The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a feature to store 'run configs' (metadata for unattended runs) using a custom shell command. When the 'command ' storage type is selected in the local configuration file, the agent executes the specified command with sub-verbs ('put', 'get', 'close') and pipes data to it. This allows the execution of arbitrary commands as defined in the project's configuration.
[DYNAMIC_EXECUTION]: The skill dynamically constructs and executes shell commands for the GitHub CLI ('gh') and uses MCP tools for the Linear backend. Although it attempts to sanitize inputs using 'jq' arguments and manual single-quoting, the generation of shell commands based on external ticket data represents a significant attack surface.
[INDIRECT_PROMPT_INJECTION]: The skill fetches data from external sources including GitHub Issues, Linear, and local markdown files. Ticket titles, bodies, and comments are untrusted inputs that are injected into the agent's context. (1) Ingestion points: GitHub Issues API, Linear API, and local issues directory. (2) Boundary markers: The skill contains natural language instructions for the agent to treat ticket content as data and ignore embedded instructions. (3) Capability inventory: The skill can execute shell commands, write to the local file system, and perform network operations via CLI tools and APIs. (4) Sanitization: The skill uses structured parameter passing via 'jq' for some commands but relies on manual escaping for others.
[EXTERNAL_DOWNLOADS]: The skill relies on external tools and services including the GitHub CLI and Linear MCP server. These are official integrations for well-known services and are used for their intended tracking purposes.

wkc-tracker