The Agent Skills Directory

[DATA_EXPOSURE_AND_EXFILTRATION]: The skill references an external documentation URL on a non-whitelisted third-party domain (jeffallan.github.io). Automated scanners have identified this specific resource as blacklisted, which may indicate it has been associated with malicious activity in the past.
Evidence: SKILL.md provides a link to https://jeffallan.github.io/claude-skills/skills/specialized/embedded-systems/.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to process user-supplied specifications for microcontroller firmware, creating a vector for indirect prompt injection if those specifications contain malicious instructions.
Ingestion points: The 'Analyze constraints' workflow step in SKILL.md consumes MCU specifications, memory limits, and timing budgets provided by the user.
Boundary markers: Absent. The instructions do not include delimiters or specific warnings for the agent to distinguish between data and embedded instructions in the user input.
Capability inventory: The skill is capable of generating executable C code and suggesting the use of system-level static analysis tools based on processed requirements.
Sanitization: Absent. There is no defined process for validating or sanitizing the input hardware specifications before they are used to influence the agent's output.

embedded-systems