The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a shell command using find to locate its reference documentation at runtime.
Evidence: The script in SKILL.md searches through multiple paths including ~/.claude/plugins, ~/.claude/skills, and ~/.codex/skills.
[COMMAND_EXECUTION]: The discovery mechanism targets hidden and potentially sensitive directories used by the AI agent platform to store plugins and skills, which could lead to unauthorized discovery of other installed extensions.
[EXTERNAL_DOWNLOADS]: The documentation in references/geo-techniques.md directs users to clone and execute code from third-party GitHub repositories (github.com/AI2HU/gego and github.com/jonradoff/llmopt). These sources are not associated with the skill author or a trusted vendor list.
[DATA_EXFILTRATION]: The skill represents a surface for indirect prompt injection by ingesting untrusted content via WebFetch or file system access and processing it alongside its own logic.
Ingestion points: URL content fetched via WebFetch or local files read at Step 3.
Boundary markers: None identified in the prompt templates to distinguish between instructions and untrusted data.
Capability inventory: Shell command execution (find) and network operations (WebFetch).
Sanitization: No explicit sanitization or validation of external content is mentioned before processing.

geo-optimizer