# Primer public-evidence diagnostic share bundle - Account: `public` - Study: `primer-public-diagnostic` - Systems: Primer - Framework: `design-system-diagnostic/1.0.0` - Evidence included: `manifests` - Drafts allowed: `no` The canonical result for each system is `reports//scorecard.json`. The HTML, Markdown, and CSV files are presentation views of the same scorecard. `bundle.json` records the payload checksums and provenance. When a reviewed implementation roadmap exists, its sanitized canonical record is included at `roadmaps/.json`. It maps each score transition to delivery actions, acceptance criteria, retained evidence, suggested owners, dependencies, effort, and conditional score lift. Agent-performance actions remain separate from readiness-score changes. When publication-ready controlled agent-task results were explicitly requested, they are included separately under `agent-reports//`. They describe observed performance for the recorded actor profile while the five-area diagnostic retains its own score and evidence trail. Open `index.html` for the customer-facing Findings & Results view. It connects each finding to the exact bundled evidence file when raw evidence is included, or to the sanitized evidence manifest when the bundle uses its safer default. `config/evaluation.shared.json` is sanitized for sharing and is intentionally not a runnable study configuration. Assessor commands, credentials, local source paths, reviewer identities, and review notes are excluded by default. Manifest-only evidence proves which immutable snapshots supported the report without redistributing their contents. A bundle marked `full` contains raw evidence and must be reviewed for customer data, licensing, and confidentiality before it is forwarded.