refactor(opencode): align output with current permissions

Stop emitting deprecated tools config for OpenCode permission modes. Update the OpenCode schema notes and add a regression that the current CE plugin outputs skills plus subagents, not command files.

docs/specs/opencode.md

@@ -26,10 +26,11 @@ https://opencode.ai/config.json
 ## Core config keys
 
 - `model` and `small_model` set the primary and lightweight models; `provider` configures provider options.
-- `tools` is still supported but deprecated; permissions are now the canonical control surface.
+- `tools` is still supported but deprecated as of OpenCode v1.1.1; permissions are now the canonical control surface.
 - `permission` controls tool approvals and can be configured globally or per tool, including pattern-based rules.
 - `mcp`, `instructions`, `disabled_providers`, `enabled_providers`, and `plugin` are supported config sections.
 - `plugin` can list npm packages to load at startup.
+- `skills.paths` and `skills.urls` can add extra skill discovery locations, but CE should not depend on them until the layout is smoke-tested locally with OpenCode.
 
 ## Tools
 
@@ -45,27 +46,31 @@ https://opencode.ai/config.json
 ## Agents
 
 - Agents can be configured in `opencode.json` or as markdown files in `~/.config/opencode/agents/` or `.opencode/agents/`.
-- Agent config supports `mode`, `model`, `temperature`, `tools`, and `permission`, and agent configs override global settings.
+- Agent config supports `mode`, `model`, `variant`, `temperature`, `top_p`, `hidden`, `steps`, `options`, `permission`, and other schema fields. `tools` still exists but is deprecated.
 - `mode` can be `primary`, `subagent`, or `all`; omitted mode defaults to `all`.
+- `hidden: true` hides subagents from the `@` autocomplete menu.
+- `permission.task` controls which subagents an agent may invoke.
 - Model IDs use the `provider/model-id` format.
 
 ## Skills
 
 - Skills are reusable `SKILL.md` definitions loaded on demand through OpenCode's native `skill` tool.
-- OpenCode searches direct child skill directories only:
+- OpenCode searches direct child skill directories in its built-in roots:
   - `.opencode/skills/<name>/SKILL.md`
   - `~/.config/opencode/skills/<name>/SKILL.md`
   - `.claude/skills/<name>/SKILL.md`
   - `~/.claude/skills/<name>/SKILL.md`
   - `.agents/skills/<name>/SKILL.md`
   - `~/.agents/skills/<name>/SKILL.md`
+- The config schema also exposes `skills.paths` and `skills.urls` for extra skill sources. Do not switch CE to those until tested against a local OpenCode install; direct `~/.config/opencode/skills/<name>/SKILL.md` remains the stable writer shape.
 - Skill frontmatter recognizes `name`, `description`, `license`, `compatibility`, and `metadata`; unknown fields are ignored.
 - Skill names must be lowercase alphanumeric with single hyphen separators and must match the directory name.
 
 ## Commands
 
 - Commands can be configured in `opencode.json` or as Markdown files in `~/.config/opencode/commands/` or `.opencode/commands/`.
-- Markdown command frontmatter can include fields such as `description`, `agent`, and `model`; the body becomes the prompt template.
+- Markdown command frontmatter can include fields such as `description`, `agent`, `model`, and `subtask`; the body becomes the prompt template.
+- If a command targets an agent whose mode is `subagent`, OpenCode invokes it as a subagent by default. `subtask: true` can force subagent invocation.
 
 ## Plugins and events
 
@@ -79,8 +84,9 @@ https://opencode.ai/config.json
 - The current CE writer shape is still appropriate in April 2026:
   - `~/.config/opencode/opencode.json`
   - `~/.config/opencode/agents/*.md`
-  - `~/.config/opencode/commands/*.md`
+  - `~/.config/opencode/commands/*.md` only when a source plugin ships commands
   - `~/.config/opencode/plugins/*.ts`
   - `~/.config/opencode/skills/*/SKILL.md`
 - OpenCode's plugin system is useful for JS/TS hooks and custom tools, but current docs do not describe a native marketplace command that consumes CE's `.claude-plugin/marketplace.json` and installs the full skills/agents/commands payload.
 - Keep the custom Bun writer until OpenCode documents a native distribution path for packaged skills and agents.
+- The `compound-engineering` plugin currently emits skills and subagent Markdown files for OpenCode. It should not emit deprecated `tools` config; permission config is enough for non-default permission modes.

