The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires access to powerful system tools including bash, run_command, write_to_file, and replace_file_content. These tools are leveraged to verify tracking implementations on live sites, run audits, and modify project source code to insert analytics tags as described in SKILL.md and references/tracking.md.
[INDIRECT_PROMPT_INJECTION]: The skill performs detailed audits of external website funnels, session recordings, and user commitment paths, creating an ingestion point for potentially untrusted data that could influence agent behavior.
Ingestion points: External website HTML, signup funnel field data, and session replay recordings analyzed in references/cro.md and references/tracking.md.
Boundary markers: The procedures do not define explicit delimiters or instructions to ignore embedded prompts when analyzing data from external site audits.
Capability inventory: The agent has bash, run_command, and broad file system write access across all modes.
Sanitization: No specific sanitization or filtering is mentioned for content ingested from audited web pages before it is processed by the agent logic.

analytics