Run (Standalone Task Execution)

Use the standalone nia run workflow to execute a task without issue, PR, or ticket context. NIA saves task input, summary output, logs, and traces under .nia/work/run/.

Overview

Run is the standalone execution workflow for focused coding tasks. Unlike nia ask, it is intended to make changes, run commands, and produce a workflow summary.

Use Run for:

• Small implementation tasks
• Quick refactors
• Documentation edits
• One-off codebase changes that do not need issue context

How It Works

1. Parse the task and command options.
2. Resolve the configured coding agent, model, role, or custom agent.
3. Validate any --context-file and --context-dir values.
4. Create or reuse .nia/work/run/ and its logs and traces directories.
5. Read the task from the CLI, --task-file, or .nia/work/run/task.md.
6. Write the resolved task to .nia/work/run/task.md.
7. Compose the Run prompt with the run path and context sources.
8. Validate required prompt inputs.
9. Run the selected coding agent unless --print-prompt is specified.
10. Save results in the Run directory.

Task Input

Provide the task as the positional argument:

nia run "Add input validation to the login form"

Or load it from a file:

nia run --task-file task.md

If omitted, NIA reads .nia/work/run/task.md.

Output

Run writes files under .nia/work/run/, including:

• task.md
• summary.md
• logs/
• traces/

Commit Behavior

nia run commits by default when changes are made. You can override that behavior with project or agent configuration.