The Agent Skills Directory

[SAFE]: The skill implements a strictly instructional workflow for project management and issue tracker coordination without executable scripts or network-based payload delivery.
[COMMAND_EXECUTION]: The skill documentation mentions a command /to-issues <engineering-handoff-issue-url>. This is a platform-specific command meant to be used by the user or agent to trigger other skills, which is standard behavior for inter-skill handoffs and does not represent a shell injection risk.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill reads from local files like SEO-STRATEGY.md. This is legitimate as it is the primary input for the skill's purpose. It does not attempt to access sensitive system files (e.g., .ssh, .env) or exfiltrate data to external domains.
[PROMPT_INJECTION]: No evidence of role-play, DAN-style overrides, or instructions to ignore safety filters were found. The 'IMPORTANT' style instructions are used appropriately for defining workflow boundaries.
[OBFUSCATION]: All files are written in clear Markdown and YAML. No Base64, hex encoding, or zero-width character obfuscation was detected.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data from SEO-STRATEGY.md and research links. While this is an ingestion point for untrusted data, the skill includes strict boundary requirements (e.g., requiring specific status and approval metadata) and focuses on drafting tracker items rather than executing raw data as code, which mitigates injection risks to a 'SAFE' level in this context.

to-seo-issues