delegate

Run altana delegate to send a task to a single named harness preset and return the structured result.

What it does:

1. Invokes altana delegate <harness> "<task>" with any supplied flags via Bash.
2. For long-running agents (timeout_s > 300 or when --write is present) runs the command in the background and streams the result when complete.
3. Parses the JSON object from stdout.
4. Surfaces a human-readable summary: harness name, status, duration, response text (trimmed to key sections), and log path for debugging.
5. On non-done status, explains the failure mode and points to the log file.

Arguments:

• <harness> — name of a configured harness preset (see altana list to enumerate presets).
• <task> — the prompt string to delegate. Wrap in quotes if it contains spaces.
• --prompt <template> — optional named prompt template from [prompt.<name>] in config; wraps the task before dispatch.
• --write — grant the agent write access to the working directory inside the container (awman executor only).

JSON result fields:

Field	Type	Description
harness	string	Harness name used
status	string	done | timeout | crash | missing_sentinel
duration_s	float	Wall-clock seconds
log_path	string	Full subprocess log path
response	string or null	Agent response (sentinel stripped); null on non-done

Status meanings:

• done — agent exited 0 and emitted the completion sentinel; response is populated.
• missing_sentinel — agent exited 0 but did not emit the sentinel; check log_path.
• crash — agent exited non-zero or failed to spawn; check log_path.
• timeout — agent exceeded the configured timeout_s; check log_path for partial output.

If altana is not installed:

Install altana (see its README — it is a Zig CLI built with zig 0.16). After building, place the binary on your PATH or invoke via mise exec -- zig-out/bin/altana.

Examples:

/delegate my-claude "explain the failing test in src/parser_test.zig"
/delegate critique-harness "review src/main.zig" --prompt critique
/delegate coding-agent "add error handling to the fetch function" --write

Task instructions:

Resolve the argument string: the first positional token is <harness>, the remaining text up to the first -- flag is <task>. Reconstruct any --prompt <template> or --write flags from the parsed argument, using space-separated form.

Run: altana delegate "<harness>" "<task>" [flags]

Parse the JSON result. Present:

• A one-line status line: harness: <name> | status: <status> | duration: <duration_s>s
• If status == "done": the response content (show ## Answer, ## Evidence, ## Confidence sections).
• If status != "done": the failure mode explanation and the log_path to inspect.

Load /altana:altana for protocol and config reference before running.