opencodekit 0.15.9 → 0.15.11

package/dist/template/.opencode/agent/vision.md

@@ -89,6 +89,7 @@ Load skill(s) → Systematic analysis → Structured findings → Recommendation
 
 - Embed CLI commands in responses (use skills/commands instead)
 - Implement changes directly (delegate to @build)
+- Generate or edit images directly (delegate to @painter)
 - Make assumptions about intent without clarifying
 - Skip skill loading for complex tasks
 - Provide generic "looks good" assessments

package/dist/template/.opencode/command/edit-image.md

@@ -1,8 +1,7 @@
 ---
 description: Edit image with natural language instructions
 argument-hint: "<image-path> <edit-instruction> [--hd] [--preserve-size] [--compare]"
-agent: vision
-model: proxypal/gemini-3-pro-image-preview
+agent: painter
 ---
 
 # Edit Image: $ARGUMENTS

package/dist/template/.opencode/command/generate-icon.md

@@ -1,8 +1,7 @@
 ---
 description: Generate app icons in multiple sizes with platform optimization
 argument-hint: "<description> [--platform=<ios|android|web|all>] [--style=<style>] [--shape=<shape>] [--color=<hex>] [--dark]"
-agent: vision
-model: proxypal/gemini-3-pro-image-preview
+agent: painter
 ---
 
 # Generate Icon: $ARGUMENTS

package/dist/template/.opencode/command/generate-image.md

@@ -1,8 +1,7 @@
 ---
 description: Generate images using Gemini image models
 argument-hint: "<prompt> [--style=<style>] [--aspect=<ratio>] [--size=<size>] [--count=<n>] [--pro]"
-agent: vision
-model: proxypal/gemini-3-pro-image-preview
+agent: painter
 ---
 
 # Generate Image: $ARGUMENTS

package/dist/template/.opencode/command/generate-pattern.md

@@ -1,8 +1,7 @@
 ---
 description: Generate seamless patterns and textures
 argument-hint: "<description> [--type=<type>] [--style=<style>] [--colors=<palette>] [--density=<level>] [--size=<px>]"
-agent: vision
-model: proxypal/gemini-3-pro-image-preview
+agent: painter
 ---
 
 # Generate Pattern: $ARGUMENTS

package/dist/template/.opencode/command/generate-storyboard.md

@@ -1,8 +1,7 @@
 ---
 description: Generate sequential images for visual storytelling
 argument-hint: "<story-description> [--frames=<N>] [--style=<style>] [--aspect=<ratio>] [--layout=<layout>]"
-agent: vision
-model: proxypal/gemini-3-pro-image-preview
+agent: painter
 ---
 
 # Generate Storyboard: $ARGUMENTS

package/dist/template/.opencode/command/implement.md

@@ -1,6 +1,6 @@
 ---
 description: Implement a bead - load context, delegate research, make changes, verify
-argument-hint: "<bead-id> [--parallel]"
+argument-hint: "<bead-id> [--parallel] [--swarm]"
 agent: build
 ---
 
@@ -14,6 +14,7 @@ You're implementing a tracked task. Stay focused, delegate research, verify as y
 | ------------ | -------- | ----------------------------------------- |
 | `<bead-id>`  | required | The bead to implement                     |
 | `--parallel` | false    | Run aggressive parallel subagent research |
+| `--swarm`    | false    | Use swarm coordination for parallel tasks |
 
 ## First: Load Skills & Context
 
@@ -21,6 +22,7 @@ You're implementing a tracked task. Stay focused, delegate research, verify as y
 skill({ name: "beads" });
 skill({ name: "test-driven-development" });
 skill({ name: "verification-before-completion" });
+skill({ name: "swarm-coordination" }); // For swarm mode
 ```
 
 Get the task details and check hierarchy:
@@ -33,8 +35,8 @@ bd list --status=in_progress  # See what else is active
 
 Check for messages from other agents:
 
-```typescript
-bd - inbox({ n: 5, unread: true });
+```bash
+bd inbox --unread -n 5
 ```
 
