The Agent Skills Directory

[PROMPT_INJECTION]: No malicious instruction overrides or safety bypass attempts were detected. The skill uses instructional markers like 'IMPORTANT' and 'CRITICAL' strictly within the context of ensuring documentation quality and adherence to ISO standards.
[DATA_EXFILTRATION]: There are no network operations, such as curl or wget, or patterns indicating the unauthorized transmission of data. The skill writes document artifacts to local project directories (e.g., docs/).
[CREDENTIALS_UNSAFE]: No hardcoded secrets, API keys, or private tokens were found. Template examples for configuration (e.g., JWT_SECRET) are appropriately documented as placeholders for user implementation.
[REMOTE_CODE_EXECUTION]: The skill explicitly states 'Do not write code' and includes boundaries that route implementation tasks to other agents. No package installations or remote script execution patterns are present.
[INDIRECT_PROMPT_INJECTION]: The skill processes project descriptions and requirements as input. While this creates an ingestion surface for potentially untrusted data, the skill employs extensive quality gates, strict template requirements, and validation steps to ensure outputs remain focused on technical specifications rather than executing instructions embedded in the input data.
[COMMAND_EXECUTION]: No shell commands or subprocess execution patterns were identified. The workflow is restricted to text processing and markdown generation.

scribe