Reviews planning artefacts against per-type quality checks and reports severity-ranked findings.
npx skills add https://github.com/DATA-AI-XYZ/Tandem --skill critique-data-ai-xyz
Please help me install this Agent Skill. Skill: critique Source: https://github.com/DATA-AI-XYZ/Tandem/tree/main/skills/critique Command: npx skills add https://github.com/DATA-AI-XYZ/Tandem --skill critique-data-ai-xyz
type: frontmatter (or infers from filename prefix) and runs the matching checks for Epic, Feature, Story, or Testplan. - Severity-ranked findings: Every check gets a PASS/WARN/FAIL verdict mapped to blocker, major, or minor findings, each with a concrete suggested fix. - Premise-vs-reality reconciliation: Resolves referenced artefact IDs and compares their actual status: against claims made in the artefact, flagging false premises as major findings. - Use Case: Before promoting STORY-05.1.02 to ready, run a critique to discover it has an XL estimate, a missing type_of_work: field, and an AC with no matching test case — then fix each with the suggested edits before the DoR gate. ## Quick Start Ask the assistant to critique the planning artefact at a given path, for example: critique the story at _00-Project-Management/32-Stories/EPIC-05/FEAT-05.1/STORY-05.1.02-login.md.High-intent search queries and answers about installing and using this skill.
Run the critique on the story's file path. It detects the artefact type from frontmatter, runs story-specific checks like machine-testable ACs, DoR readiness, and estimate sanity, then reports blocker, major, and minor findings with concrete suggested fixes.
Story checks cover machine-testable acceptance criteria, Definition of Ready readiness, estimate sanity with XL flagged for splitting, a maximum of five ACs, AC-to-test-case coverage against the paired testplan, a set type_of_work field, and premise-versus-reality reconciliation of referenced artefacts.
No, the critique is strictly advisory and read-only. It never edits files, never changes status fields, and never promotes artefacts; all improvements appear as suggested fixes in the findings report for the user to apply.
Yes, but only when explicitly requested. The report is written to 41-Reports/reviews/CRITIQUE-<artefact-id>-<date>.html using semantic HTML with no external CSS, containing the same severity-ranked findings as the chat output.
The skill first reads the type: frontmatter field, then falls back to inferring from the filename prefix such as EPIC-, FEAT-, STORY-, or TESTPLAN-. If both fail, it stops and asks the user which type to treat the artefact as.
ShoebtamboliConvert planning lists into structured markdown task files with YAML frontmatter.
TencentBlueKingClarifies TAPD backlog stories and generates standardized requirement documents with DoR gating.
Obsidian-OwlMap acceptance criteria to implementation and test evidence in a codebase.
heymishyExecute Definition of Ready checks and generate coding agent instructions.