The Agent Skills Directory

[DATA_EXFILTRATION]: The skill is configured to read from sensitive local directories to aggregate personal information, specifically accessing ~/.claude/LIFEOS/USER/TELOS for missions and goals, PRINCIPAL_IDENTITY.md for bio data, and MEMORY/WORK for session themes.
[DATA_EXFILTRATION]: Deployment workflows transmit aggregated data to external destinations, including a remote Git repository via git push and an API endpoint at https://mcp.daemon.example.com using curl POST requests.
[DATA_EXFILTRATION]: The skill performs local network requests to http://localhost:31337/notify to send status updates during workflow execution.
[COMMAND_EXECUTION]: The skill executes several shell commands to manage data and deployment, including git commit, git push, and cp operations, as well as executing local TypeScript tools through the bun runtime.
[EXTERNAL_DOWNLOADS]: The skill runs bun install during its deployment phase, which retrieves external package dependencies from the public NPM registry.
[PROMPT_INJECTION]: The skill ingests content from various local files which could serve as a vector for indirect prompt injection if those files contain untrusted data. This risk is addressed by the skill's architecture, which includes a deterministic SecurityFilter for redaction and a requirement for explicit user approval before any data is written or deployed.

Daemon