The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection because it ingests and processes untrusted scraped content from external business websites to generate project code and marketing copy.
Ingestion points: Business data, content inventories, and audit reports are read from the ~/prospect-pipeline/prospects/ directory.
Boundary markers: The instructions do not define explicit delimiters or 'ignore embedded instructions' warnings for the data being processed.
Capability inventory: The agent has the ability to execute shell commands (npm, curl) and perform local file system writes.
Sanitization: There are no mandated sanitization or validation steps for external content before it is interpolated into Astro component templates.
[COMMAND_EXECUTION]: The skill instructs the agent to execute standard shell commands using npm, npx, and curl for project initialization, package installation, and asset acquisition. These operations are routine for website development but represent a capability surface if an injection occurs.
[EXTERNAL_DOWNLOADS]: Asset and library downloads are performed from well-known services. The skill uses curl to fetch stock photography from Unsplash and dynamically loads the axe-core accessibility engine from the Cloudflare CDN for automated QA testing.

local-business-rebuild