opencode-bridgelisted
Install: claude install-skill darkstar1227/bridge
# OpenCode Bridge
**Announce at start:** "I'm using the opencode-bridge skill to delegate this task to OpenCode."
## Step 0: Check whether this should be a standalone call
This skill only does one thing: hand a task to OpenCode and return a handoff report. It has **no spec-compliance review, no code-quality review, and no test enforcement** — `test_results` and `self_review_notes` in the report are literally "not reported by OpenCode". Those review gates only exist in `superpowers:subagent-driven-development` (or `superpowers:executing-plans`), which can use this skill as their implementer step instead of a Claude subagent.
Before running Step 1, check: is this invocation already happening as the implementer step inside an active `/subagent-driven-development` or `/executing-plans` run for this task (i.e. that skill's process is already driving this task, and it will handle spec/quality review after this skill returns)?
- **Yes** → proceed directly to Step 1, no prompt needed.
- **No** (this is a bare/ad-hoc request to "delegate this to opencode" for a single task, with no surrounding plan-execution flow tracking review gates) → stop and ask the user via AskUserQuestion:
- **Question**: "opencode-bridge alone skips spec-compliance and code-quality review. Run this through `bridge:opencode-subagent-driven-development` instead (recommended — same review gates as subagent-driven-development, with OpenCode as the implementer), or continue standalone?"
- Options: "Use bri