---
title: "Harnie Use Cases"
description: "Cross-agent work continuity with evidence-backed handoffs."
---
## Cross-agent continuity

Start an investigation in Pi, continue in OpenCode without re-explaining context:

```bash
# After finishing work in Pi
harnie import pi ~/.pi/sessions/abc123.jsonl
harnie handoff work:pi:abc123 --to opencode
# Paste the Markdown output into a fresh OpenCode session
```

## Harness independence

Your work persists when switching coding agents. The engineering context is separate from whichever harness produced it.

## Evidence-backed state

Every derived claim — goal, findings, next steps — links back to observed events. No opaque LLM summaries without provenance.

```bash
harnie show work:pi:cfef1a72-…
```

## Session archaeology

Inspect what an agent actually did versus what it claimed:

```bash
harnie import codex session.jsonl
harnie show work:codex:session-id
```

## Team handoffs

Export structured work-state for a teammate starting in a different agent:

```bash
harnie import opencode teammate-session.jsonl
harnie handoff work:opencode:xyz --to codex
```

## Pairing with Root

Use Root to install and manage your agent CLIs, then Harnie to carry work between them:

```bash
root install opencode
# ... work in OpenCode ...
harnie import opencode session.jsonl
harnie handoff work:opencode:abc --to pi
```