Run an ent task and read its log

ent is an Emacs Lisp build tool. A project defines tasks in a .ent.el file at its root (each task's :action is a shell command or an elisp function, with optional :deps), and running a task executes it — plus all its dependencies — reporting output to the *ent-log* buffer. This skill runs a task for the agent's current project directory in the running Emacs server (via emacsclient), waits for it to complete, and returns the log so you can process it (e.g. diagnose a failed step).

This skill is not just for explicit user requests. Use it proactively during coding tasks whenever the project uses ent and you need to run or verify a build step. Before running a build, test, lint, or cleanup command yourself, check whether the current directory or its parents contain a .ent.el project file (e.g. via ls or the existence of the file). If it does, the ent tasks are the project's canonical way to run those steps — use this skill instead of ad-hoc shell commands, and report the results back from the log.

First, locate agent-skill-ent.el, which lives alongside this skill file at skills/ent/agent-skill-ent.el in the emacs-skills plugin directory.

Run a task

agent-skill-ent-run takes the task name and the project directory. Use the agent's current working directory as :dir unless the user says otherwise (get it with pwd if unsure).