v0.8.0

Summary

• Add static skill content scanning for prompt-injection, credential access, exfiltration wording, and stealth Unicode.
• Risky skills now require explicit confirmation on import and per-agent attach.
• Scan findings are surfaced through CLI, API, and web UI.

Validation

• Local: pnpm typecheck passed.
• Local: pnpm test passed (80 files, 637 tests).
• Local: pnpm build passed.
• GitHub Actions Release workflow passed: https://github.com/rullopat/bazilion/actions/runs/28737186612

Published packages

• bazilion@0.8.0
• @bazilion/client@0.8.0
• @bazilion/api-types@0.8.0