The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructions are focused entirely on the PRD generation workflow. There are no attempts to override agent behavior, bypass safety filters, or ignore previous instructions.\n- [DATA_EXFILTRATION]: The workflow does not access sensitive local file paths (such as SSH keys or AWS credentials) and does not contain hardcoded API keys or secrets. No unauthorized network operations were detected.\n- [OBFUSCATION]: All content, including templates and examples, is provided in clear, human-readable Markdown. No Base64, zero-width characters, or homoglyph-based obfuscation techniques were identified.\n- [REMOTE_CODE_EXECUTION]: While the skill orchestrates several other PM-related skills, it does so using local file paths and does not download or execute scripts from external or untrusted sources.\n- [PRIVILEGE_ESCALATION]: The skill does not utilize commands for elevated system privileges or modifying file permissions (e.g., sudo or chmod).\n- [PERSISTENCE]: No mechanisms for establishing persistence, such as modifying startup scripts or crontabs, are present in the skill files.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided discovery notes and research data to populate a template. It lacks the system-level capabilities (e.g., network access, arbitrary file writing) necessary to execute an attack if malicious instructions were present in the processed data.

prd-development