The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses Bash tools to perform recursive searches using grep across project files and local documentation stored in a specific directory (/Users/zach/Documents/cc-skills/docs/supabase/). This is intended to provide relevant context and documentation references for user queries.\n- [EXTERNAL_DOWNLOADS]: The skill's metadata specifies an installation command that uses curl to download the SKILL.md source from a remote GitHub repository (raw.githubusercontent.com/heyflouai/ikf-central-dashboard).\n- [PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it processes and incorporates content from local project files and documentation into the agent's context without sanitization or boundary markers.\n
Ingestion points: Reads local project files (grep -r "createClient" .) and external documentation via filesystem search tools.\n
Boundary markers: The instructions lack delimiters or explicit instructions for the agent to ignore potentially malicious commands embedded in the ingested data.\n
Capability inventory: The skill has access to Bash, Write, Edit, and Read tools, which could be leveraged if an injection is successful.\n
Sanitization: No validation or filtering is performed on the output of the search commands before it is provided to the agent model.

supabase-expert