The Agent Skills Directory

[SAFE]: The skill performs business matching logic for startup accelerators. It reads from and writes to specific project paths like './business/' and './portfolio/'. No unauthorized file access, persistence mechanisms, or hardcoded credentials were found.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface. 1. Ingestion points: User responses to profiling questions (AL-1 to AL-6) and internal project files in the './business/' directory. 2. Boundary markers: Absent. 3. Capability inventory: Local file read/write operations restricted to markdown analysis. 4. Sanitization: Absent. The attack surface is restricted to report generation, posing no significant risk of exploitation.
[EXTERNAL_DOWNLOADS]: The skill contains references to official websites for established accelerator programs (e.g., ycombinator.com, techstars.com, 500.co). No automated downloads, remote code execution, or piped command patterns were detected.

accelerator-launchpad