GitHub - LiteLLM-Labs/lite-harness: Unified Server for running OpenCode, Claude Code, Codex agents

Deploy agents from any AI coding agent (Claude Code, Codex, OpenCode) to a unified server, for use across your team. Agents are available through CLI, API (Open-code API Spec), and Slack.

ui

usage

Install the skills, then deploy agents from any AI coding agent (Claude Code, Codex, Cursor, etc.):

npx skills add LiteLLM-Labs/lite-harness -g

/deploy-agent

> Deploy a GitHub stargazer outreach agent that DMs new stars on LinkedIn,
  runs every 4 hours on weekdays, requires human approval before each send.

  ✓ Checking capabilities... sandbox=e2b vault=available cron=supported
  ✓ Stored vault key BROWSER_USE_API_KEY
  ✓ Stored vault key LINKEDIN_PROFILE_ID
  ✓ Created agent: agent_abc123 (opencode · claude-sonnet-4-6)
  ✓ Attached vault keys + scheduled 0 */4 * * 1-5 America/Los_Angeles
  ✓ Test run started: run_xyz789
  → Logs: https://lite-harness.onrender.com/api/agents/agent_abc123/runs/run_xyz789/logs
  → UI:   https://lite-harness.onrender.com/agents

Agents run on a cron schedule in an isolated sandbox. Human-in-the-loop approval flows through the Inbox UI.

Supported harnesses: opencode claude-code github-copilot codex — full CLI docs

setup

npx skills add LiteLLM-Labs/lite-harness -g

Then in any AI coding agent (Claude Code, Codex, Cursor, etc.):

Or follow the manual setup guide below.

export MASTER_KEY=$(openssl rand -hex 32)
echo "MASTER_KEY: $MASTER_KEY"

docker run -p 4096:4096 \
  -e LITELLM_API_BASE=https://your-litellm-gateway \
  -e LITELLM_API_KEY=sk-... \
  -e MASTER_KEY="$MASTER_KEY" \
  ghcr.io/litellm-labs/lite-harness:latest

Needs a LiteLLM gateway. Full config: docs/configuration.md.

persistence

By default, sessions (history, model context) are lost when the server restarts. To keep them across restarts, mount persistent storage at the data directory:

Docker:

docker run -p 4096:4096 \
  -v ./data:/home/sandbox/.local/share/lite-harness \
  -e LITELLM_API_BASE=... \
  -e LITELLM_API_KEY=... \
  -e MASTER_KEY=... \
  ghcr.io/litellm-labs/lite-harness:latest

Custom path (e.g. a mounted cloud volume at /mnt/data):

-e DB_PATH=/mnt/data/db.db

On restart the server logs hydrated N session(s) from db and all prior sessions are immediately available.

sandboxing

Set E2B_API_KEY or DAYTONA_API_KEY and agents get an isolated Linux sandbox automatically. Full options (templates, snapshots, vault): docs/configuration.md.

about

We built lite-harness because running opencode and claude-code as separate servers got hard to maintain - multiple services, different API specs, unreliable session management, different inputs for MCP tools and system prompts.

So we wrapped all harnesses in an OpenCode-compatible server and put it in one Dockerfile, giving us one service to maintain, with shared MCP tools, prompts, and session management across all harnesses.

docs

API reference · Architecture · Configuration · CLI · Add a harness

license

MIT