Release And Versioning

Issue 91 Edition 2026-04-01 3 min read

Not accepted General

Sources: 1 • Confidence: Medium • Updated: 2026-04-13 03:32

Key takeaways

datasette-enrichments-llm version 0.2a1 has been released.
When an enrichment is triggered, the triggering actor is passed through to the LLM call as the actor argument to llm.mode(...).

When an enrichment is triggered, the triggering actor is passed through to the LLM call as the actor argument to llm.mode(...).

What other behavioral changes, fixes, or breaking changes (if any) are included in datasette-enrichments-llm 0.2a1 besides actor propagation?
How is the actor value defined/serialized, and under what conditions might it be null, anonymized, or unavailable during enrichment triggers?
Does passing actor into llm.mode change authorization boundaries (for example, enabling per-actor access control) or is it informational only?
Is there any direct decision read-through (operator, product, or investor) tied to these changes, such as recommended upgrade timing, migration steps, or compatibility guarantees?

Actor propagation into LLM calls could enable better auditability and per-user context handling in integrations, potentially making the enrichment system more suitable for multi-user or compliance-sensitive deployments.
The prerelease nature of 0.2a1 suggests possible near-term iteration risk for adopters, which can affect upgrade timing and integration stability for teams building around enrichments.

Release notes or documentation for 0.2a1 describing additional changes beyond actor propagation, including any compatibility guarantees or migration guidance.
Examples or guidance showing actor used for access control, logging, rate limits, or policy enforcement in llm.mode integrations.
Evidence that actor is consistently available and well-defined across enrichment triggers, including how it is serialized and handled for anonymous users.

Actor is frequently null, anonymized, or inconsistently provided, limiting practical utility for audit or per-user behavior.
Documentation clarifies actor is informational only and not used by llm.mode or downstream tooling, reducing product or adoption impact.
0.2a1 includes breaking changes or instability that discourages upgrades or causes integration regressions.