DATASETTE ENRICHMENT PLUGIN 0.2A1 PASSES ACTOR TO LLM.MODE FOR BETTER AUDITABILITY
datasette-enrichments-llm 0.2a1 passes the triggering actor to llm.mode(), tightening auditability and enabling per-user behavior in LLM enrichments. The relea...
datasette-enrichments-llm 0.2a1 passes the triggering actor to llm.mode(), tightening auditability and enabling per-user behavior in LLM enrichments.
The release of datasette-enrichments-llm 0.2a1 adds a small but useful change: the actor who kicks off an enrichment is now available in llm.mode(..., actor=actor). Your custom modes can now log who did what and adjust prompts or routing based on the caller.
If you run multi-tenant Datasette or care about audit trails, this gives you cleaner provenance and an easy hook for quotas or policy checks. It’s an alpha bump, but the surface area is tiny and simple to test.
Clear actor propagation improves audit trails and makes per-user prompt policies or quotas straightforward.
Multi-tenant deployments can condition LLM behavior on the requester without plumbing custom identity logic.
-
terminal
Verify actor is received in custom llm.mode handlers and affects prompts/output as expected across different users.
-
terminal
Check logs/metrics to ensure actor is recorded for every enrichment; confirm no leakage between tenants.
Legacy codebase integration strategies...
- 01.
Upgrade the plugin and ensure existing enrichments still run; add actor-aware logging where helpful.
- 02.
If you previously inferred user from context, migrate to the actor parameter and add fallbacks for historic runs.
Fresh architecture paradigms...
- 01.
Design enrichments with actor-aware prompts, policies, and per-user routing from the start.
- 02.
Add per-actor quotas and provenance logs to simplify compliance and debugging.