Skip to content

timeline-report skill: example SQL references non-existent observations.source_tool column #3332

Description

@lunetics

Plugin version: 13.10.4 (marketplace install, plugins/cache/thedotmack/claude-mem/13.10.4)

The Token Economics section of skills/timeline-report/SKILL.md tells the analysis agent that the observations table has columns source_tool and source_input_summary, and the "Explicit recall events" example query filters on source_tool LIKE '%search%' ….

Verified against a live 13.10.4 database (PRAGMA table_info(observations)): neither column exists. The schema instead carries agent_type, generated_by_model, relevance_count (alongside discovery_tokens, created_at_epoch, etc.).

Effect: an agent following the skill instructions either fails on the query or silently reports 0 explicit recall events. In our run the agent noticed the mismatch and flagged the metric as unverifiable, but the skill doc steers less careful runs into fabricating that section.

Suggestion: update the column enumeration and the example SQL in the skill to the current schema, and point the recall-event heuristic at whichever field now carries tool/provenance information (perhaps agent_type or narrative-text matching only).

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions