# Opencode

> Delegate tasks or get a second opinion from OpenCode CLI. Infers mode (delegation vs validation) from prompt context. Detects available models at runtime via the CLI. Use when cross-validating code, plans, or architecture with OpenCode, or offloading work to it. Do not use when already running inside OpenCode CLI or for tasks that do not benefit from cross-model input.

- Skill: `darshitpp/opencode` (Agent Skill, multi-file: 4 files)
- Install (CLI): `npx skillmds@latest add darshitpp/opencode`
- Raw SKILL.md: https://api.skillmd.com/api/skills/darshitpp/opencode/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: darshitpp (https://skillmd.com/u/darshitpp)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/darshitpp/opencode

---


# Cross-Agent Task Runner — OpenCode CLI

## Procedures

**Step 1: Self-Call Detection**

1. Check for `OPENCODE` environment variable, or `opencode` in process ancestry.
2. If detected, stop and inform the user:
   "Cannot invoke OpenCode from within OpenCode. Use a different target CLI."

**Step 2: Execute Shared Procedure**

1. Read `references/shared-procedure.md` for the core workflow
   (mode inference, context gathering, prompt construction, result presentation).
2. Read `references/cli-opencode.md` for OpenCode-specific flags, model selection
   heuristics, and version compatibility matrix.
3. Follow the shared procedure using the OpenCode-specific details.