linear-sdk-scripting
Fail
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: HIGHPERSISTENCECREDENTIALS_UNSAFEDYNAMIC_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PERSISTENCE]: The skill provides instructions and commands to append environment variable exports (specifically
LINEAR_API_KEY) to shell configuration files including~/.zshrc,~/.bashrc, and~/.config/fish/config.fish. This allows sensitive credentials to persist across system sessions. - [CREDENTIALS_UNSAFE]: The skill facilitates the storage of sensitive Linear API keys in plain text within global shell profile files.
- [DYNAMIC_EXECUTION]: The skill's core workflow involves writing JavaScript code to a temporary file (
task.mjs) in a cache directory and executing it using the Node.js runtime based on task requirements. - [COMMAND_EXECUTION]: The execution pattern involves sourcing multiple shell profiles (
~/.zshenv,~/.zshrc,~/.zprofile,~/.profile,~/.bashrc,~/.bash_profile) using thesourcecommand. This executes the contents of these files within the agent's environment, which can trigger unintended side effects or execute arbitrary commands if those files contain complex scripts. - [EXTERNAL_DOWNLOADS]: The skill downloads the official
@linear/sdkpackage from the NPM registry. As Linear is a well-known service, this reference is documented neutrally as standard dependency acquisition. - [INDIRECT_PROMPT_INJECTION]: The skill acts as an ingestion point for untrusted data from the Linear API (e.g., issue titles, descriptions, and comments).
- Ingestion points: Data is fetched via SDK methods like
linear.issues(),linear.issue(), andlinear.searchIssues()as described inSKILL.mdandreferences/recipes.md. - Boundary markers: None. The skill does not implement delimiters or safety instructions when processing or displaying data retrieved from Linear.
- Capability inventory: The skill has the capability to write files, execute Node.js scripts, perform network operations via the Linear SDK, and modify system shell configuration files.
- Sanitization: There is no evidence of sanitization, escaping, or validation of the content retrieved from Linear before it is logged to the console or used in subsequent logic.
Recommendations
- AI detected serious security threats
Audit Metadata