The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external, untrusted resume files, which represents a potential surface for indirect prompt injection. However, it includes explicit instructions to the AI to treat all parsed content strictly as data and to ignore any instructions found within the documents.
Ingestion points: User-provided resume files are read by resume_parser.py and then processed as Markdown by the agent.
Boundary markers: The skill uses a clear instruction to the agent to treat document content as data-only.
Capability inventory: The skill can read local files and perform network requests to the vendor's API endpoints.
Sanitization: The agent is directed to extract specific fields into a structured JSON schema, reducing the likelihood of executing malicious content.
[DATA_EXFILTRATION]: The skill sends resume data to somark.cn and somark.ai. These are official domains associated with the skill's author (somarkai) and are required for the core parsing functionality. No unauthorized data exfiltration was detected.
[CREDENTIALS_UNSAFE]: The skill correctly uses environment variables (SOMARK_API_KEY) for authentication and provides instructions to the user on how to set this up safely without exposing the key in chat history.

resume-parser