Skip to content

eds: 24h auto-refresh for karen knowledge base#344

Draft
deepmasq wants to merge 2 commits intomainfrom
feat/karen-plan-26
Draft

eds: 24h auto-refresh for karen knowledge base#344
deepmasq wants to merge 2 commits intomainfrom
feat/karen-plan-26

Conversation

@deepmasq
Copy link
Copy Markdown
Contributor

Summary

  • Add rescan_period: 86400 instruction to Karen's collect-support-knowledge-base skill
  • The auto-rescan infrastructure already exists in both crawler (service_crawler.py:periodic_rescan_loop) and unstructured (service_unstructured.py:periodic_rescan_loop) services
  • Default is 604800 (7 days) — too slow for support KB. Karen needs 24h freshness.
  • Zero backend changes needed — just telling Karen to set the config when creating EDS

Karen plan item

Issue #26 (Fibery #2465) — EDS auto-refresh 24h

Test plan

  • Create a crawler EDS via Karen's KB collection skill
  • Verify eds_json.rescan_period is 86400 on the created EDS
  • Wait >24h or manually set eds_last_scan_ts = 0 to trigger rescan
  • Verify crawler re-indexes the source

🤖 Generated with Claude Code

the auto-rescan infrastructure already exists in both crawler and
unstructured services (periodic_rescan_loop). default is 7 days.
karen needs 24h for support kb freshness.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
If you have a documentation website, that's perfect. Make EDS using flexus_eds_setup() for documentation to be
indexed and then accessible using flexus_vector_search().

ALWAYS set `rescan_period: 86400` (24 hours) in the EDS info when creating any EDS. This ensures the knowledge
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ask the user every how much it should be refreshed proposing a default of 24h?

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done in f48bc8c — converted the hard ALWAYS set 86400 into "ask the user, propose 24h as default" with a small conversion guide (1h / 24h / 7d / one-time) and 86400 as the fallback when the user is indifferent. Kept it in the SKILL.md prompt rather than setup_schema.json because the rescan period is per-EDS and the user may set up several with different freshness needs (status page hourly vs. policy archive weekly), so a single global setup field would be the wrong shape.

@deepmasq
Copy link
Copy Markdown
Contributor Author

Fair clarifying question from @humbertoyusta. I'd lean toward landing the 24h default as-is in this PR and adding the "ask the user with 24h default" prompt as a small follow-up — keeps this one focused on the EDS plumbing. WDYT, ok to ship and chase the prompt question separately?

humberto on #344: instead of hardcoding rescan_period=86400, ask the
user during kb collection how often to refresh and propose 24h as the
default. different sources have different freshness needs (status pages
hourly, archives weekly), so a one-size global default is wrong.

skill text now tells karen to ask, lists the common conversions, and
falls back to 86400 if the user is indifferent.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants