| name | sage-feed-researcher |
| description | Use Sage's existing Feed engine for fresh research, trend, and topic questions. Prefer Sage Feed MCP tools over ad hoc web fetching when the user asks what is new, relevant, or worth reading. |
| version | 0.1.0 |
| metadata | {"openclaw":{"emoji":"📰","homepage":"https://github.com/EvanL1/sage","tags":["feed","research","mcp","trends"]}} |
Sage Feed Researcher
Use this skill when the user asks about fresh research, current trends, new
repos, papers, articles, or topic monitoring.
Workflow
- Search existing feed context first:
sage_feed_search(query, limit)
sage_feed_items(query, limit)
sage_feed_digest(date)
- If results are stale, sparse, or missing the topic, ask permission to refresh
and call:
sage_feed_pull
This reuses Sage's daemon flow: configured sources, feed filtering, deep-read
budget, observations, and digest generation. Do not reimplement crawling inside
the skill.
- Answer with citations from feed item URLs. Distinguish:
- collected facts from feed items
- Sage's generated insight/summary
- your inference
- Record feedback when the user reacts:
sage_feed_feedback(action="bookmark" | "archive" | "unarchive" | "learned")
Rules
- Prefer Sage Feed MCP tools over raw browser/search when the topic matches the
user's configured interests or active projects.
- Do not claim real-time freshness unless
sage_feed_pull succeeded or the item
timestamps support it.
- Keep answers useful: summarize the signal, why it matters to the user, and the
next concrete reading or action.
