The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from a local file to tailor its marketing recommendations, creating a potential surface for indirect injection if that file's content is untrusted.
Ingestion points: The skill is instructed to read .agents/product-context.md (referenced in SKILL.md and PSYCHOLOGY.md) to determine the marketing context.
Boundary markers: No explicit delimiters or instructions are used to separate the ingested data from the agent's internal logic or to warn the agent about potentially malicious instructions within the context file.
Capability inventory: This skill primarily generates text-based suggestions and provides links to other skills. It does not perform subprocess calls, file writes, or network operations directly.
Sanitization: There is no evidence of sanitization, escaping, or validation of the content read from the product context file.
[SAFE]: The skill actively promotes ethical and lawful marketing practices through extensive documentation of official rules.
It includes references to Google Search spam policies regarding programmatic and parasite SEO in IDEAS.md.
It documents FTC disclosure requirements for influencers and reviews, as well as legal guidelines for pricing claims.
It cites GDPR requirements regarding opt-out models and consent in PSYCHOLOGY.md.
It identifies and discourages the use of artificial engagement tactics like engagement pods that violate platform rules.

growth-ideas