src/converters/claude-to-opencode.ts

@@ -362,11 +362,6 @@ function applyPermissions(
   }
 
   const permission: Record<string, "allow" | "deny" | Record<string, "allow" | "deny">> = {}
-  const tools: Record<string, boolean> = {}
-
-  for (const tool of sourceTools) {
-    tools[tool] = mode === "broad" ? true : enabled.has(tool)
-  }
 
   if (mode === "broad") {
     for (const tool of sourceTools) {
@@ -415,7 +410,6 @@ function applyPermissions(
   }
 
   config.permission = permission
-  config.tools = tools
 }
 
 function normalizeTool(raw: string): string | null {

src/types/opencode.ts

@@ -4,19 +4,37 @@ export type OpenCodeConfig = {
   $schema?: string
   model?: string
   default_agent?: string
+  /** @deprecated OpenCode v1.1.1+ uses permission as the canonical control surface. */
   tools?: Record<string, boolean>
   permission?: Record<string, OpenCodePermission | Record<string, OpenCodePermission>>
   agent?: Record<string, OpenCodeAgentConfig>
   mcp?: Record<string, OpenCodeMcpServer>
+  skills?: OpenCodeSkillsConfig
 }
 
 export type OpenCodeAgentConfig = {
   description?: string
-  mode?: "primary" | "subagent"
+  mode?: "primary" | "subagent" | "all"
   model?: string
+  variant?: string
   temperature?: number
+  top_p?: number
+  prompt?: string
+  disable?: boolean
+  hidden?: boolean
+  color?: string
+  steps?: number
+  /** @deprecated Use steps instead. */
+  maxSteps?: number
+  options?: Record<string, unknown>
+  /** @deprecated OpenCode v1.1.1+ uses permission as the canonical control surface. */
   tools?: Record<string, boolean>
-  permission?: Record<string, OpenCodePermission>
+  permission?: Record<string, OpenCodePermission | Record<string, OpenCodePermission>>
+}
+
+export type OpenCodeSkillsConfig = {
+  paths?: string[]
+  urls?: string[]
 }
 
 export type OpenCodeMcpServer = {

tests/converter.test.ts

@@ -15,6 +15,24 @@ const compoundEngineeringRoot = path.join(
 )
 
 describe("convertClaudeToOpenCode", () => {
+  test("current compound-engineering output is skills and subagents, not commands", async () => {
+    const plugin = await loadClaudePlugin(compoundEngineeringRoot)
+    const bundle = convertClaudeToOpenCode(plugin, {
+      agentMode: "subagent",
+      inferTemperature: true,
+      permissions: "none",
+    })
+
+    expect(bundle.agents.length).toBeGreaterThan(0)
+    expect(bundle.skillDirs.length).toBeGreaterThan(0)
+    expect(bundle.commandFiles).toHaveLength(0)
+    expect(bundle.plugins).toHaveLength(0)
+    expect(bundle.config.tools).toBeUndefined()
+
+    const parsedAgents = bundle.agents.map((agent) => parseFrontmatter(agent.content))
+    expect(parsedAgents.every((agent) => agent.data.mode === "subagent")).toBe(true)
+  })
+
   test("from-command mode: map allowedTools to global permission block", async () => {
     const plugin = await loadClaudePlugin(fixtureRoot)
     const bundle = convertClaudeToOpenCode(plugin, {
@@ -24,6 +42,7 @@ describe("convertClaudeToOpenCode", () => {
     })
 
     expect(bundle.config.command).toBeUndefined()
+    expect(bundle.config.tools).toBeUndefined()
     expect(bundle.commandFiles.find((f) => f.name === "workflows:review")).toBeDefined()
     expect(bundle.commandFiles.find((f) => f.name === "plan_review")).toBeDefined()
 
@@ -275,6 +294,7 @@ describe("convertClaudeToOpenCode", () => {
       inferTemperature: false,
       permissions: "broad",
     })
+    expect(broadBundle.config.tools).toBeUndefined()
     expect(broadBundle.config.permission).toEqual({
       read: "allow",
       write: "allow",