The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches the Snyk CLI from the npm registry using npx to provide the underlying scanning functionality.
[REMOTE_CODE_EXECUTION]: Executes the Snyk command-line tool to perform security analysis and provide the MCP server interface.
[DATA_EXFILTRATION]: Transmits source code, manifest files, and project metadata to Snyk's cloud platform for processing. The skill includes explicit warnings to avoid scanning directories containing secrets or .env files to mitigate data exposure.
[INDIRECT_PROMPT_INJECTION]: The skill creates a surface where the agent processes external data (scan findings and source code) to suggest and apply code fixes, which could be exploited to influence agent behavior.
Ingestion points: Scan output from tools such as snyk_code_scan and snyk_sca_scan (documented in references/tools.md).
Boundary markers: The skill instructions do not specify delimiters to separate external scan findings from the agent's primary instructions.
Capability inventory: The agent is expected to write to the local filesystem to apply suggested remediations as described in SKILL.md.
Sanitization: No specific sanitization or filtering of the content returned by the Snyk API is described.
[PRIVILEGE_ESCALATION]: Includes the snyk_trust tool, which modifies local Snyk configuration to permit scanning of directories that the security tool would otherwise consider untrusted.

snyk-security-scan