The Agent Skills Directory

[DATA_EXPOSURE]: The skill includes instructions to verify the FRAMER_API_KEY by echoing its first 20 characters to the terminal. While intended for debugging and used with a length limit, this practice exposes a portion of a sensitive credential to the agent's logs.
Evidence: echo $FRAMER_API_KEY | head -c 20 in SKILL.md.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes local source code using Grep and Read to diagnose errors. Malicious content embedded in the project files (such as comments in components) could potentially influence the agent's behavior during a diagnostic session.
Ingestion points: Uses Grep and Read on local source files (e.g., src/*.tsx).
Boundary markers: None identified in the prompt templates.
Capability inventory: Can execute Bash commands, including curl and npm.
Sanitization: No explicit sanitization or instruction to ignore embedded directives in the files being read.
[COMMAND_EXECUTION]: The skill uses Bash for environment checks and local development tasks, which is expected for its stated purpose as a developer tool.
Evidence: Use of npm run dev, npm list, and grep commands.

framer-common-errors