The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through data ingested from the Linear platform.\n
Ingestion points: Commands such as lin issue get, lin project get, and lin document get (described in SKILL.md) retrieve external markdown content and comments into the agent's context.\n
Boundary markers: There are no explicit instructions or delimiters provided to the agent to distinguish between its primary instructions and potential commands embedded in the fetched Linear content.\n
Capability inventory: The agent can use the lin tool to perform actions in the workspace and has access to local files through the Read, Grep, and Glob tools.\n
Sanitization: The skill does not implement sanitization or filtering for the markdown content retrieved from Linear.\n- [COMMAND_EXECUTION]: The skill facilitates interaction with the Linear platform through a dedicated CLI binary.\n
Evidence: The allowed-tools section in SKILL.md permits Bash(lin *) commands, enabling the agent to manage issues and projects.\n
Note: A raw GraphQL escape hatch is provided via lin api query for authenticated API access.\n- [DATA_EXFILTRATION]: The skill allows the agent to upload local files to Linear's CDN.\n
Evidence: The lin file upload command and the --file flag in lin issue comment new facilitate the transmission of local data to a remote service.\n
Context: Linear is recognized as a well-known service, and this functionality is a core component of the intended issue-tracking workflow.\n- [EXTERNAL_DOWNLOADS]: The skill provides capabilities to download assets from Linear's infrastructure.\n
Evidence: The lin file download command fetches content from the uploads.linear.app domain.\n
Context: This interaction targets a well-known service and is a legitimate part of the skill's functionality.

lin