The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the retrieval and processing of data from the Linear platform (such as issue titles, descriptions, and comments). This data, being externally controlled, presents a potential vector for indirect prompt injection if it contains instructions intended to manipulate the agent's logic.
Ingestion points: Content is ingested through commands like linear issue list, linear issue comment add, and raw GraphQL queries using linear api.
Boundary markers: The instructions do not specify the use of delimiters or clear separation between system instructions and data retrieved from the API.
Capability inventory: The skill possesses the capability to execute shell commands (linear), make network requests (curl), and manipulate the file system (cat, grep), which are potential targets for an injection attack.
Sanitization: The skill provides best practices for preventing shell escaping issues when sending data (by using file-based flags), but does not define sanitization procedures for data received from the Linear API.
[COMMAND_EXECUTION]: The skill relies on the execution of the linear CLI and various system utilities (curl, jq, grep). These tools are essential for the skill's intended functionality but provide a broad range of interaction with the local system and the Linear API.

linear-cli