docs-updater
Pass
Audited by Gen Agent Trust Hub on Nov 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is designed to execute arbitrary shell commands defined in repository settings files. For example, the
Drift probessetting insettings/docs.mdallows users to specify commands (e.g.,gh api ...) that the agent runs to verify documentation accuracy. Similarly, theDocs validationsetting executes repository scripts likenode scripts/check-docs.mjs. If these configuration files or scripts are maliciously modified, the agent will execute the attacker-controlled commands. - [DYNAMIC_EXECUTION]: The skill dynamically assembles and executes commands based on string values stored in Markdown files (e.g.,
Workspace status command,Sync with main, andGenerated-config checkinsettings/workspace.md). This allows for runtime code execution where the logic is stored outside of the core skill instructions. - [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest untrusted data from the repository—including source code, git diffs, and existing documentation—and then use that data to generate or update other documentation.
- Ingestion points:
docs/agents/*.mdsettings files,git diffoutput, and the content of all tracked Markdown files. - Boundary markers: The skill lacks explicit boundary markers or instructions to treat ingested data as potentially malicious when interpolating it into prompts.
- Capability inventory: The skill has full access to the shell (bash), git operations, and the ability to write to the file system.
- Sanitization: The skill contains high-level instructions to 'Keep private things private' and 'never invent a feature', but it does not employ technical sanitization or escaping for the external content it processes.
Audit Metadata