The Agent Skills Directory

[PROMPT_INJECTION]: No evidence of instructions designed to bypass safety filters or override agent behavior was found. The content is strictly instructional and focused on integration development.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill correctly handles sensitive information. It instructs users and agents to use environment variables (process.env) for secrets like CLIENT_SECRET and provides patterns for secure OAuth2 management. No hardcoded credentials or unauthorized network exfiltration patterns were detected.
[OBFUSCATION]: The documentation and code examples are provided in clear, plain-text TypeScript and Markdown. No Base64-encoded payloads, hex escapes, zero-width characters, or homoglyph attacks were identified.
[UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: All referenced dependencies and tools are official components of the Prismatic ecosystem (e.g., @prismatic-io/spectral, @prismatic-io/prism). These are established, well-known services associated with the skill's author.
[INDIRECT_PROMPT_INJECTION]: While the skill teaches the agent how to process external data (webhooks), it includes dedicated sections on security best practices, such as webhook signature verification and schema validation using Ajv, to mitigate risks associated with untrusted data ingestion.
[DYNAMIC_CONTEXT_INJECTION]: No use of shell-command execution markers (e.g., !command) was found in the main SKILL.md or other reference files. The skill uses custom XML-like tags (e.g., <rule>, <connection-code-rules>) solely for structural organization of documentation.

integration-patterns