The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from sales call transcripts and notes via the list_findings tool. This creates a surface for indirect prompt injection if a transcript contains malicious instructions designed to alter the agent's behavior during microsite generation.
Ingestion points: list_findings (call transcripts) referenced in references/coaching-agents.md and references/coaching-microsite-sections.md.
Boundary markers: The skill does not utilize specific delimiters or instructions to ignore embedded commands within the ingested findings.
Capability inventory: The skill is capable of generating HTML/JavaScript artifacts and accessing sensitive CRM data through tools like find_crm_records and find_crm_activities.
Sanitization: The instructions do not specify a sanitization process for findings before they are rendered into the Job 2 ('Where the buyer's head is') and Job 5 ('Top risks') sections.
[COMMAND_EXECUTION]: The skill performs authorized tool calls to retrieve business data, including find_crm_records, find_crm_activities, and enrich_company. These operations are consistent with the skill's stated purpose of deal coaching and are managed by the host environment's tool permissions.
[EXTERNAL_DOWNLOADS]: The HTML scaffold in references/html-scaffold.md includes references to Google Fonts (fonts.googleapis.com and fonts.gstatic.com). These are well-known services and do not represent a security risk.

deal-coach