claude-code-plugin-ref…
Explain plugin, skill, command, agent, and hook mechanics used here. Use when authoring or debugging plugins. Do not use for ops; use night-market-operations.
Delegates tasks to OpenCode (opencode run) via delegation-core. Use when delegation-core selects OpenCode or a provider-agnostic harness is wanted.
$ npx -y skills add athola/claude-night-market --skill opencode-delegation --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/opencode-delegationContext preview
The summary Claude sees to decide when to auto-load this skill.
Delegates tasks to OpenCode (opencode run) via delegation-core. Use when delegation-core selects OpenCode or a provider-agnostic harness is wanted.
name: opencode-delegation description: Delegates tasks to OpenCode (opencode run) via delegation-core. Use when delegation-core selects OpenCode or a provider-agnostic harness is wanted. alwaysApply: false category: delegation-implementation tags: - opencode - cli - delegation - provider-agnostic dependencies: - delegation-core tools: - opencode - delegation_executor.py usage_patterns: - opencode-cli-integration - provider-agnostic-routing complexity: intermediate model_hint: standard estimated_tokens: 550 references: - delegation-core/shared-shell-execution.md
OpenCode is an open-source, provider-agnostic terminal agent published as npm `opencode-ai`. Its scripting entry point is `opencode run`, which takes the prompt as positional arguments.
Being provider-agnostic makes it the most flexible entry in the registry: the model behind it is whatever the local OpenCode config selects, including a local one.
OpenCode already has it configured
what is present
to OpenCode's own config, not to the delegation call
npm install -g opencode-ai@latest
Homebrew, pnpm, bun, scoop and chocolatey are also documented upstream.
Credentials resolve per provider through environment variables such as `ANTHROPIC_API_KEY` or `OPENAI_API_KEY`, through a `.env` file, or through `opencode auth login`. Stored credentials are written to `~/.local/share/opencode/auth.json`, and the config file supports `{env:VARIABLE_NAME}` interpolation.
Verification runs `opencode auth list`, because there is no single variable to check when the provider set is open-ended.
uv run python scripts/delegation_executor.py opencode "Find the bug" \ --files src/
make -C plugins/conjure delegate-opencode PROMPT='Find the bug'
opencode run "Explain the use of context in this package"
`opencode` declares the `code_execution` strength and ranks last among the network providers in the candidate order, ahead only of the local `glimmer` service. It declares no model ids, since model choice belongs to OpenCode's own configuration.
| Property | Value | |----------|-------| | Binary | `opencode` | | Package | `opencode-ai` | | Headless form | `opencode run <prompt>` | | Prompt delivery | positional | | Version probe | `opencode --version` | | Auth probe | `opencode auth list` |
`opencode run --file <path>` attaches a file upstream, but this service inlines file context into the prompt instead, which keeps one delivery path across every provider that lacks an `@path` syntax.
`opencode serve` starts a headless HTTP server that `opencode run --attach <url>` can target, and it honors `OPENCODE_SERVER_PASSWORD`. That path is not modeled here: the registry spawns a process per delegation rather than holding a server.
An output-format flag is documented for the `session list`, `export` and `db` subcommands, but not confirmed for `run`, so none is declared.
and no prompt flag before it
`--file`
A plugin marketplace for Claude Code. Install only the plugins you need to run git workflows, code review, spec-driven development, and autonomous agents from inside your Claude Code session.
Explain plugin, skill, command, agent, and hook mechanics used here. Use when authoring or debugging plugins. Do not use for ops; use night-market-operations.
States load-bearing decisions, invariants, and weak points. Use when judging a design change. Do not use for gating; use night-market-change-control.
Rebuild the dev environment: uv, Python tiers, pins, traps. Use when onboarding or toolchain breaks. Do not use for daily commands; use night-market-operations.
Classify, gate, and review changes. Use when landing a PR, releasing, or amending rules. Do not use for failure triage; use night-market-debugging-playbook.
Search and record project memory (Discussions, journal, ADRs). Use before re-investigating anything. Do not use for settled battles; see failure-archaeology.
Bind loop 'done' to unfakeable gates. Use to harden egregore/herald loops or promote completion_integrity. Not for QA gates; use night-market-validation-and-qa.