A durable library of gradeable rubrics, straight from the Managed Agents schema.
This repo has a pattern: take something real and durable, and turn it into a small D1-backed primitive the rest of the ecosystem can reference — subagentcontracts.com did it for the Contract, subagenttasks.com did it for the Task, subagentcache.com did it for real Cloudflare cache analytics. subagentrubrics.com does the same thing for Anthropic's own Managed Agents outcome schema.
What a rubric actually is, per the real docs
Per platform.claude.com/docs/en/managed-agents/define-outcomes.md: “A rubric is a markdown document describing per-criterion scoring… The rubric is required.” The same doc's guidance on how to write one: “Structure the rubric as explicit, gradeable criteria” rather than vague ones. Those are the two ideas this whole site generalizes into a stored, queryable primitive — a rubric is just a markdown document with a countable number of gradeable criteria, typed as either inline text or an uploaded file.
Latest rubric text · rbc_cowork_feature_adoption · 9 criteria
Seeded with the rubric actually in use this session
rbc_worker_quality is not a placeholder — it's the real rubric subagentevaluations.com scores this session's 9 shipped workers against: durable persistence, real API contract, design-system compliance, provenance and grounding, and family integration. Five criterion groups, fifteen gradeable bullet points, stored here as one markdown document per the schema's own shape.
Explore
Browse the rubric library → · Read how the real rubric field is shaped → · Hit the JSON API →