The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes content from untrusted external sources, which creates a surface for potential indirect prompt injection attacks. 1. Ingestion points: The skill processes data from external URLs, search engine results page (SERP) snapshots, AI prompt logs from platforms like ChatGPT and Claude, and customer support transcripts during keyword and SEO audits (referenced in reference/keyword-research.md and reference/seo-audit.md). 2. Boundary markers: Absent; there are no explicit instructions or delimiters provided to separate untrusted data from the agent's internal instruction context. 3. Capability inventory: The skill maintains the ability to write analysis logs to the file system (.agents/growth.md) and orchestrate actions across other functional agents such as 'Artisan', 'Bolt', and 'Pulse'. 4. Sanitization: Absent; the instructions do not specify methods for sanitizing, escaping, or filtering ingested data before it is processed by the AI model.
[EXTERNAL_DOWNLOADS]: The skill references and fetches documentation, benchmarks, and configuration guidelines from well-known technology organizations and reputable industry sources, including Google, Princeton University, and established marketing research entities like Ahrefs and the Baymard Institute.

growth