The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The references/troubleshooting.md file contains instructions to bypass sandbox security restrictions. Specifically, it suggests using the --sandbox danger-full-access flag in the OpenAI Codex environment to resolve network connectivity issues.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external, potentially untrusted sources, creating an attack surface for indirect prompt injection.
Ingestion points: Data enters the context via URLs (web/YouTube), Google Drive documents, and local files as defined in SKILL.md and references/mcp_tools.md.
Boundary markers: There are no instructions provided to delimit untrusted source content or warn the agent to ignore embedded instructions within that content.
Capability inventory: The skill utilizes bash for CLI command execution and multiple MCP tools that perform file system operations, network requests, and content generation.
Sanitization: No sanitization, validation, or filtering of external source content is implemented before it is processed by the agent.
[COMMAND_EXECUTION]: The skill heavily relies on the execution of shell commands through the nlm CLI via bash. While this is the primary purpose of the skill, it requires high trust in the nlm binary and the environment where it runs.

nlm-skill