Skip to content

assess-strat

Assess RHAISTRAT strategies against quality criteria using a scored rubric with calibration examples. Scores across four dimensions: feasibility, testability, scope, and architecture.

Plugin Details

Skills

Skill Description Invocable
/assess-strat Assess strategies against quality criteria using a structured rubric
/export-rubric Export the assessment rubric

Contract Summary

Headline view only. Individual skill pages carry the detailed measures, references, success conditions, and invariants.

Focus Functions

  • review — Assess an artifact against expectations and identify issues, risks, or fit.
  • generate — Produce a new artifact for the user or another tool to consume.

Focus Metrics

  • task_success — Whether the skill completes the intended job correctly for the task. (Prefer deterministic or verifier-backed checks; use judge only as a fallback.)
  • evidence_completeness — Whether claims and verdicts are backed by enough concrete evidence. (Use verifier-backed checks when evidence can be counted; otherwise use a rubric-backed judge.)
  • output_quality — Human-judged quality of the final artifact when deterministic checks are insufficient. (Judge only; always pair it with a stable rubric_ref and, when available, calibration data.)
  • latency — How quickly the skill produces the final usable result. (Deterministic only; measure elapsed wall-clock time for the user-visible outcome.)

Notes

Strategy assessment against a scored rubric and rubric export.

Installation

Claude Code

/plugin install assess-strat@opendatahub-skills

OpenAI Codex — add the marketplace, then enable assess-strat from the /plugins browser:

codex plugin marketplace add opendatahub-io/skills-registry