persona-evaluatorv0.3.4

Automate persona quality checks in a CI/CD pipeline using the structural mode to enforce minimum dimension scores before deployment. Developers commit persona changes, and the pipeline runs npx openpersona evaluate to fail builds if scores drop below thresholds.

A persona creator uses the semantic white-box mode to get a narrative quality review of their own pack, receiving scores and actionable improvements for dimensions like Soul and Identity. This helps refine the agent's character and consistency before publishing.

A marketplace for agent personas allows creators to request peer reviews from other developers. The peer uses the semantic white-box mode with --pack-content to evaluate narrative quality, building trust among users.

A security auditor evaluates an unknown remote agent (non-OpenPersona) using the semantic black-box mode, performing A2A handshake, consent probing, and passive observation to assess trustworthiness and safety without filesystem access.

An enterprise assesses third-party AI agents before integrating them into their workflow. Using semantic black-box evaluation, they probe for compliance with corporate policies and safety standards, with confidence-capped scores for decision-making.

Offer a subscription service where personas are automatically re-evaluated daily/weekly in a CI pipeline, with alerts on quality regressions. Revenue comes from tiered monthly fees based on number of personas monitored.

A persona marketplace licenses the evaluator to provide verified quality badges. Creators pay per evaluation for a badge that increases buyer trust. Revenue is shared per evaluation or via flat licensing fee.

Offer one-time or recurring black-box audits for companies that want to assess external agents before deployment. Includes a detailed report with risk scores and recommendations. Revenue per audit engagement.