agent-chronicle

AI-powered diary generation for agents - creates rich, reflective journal entries (400-600 words) with Quote Hall of Fame, Curiosity Backlog, Decision Archae...

Key things to consider before installing: - Metadata mismatch: the registry metadata claims no required env vars, but the SKILL.md and scripts require a SKILLBOSS_API_KEY. Do not supply secrets until you verify the publisher and understand the endpoint. - Data sent externally: generate.py aggregates session logs and persistent diary files and POSTs them to https://api.heybossai.com/v1/pilot. If your session logs contain sensitive user data, credentials, or private content, those will be transmitted to SkillBoss. Consider whether that is acceptable for your environment. - Privacy controls: the skill has a privacy level setting and an interactive fallback (you can run generate.py --interactive or pipe content via --from-stdin). Use interactive/local modes if you prefer not to send logs externally. Review config.json options (privacy_level, memory_integration) and disable automatic memory integration or auto-generation if undesired. - Verify provenance: the package references GitHub URLs and authors, but the registry owner and homepage are not authoritative. If this skill will run in production or with sensitive data, verify the repository upstream (review commits, author identity) before trusting it. - Exports and dependencies: export functionality invokes pandoc or WeasyPrint and will call subprocesses. Ensure those binaries/libraries are installed from trusted sources; exports will create temporary files under the workspace. - If you need stronger assurance: inspect scripts/generate.py and scripts/export*.py yourself (they're included) or run the generation in a sandboxed workspace containing only non-sensitive session files. If you want the skill but not remote calls, consider editing the code to use interactive/local generation or to route to an internal model endpoint. If you want, I can list the exact lines that perform external network calls and file reads so you can audit them or produce a safe, local-only variant of the script.

Type: OpenClaw Skill Name: marjorie-agent-chronicle Version: 1.0.0 The agent-chronicle skill is a functional tool designed to generate reflective diary entries for AI agents by analyzing session logs. It uses the SkillBoss API Hub (api.heybossai.com) for AI generation, a dependency clearly disclosed in SKILL.md and README.md. The Python scripts (generate.py, export.py, setup.py) are well-structured, transparently log their activities to the console, and employ security best practices such as using the --sandbox flag with pandoc. No evidence of malicious intent, unauthorized data exfiltration, or prompt injection was found.