feat(skills): add dq-job-from-data-contract skill + behavioral evals - #138
Open
BethHoffmann wants to merge 1 commit into
Open
feat(skills): add dq-job-from-data-contract skill + behavioral evals#138BethHoffmann wants to merge 1 commit into
BethHoffmann wants to merge 1 commit into
Conversation
Adds a new collibra/dq-job-from-data-contract skill that turns a Data Contract's manifest clauses into an enforcing DQ job and tolerance-0 rules, and ports its test plan into chip's existing Python behavioral eval harness as a new skill-level suite (5 task scenarios + 20 trigger scenarios covering when the skill should/shouldn't load). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
collibra/dq-job-from-data-contractskill that turns a Collibra Data Contract's manifest clauses into an enforcing Data Quality job plus tolerance-0 rules (monitors), reusingcollibra/dq-rules/collibra/dq-rule-workbenchfor the per-rule mechanics.SKILLS.mdand thecollibra/indexnavigator, and cross-links it fromcollibra/dq-rules/collibra/dq-rule-workbench.evals/dq_job_from_data_contract/— 5 task scenarios (full flow, existing-job reuse, verify-and-results, impossible time-slice, referential-integrity clause blocked) plus 20 trigger scenarios (10 should-trigger / 10 should-not-trigger), all graded deterministically off the tool-call transcript against a dedicated mock Collibra fixture (EMPLOYEES table + data contract). No LLM-graded rubrics, matching this repo's existing convention.confirm=true), since the underlying skill performs live writes in Collibra.Known simplifications are called out in
evals/dq_job_from_data_contract/README.md's "Known simplifications" section (e.g. prompts pass the fixture's data contract ID directly rather than exercising asset-search discovery, which is out of scope for this skill and already covered bycollibra/discovery).Test plan
go build ./...go vet ./...go test ./pkg/skills/...python3 -m py_compileon every new eval Python filescenarios.yaml/trigger_scenarios.yamland confirmed scenario counts (5 task, 20 trigger — 10/10 split)evals/dq_job_from_data_contract/runner.pyagainst a live model (ANTHROPIC_API_KEY) — not run as part of this PR, same as other suites inevals/🤖 Generated with Claude Code