agent-harness

When you give the lead agent a work item URL, execution follows this pipeline. If the backlog platform is None, skip the work item stage. If the repository platform is None, skip the pull request stage.

lead
                   ↓
          parse work item via userstory skill
                   ↓
               plan-propose
         proposal + specs + tasks
                   ↓
              [confirm with user]
                   ↓
   wave of subagents (*-engineer, per-tier model)
  each implements its assigned tasks → returns result → lead commits group
                   ↓
        verify (tests, build, lint as needed)
                   ↓
     lead (ship mode, if configured)
   commit → push → pull request → feedback loop

Load the platform userstory skill (installed as pc-userstory, from the variant matching your backlog platform)
Run /plan-propose to produce proposal.md, specs and tasks.md
Confirm with the user before implementation
Run /plan-apply to orchestrate implementation in waves
Each wave spawns engineers in parallel (custom *-engineer specialists, each carrying its own tier model), capped at agents.maxConcurrent
Each subagent receives its task IDs in its prompt, loads relevant abilities, implements, and returns; the lead commits each group
Verify with tests, build and lint according to task scope
Ship or update the pull request via the lead flow

Unattended runs

Keep a shell task as the final gate

For unattended loop runs, end with a concrete shell verification task as the exit-code gate. Inside /plan-goal, /plan-apply step 10 runs lint, typecheck, test, and build commands and is the checks gate; /repo-verify then writes verification-plan.md, the journey the evidence workflow replays in headless Chromium once the pull request merges. The shell task independently proves the critical commands passed before any commit or pull request action.

Watching a run

Agents run as native OpenCode subagents in parallel waves: no external plugin, no git worktrees. The lead's Todo pane is the live board, and the pc-subagent-monitor plugin mirrors state to .opencode/harness-run.json. Navigate into any running subagent with ctrl+x ↓ then ←/→.

The pipeline

Unattended runs

Watching a run

On this page