sales-sequences
Pass
Audited by Gen Agent Trust Hub on Oct 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from sources such as fetched web pages, CRM exports, and user-pasted notes.
- Ingestion points: External data is ingested during Step 1 (Read the request) and Step 3 (Draft the emails) from fetched pages and CRM data.
- Boundary markers: The skill implements a 'Golden Rule' and a mandatory 'Notice line' output format to explicitly ignore any instructions found within the processed data.
- Capability inventory: The skill executes a local Python script for logic and utilizes the
web_fetchtool for data gathering. - Sanitization: The instructions mandate the use of
[User to add: ...]placeholders for missing facts and explicitly command the agent to refuse any actions requested within the data. - [COMMAND_EXECUTION]: The skill executes a local script
scripts/sequence_days.pyto calculate email send days and perform mechanical checks on drafts. The script is provided as part of the skill, uses only the Python standard library, and performs no network or file-writing operations.
Audit Metadata