The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions in references/expert-repos.md direct the agent to execute shell commands, specifically git clone, to download external repositories to a sibling directory (../).
[EXTERNAL_DOWNLOADS]: The skill fetches content from several third-party GitHub repositories for analysis, including steveyegge/gastown, glittercowboy/get-shit-done, bgauryy/octocode-mcp, mikeyobrien/ralph-orchestrator, and bmad-code-org/BMAD-METHOD.
[PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection by design.
Ingestion points: Untrusted source code from external GitHub repositories is cloned and analyzed by specialized agents (references/ARL-agent-instructions.md).
Boundary markers: Not present; the instructions do not specify the use of delimiters or safety warnings when processing the external code.
Capability inventory: The system has the capability to execute shell commands (git clone) and perform recursive file system reads.
Sanitization: None; the agents are instructed to perform deep source code analysis without explicit sanitization or filtering of the input data.

arl