autorekt
Pass
Audited by Gen Agent Trust Hub on Nov 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill documentation describes a
--hookparameter for theautorekt watchcommand. This feature allows the user (or an agent following these instructions) to specify a shell command that will be executed by a background daemon whenever a specific market condition is met. While the skill advises obtaining user approval, this creates a mechanism for persistent, conditional command execution. - [INDIRECT_PROMPT_INJECTION]: The skill processes live, third-party market data from various external sources (exchanges, news feeds, prediction markets). This data is ingested into the agent's context to answer questions.
- Ingestion points: Market data, news headlines, and calendar events fetched via
autorekt showandautorekt why(SKILL.md). - Boundary markers: The skill instructs the agent to quote values and timestamps but does not specify strict delimiters for the external text.
- Capability inventory: The skill uses
autorektCLI which can perform network requests and, via thewatchcommand, execute shell hooks. - Sanitization: No explicit sanitization of the fetched text/headlines is mentioned before the agent processes it.
- [EXTERNAL_DOWNLOADS]: The skill requires a binary that is not included. It directs users to download and install the
autorektbinary from GitHub (github.com/autorekt/autorekt) orautorekt.com. Following the trust-scope-rule, these are documented neutrally as they are the official sources for the tool described in the skill.
Audit Metadata