Evaluates task commits against specifications using dual logic and consistency checks.
npx skills add https://github.com/dcarmitage/claws --skill judge-dcarmitage
Please help me install this Agent Skill. Skill: judge Source: https://github.com/dcarmitage/claws/tree/main/skills/dual-judge Command: npx skills add https://github.com/dcarmitage/claws --skill judge-dcarmitage
Run dual-judge evaluation (logic + consistency) on a task's commit to verify whether the output meets the specification. This is triggered automatically after a task is marked done, or can be invoked manually with /judge.
Use the judge skill to evaluate the latest commit against its spec via /judge.
High-intent search queries and answers about installing and using this skill.
Automate code review by running dual-judge evaluation on a task commit to verify logic and consistency against the specification. This QA workflow checks if outputs meet design requirements before progressing.
Dual-judge evaluation is a QA process that runs parallel LLM-based logic and consistency checks on a commit. It assesses code quality and alignment with specifications to generate a pass or fail verdict.
Run a consistency and logic judge after marking a task done to ensure the submission satisfies the spec. It can be triggered automatically in task pipelines or invoked manually to validate outputs.
Yes, you can run LLM-based quality assurance without external dependencies. The judge skill operates standalone using internal logic_judge and consistency_judge scripts to evaluate repository commits.
Evaluate the latest commit by invoking the manual /judge command. This triggers the dual-judge evaluation to validate the commit against an optional spec path and taskboard context.
Automated LLM judges require access to the repository commit, optional spec path, and taskboard context to generate a verdict. Without these inputs, the logic and consistency checks cannot evaluate alignment.
danielgapRuns two parallel LLM reviewers to produce a fused verdict report.
tile-aiValidate Python and C++ code formatting against CI rules and generate JSON/Markdown reports.
am-willOrchestrate parallel subagents to execute development plan tasks in waves.
SteelMorganSubmit project artifacts to external LLMs via Codex CLI for review.
