The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were identified. The skill functions as a specialized measurement consultant for developer relations, generating Markdown tracking plans based on user input.\n- [SAFE]: The skill does not incorporate any executable scripts, remote code downloads, or third-party package dependencies, eliminating risks associated with command execution or supply chain compromise.\n- [SAFE]: The use of a local file (devrel-context.md) for session memory is a standard practice for AI agent state management and does not involve unauthorized data access or persistence mechanisms.\n- [SAFE]: The skill includes explicit instructions to protect user privacy by banning PII (Personally Identifiable Information) from tracking properties and requiring consent before publishing community leaderboards.\n- [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests user requirements to build event taxonomies, but it lacks the capabilities to be exploited. Evidence chain: 1. Ingestion points: User responses to the Interview section in SKILL.md. 2. Boundary markers: Absent. 3. Capability inventory: No tool access (no network operations, shell command execution, or system file system writes). 4. Sanitization: Instructions in the Event Taxonomy section of SKILL.md to exclude PII.

devrel-analytics