 scripts/rss_subscribe.py list-entries --db "$AI_RSS_DB_PATH" --limit 100
```
## Input Requirements
- Supported inputs:
- RSS XML feed URLs.
- OPML feed list files.
## Output Contract (Metadata Only)
- Persist `feeds` metadata to SQLite:
- `feed_url`, `feed_title`, `site_url`, `etag`, `last_modified`, status fields.
- Persist `entries` metadata to SQLite:
- `id`, `dedupe_key` (compat primary identity snapshot), `guid`, `url`,
`canonical_url`, `title`, `author`, `published_at`, `updated_at`, `summary`,
`categories`, `content_hash`, `match_confidence`, timestamps.
- Persist `entry_identities` mapping table to SQLite:
- `entry_id`, `key_type`, `key_value`, `created_at`.
- Supported key types: `guid`, `canonical_url`, `legacy_guid`, `fallback_hash`.
- Do not store generated summaries and do not create archive markdown files.
## Configurable Parameters
- `db_path`
- `AI_RSS_DB_PATH` (recommended absolute path in multi-agent runtime)
- `opml_path`
- `feed_urls`
- `max_feeds_per_run`
- `max_items_per_feed`
- `user_agent`
- `seen_ttl_days`
- `enable_conditional_get`
- Example config: `assets/config.example.json`
## Error and Boundary Handling
- Feed HTTP/network failure: keep syncing other feeds and record `last_error`.
- Feed `304 Not Modified`: skip entry parsing and keep state.
- Missing `guid` and `link`: use hashed fallback identity and set `match_confidence=low`.
- Dependency missing (`feedparser`): return install guidance.
## Final Output Checklist (Required)
- core goal
- trigger conditions
- input requirements
- metadata schema
- dedupe and sync rules
- command workflow
- configurable parameters
- error handling
Use the following simplified checklist verbatim when the user requests it:
```text
核心目标
输入需求
触发条件
元数据模型
去重与同步规则
命令流程
可配置参数
错误处理
```
## References
- `references/input-model.md`
- `references/output-rules.md`
- `references/time-range-rules.md`
## Assets
- `assets/hn-popular-blogs-2025.opml` (candidate feed pool)
- `assets/config.example.json`
## Scripts
- `scripts/rss_subscribe.py`
© tiangong-ai · MIT · tiangong-ai/agent-skills