Generate a generative-adversarial agent pair in a Ratchet configuration.
npx skills add https://github.com/netbrain/ratchet --skill ratchet-pair
Please help me install this Agent Skill. Skill: ratchet:pair Source: https://github.com/netbrain/ratchet/tree/main/skills/pair Command: npx skills add https://github.com/netbrain/ratchet --skill ratchet-pair
The ratchet:pair Skill streamlines adding a new generative-adversarial agent pair to an existing Ratchet project so teams can expand coverage for specific quality dimensions without manual boilerplate or configuration errors.
Ask Ratchet to add a new pair by invoking /ratchet:pair security to create a security-focused generative and adversarial agent pair.
High-intent search queries and answers about installing and using this skill.
To add a generative-adversarial agent pair to a Ratchet workflow, use the interactive pair creation process which proposes names and scopes, generates agent prompts, and updates workflow.yaml automatically.
Generative-adversarial agent pairs are used in automated workflows to expand quality coverage for specific dimensions like security, style, or performance by pairing a builder agent with a critic agent.
Yes, you need an existing Ratchet project configuration because the generation process reads .ratchet/project.yaml and .ratchet/workflow.yaml to avoid overlap with existing pairs and bake project-specific knowledge into the prompts.
To automatically register and verify a new agent pair, the workflow writes generative.md and adversarial.md files, updates the workflow.yaml configuration, and runs shell tests to verify successful creation.
Yes, when a name is not provided, the system uses analyst prompts and AskUserQuestion interactions to propose names, scopes, and quality focuses interactively for the new agent pair.
netbrainManage deterministic lint, test, and security guards in Ratchet workflows.
netbrainRecords ad-hoc discoveries into the Ratchet plan's .ratchet/plan.yaml.
netbrainAnalyze Ratchet improvement signals and generate prioritized fixes for prompts, guards, and workflows.
EarthmanWebExecute sequential multi-agent workflows via custom prompt chains and predefined pipelines.
