Harn Decide

---
name: harn-decide
description: "File a decision into the docket instead of blocking on a human: search precedent, file it, and proceed on a reversible default; or pick up and resolve an open decision with cited evidence. Use whenever you're about to ask a human a decision-shaped question you could resolve yourself."
---
<!-- harnery:generated harn-decide v=75aa315d; machine-owned; regenerated by `harn init`, removed by `harn deinit`. Edit the harnery template, not this file. -->

The decision docket is a persistent queue for decisions you would otherwise
route to a human. It's built on the `harn decision` engine. This skill is the
mechanics: file, find precedent, claim, and resolve with evidence. *When* a
decision needs a human at all (versus one you settle yourself) is host policy;
if this project defines that rubric, follow it.

## Modes

- **A decision you're facing (default)** → capture: record it and proceed.
- **`resolve <id>`** → pick up an open decision, research it, resolve it.
- **`review`** → surface resolved-but-unreviewed decisions for a human to skim.

## Capture (default)

1. **Check precedent first.** `harn decision search "<key terms>"`. If a resolved
   decision already answers this, cite it; don't re-litigate.
2. **File it** when the choice has a second consumer (a human will want to see it,
   or a future agent will face it again) or reversal is expensive. Skip the
   docket for pure local mechanics (a variable name, one of two equivalent idioms).

   ```bash
   harn decision file "<the decision as a clear question>" \
     --context "why it matters / what's blocked / the options you see" \
     --default-taken "<what you're proceeding with>"
   ```

   For a decision with real substance, write a brief to a file and pass
   `--brief <path>` so the reviewer sees options + evidence, not a cold prompt.
3. **Proceed on your default.** Filing does not mean blocking. Note the id in your
   reply and keep working.

## Resolve (`resolve <id>`)

```bash
harn decision show <id>     # read the question + context + any brief
harn decision claim <id>    # mark it deliberating (claimed by you)
```

Research it for real: run the queries, read the files, compute the costs. Then
resolve with **cited evidence** (the engine rejects an evidence-free resolution):

```bash
harn decision resolve <id> \
  --recommendation "<the call>" \
  --evidence "<a fact you established: a query run, a file read, a cost computed>" \
  --evidence "<another>" \
  --reversal-cost "<cost to undo if wrong>" \
  --wrong-if "<what would make this wrong>" \
  --revisit-when "<trigger to revisit>"
```

If the decision is genuinely hard or contested, escalate to a council
(`harn council create "<objective>"`) and link it rather than forcing a thin
single-agent resolution.

## Review (`review`)

```bash
harn decision list --status resolved
```

For each one a human reacts to, record the verdict so triage self-corrects:

```bash
harn decision review <id> --verdict ratified              # agreed; no action
harn decision review <id> --verdict overridden --note "…" # disagreed
harn decision review <id> --verdict wrong-tier-low        # "didn't need to see this"
harn decision review <id> --verdict wrong-tier-high       # "should have seen it sooner"
```