The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from multiple sources, including user-provided CSV lists, CRM notes, and external web content scraped at runtime.
Ingestion points: Data enters the agent context through the account list (Step 1.1), first-party evidence such as call quotes and CRM notes (Step 1.5), and public URLs processed via the scrape-website function (Step 5).
Boundary markers: While the instructions emphasize identifying "verbatim buyer statements," they do not provide explicit structural delimiters (e.g., XML tags or unique string markers) or specific system instructions to ignore potentially malicious commands embedded within the external content.
Capability inventory: The skill has access to shell command execution through the clay CLI and network access via platform-managed functions like scrape-website and http-api-v2.
Sanitization: The instructions do not describe any sanitization, filtering, or validation of the ingested text before it is interpolated into prompts for hypothesis generation or message drafting.
[COMMAND_EXECUTION]: The skill executes local shell commands using the clay CLI to verify platform readiness and retrieve technical metadata.
Evidence: Instructions include running clay whoami, clay --version, clay routines list, clay routines get, and clay workflows actions schema.
Context: These commands are used for environment setup and discovery of internal Clay functions, which is consistent with the skill's purpose as a platform integration tool for 'clay-run'.

account-priority-buckets