The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes system commands including git and pnpm (e.g., pnpm build, pnpm test:all, pnpm redeploy) to manage repository state and build processes.
[EXTERNAL_DOWNLOADS]: Communicates with local infrastructure via localhost:3737 and localhost:27141 to fetch system status, alerts, and telemetry data.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface by processing untrusted data from the vault.
Ingestion points: Reads content from 00_Inbox/**/*.md, .claude/skills/**/*.md, .claude/plugins/**/*.md, and project/goal files.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are defined for the LLM context.
Capability inventory: The skill can trigger build scripts, deployment commands, and git operations based on its analysis of vault state.
Sanitization: No content validation or filtering is performed before the data is processed by the agent or the underlying LLM (Ollama).

jarvis