 ## Check Hierarchy Position
@@ -55,7 +57,131 @@ bd ready --json | grep -q "$ARGUMENTS"
 
 → Work on ready subtasks instead: `/implement <subtask-id>`
 
-**Or execute READY subtasks in parallel:**
+**Or use SWARM MODE for parallel execution (recommended for 3+ subtasks):**
+
+## Swarm Mode Execution (--swarm flag or auto-detect)
+
+When the task has 3+ independent subtasks that can run in parallel, use swarm coordination:
+
+### Step 1: Evaluate Swarm Eligibility
+
+```typescript
+// Check if subtasks are parallelizable:
+// - Do they modify different files? → YES = parallel OK
+// - Are they sequentially dependent? → NO = parallel OK
+// - Are there 3+ independent tasks? → YES = swarm beneficial
+```
+
+### Step 2: Create Delegation Packets
+
+For each parallelizable subtask:
+
+```typescript
+swarm_delegate({
+  bead_id: "<subtask-id>",
+  title: "Subtask title from bead",
+  expected_outcome: "Measurable completion criteria",
+  required_tools: "read, grep, lsp, edit, bash",
+  must_do: "LSP before edits, run project verification commands",
+  must_not_do: "No new deps, don't edit files outside scope",
+  acceptance_checks: "typecheck: npm run typecheck, lint: npm run lint, test: npm test",
+  context: "Part of epic $ARGUMENTS. See spec for requirements.",
+  write: true,
+});
+```
+
+### Step 3: Reserve File Sets
+
+Prevent conflicts between workers:
+
+```bash
+# Reserve files for each worker
+bd reserve "src/domain-a/*" --reason "swarm: subtask-1"
+bd reserve "src/domain-b/*" --reason "swarm: subtask-2"
+```
+
+### Step 4: Spawn Worker Swarm
+
+Launch all workers in parallel (multiple Task calls in one message):
+
+```typescript
+// All run simultaneously
+Task({
+  subagent_type: "general",
+  description: "Execute subtask-1",
+  prompt: `Execute bead <subtask-1>: <title>
+
+## Delegation
+Read: .beads/artifacts/<subtask-1>/delegation.md
+
+## Coordination
+Team: $ARGUMENTS-swarm
+Worker: worker-1
+
+## Protocol
+1. Read delegation packet FIRST
+2. Announce start: swarm-helper sendTeamMessage
+3. Execute following MUST DO / avoiding MUST NOT DO
+4. Run ALL acceptance checks
+5. Report completion: swarm-helper sendTeamMessage
+
+Report format:
+- DONE: <summary> (if successful)
+- ERROR: <issue> (if failed)`,
+});
+
+Task({
+  subagent_type: "general",
+  description: "Execute subtask-2",
+  prompt: `Execute bead <subtask-2>: <title>
+  ...same pattern, different worker name...`,
+});
+
+Task({
+  subagent_type: "general",
+  description: "Execute subtask-3",
+  prompt: `Execute bead <subtask-3>: <title>
+  ...same pattern, different worker name...`,
+});
+// Results available when all complete
+```
+
+### Step 5: Monitor Progress
+
+Check mailbox for worker status:
+
+```typescript
+swarm_helper({
+  operation: "getTeamStatus",
+  team_name: "$ARGUMENTS-swarm",
+  limit: 20,
+});
+```
+
+### Step 6: Handle Worker Reports
+
+| Message | Action                                  |
+| ------- | --------------------------------------- |
+| DONE    | Mark subtask complete, continue waiting |
+| ERROR   | Analyze, fix locally or respawn         |
+| BLOCKED | Resolve blocker, send guidance          |
+
+### Step 7: Final Verification
+
+After all workers complete:
+
+```bash
+# Full verification suite
+npm run typecheck && npm run lint && npm test
+```
+
+### Step 8: Close Epic
+
+```bash
+bd close $ARGUMENTS --reason "Swarm completed all subtasks: <summary>"
+```
+
+**Or execute READY subtasks sequentially (non-swarm mode):**
 
 ```typescript
 // Get all READY subtasks for this epic
@@ -264,14 +390,14 @@ If you hit 80% of budget without finishing, create a handoff. Don't push past li
 
 For shared files or multi-agent coordination:
 
-```typescript
-bd - reserve({ paths: ["src/file-to-edit.ts"], ttl: 600 });
+```bash
+bd reserve src/file-to-edit.ts --ttl 600
 ```
 
 Release after completing edits:
 
-```typescript
-bd - release({ paths: ["src/file-to-edit.ts"] });
+```bash
+bd release src/file-to-edit.ts
 ```
 
 ## Do The Work
@@ -417,6 +543,6 @@ Then start a fresh session. Don't grind past diminishing returns.
 
 Before ending session:
 
-```typescript
-bd - release({ _: true }); // List and release all locks
+```bash
+bd release --all  # Release all your locks
 ```

package/dist/template/.opencode/command/restore-image.md

@@ -1,8 +1,7 @@
 ---
 description: Restore, enhance, upscale, or improve image quality using AI
 argument-hint: "<image-path> [mode: enhance|upscale|restore|denoise|colorize]"
-agent: vision
-model: proxypal/gemini-3-pro-image-preview
+agent: painter
 ---
 
 # Restore Image: $ARGUMENTS

package/dist/template/.opencode/dcp.jsonc

@@ -3,7 +3,7 @@
   "enabled": true,
   "debug": false,
   // "minimal" shows prune activity without noise; "detailed" shows token counts
-  "pruneNotification": "minimal",
+  "pruneNotification": "off",
   "turnProtection": {
     "enabled": true,
     // 3 turns = faster cleanup while still protecting recent context
@@ -19,20 +19,6 @@
     "**/tsconfig.json",
     "**/biome.json",
   ],
-  // Custom slash commands (v1.2.4+)
-  "commands": {
-    "sweep": {
-      "description": "Aggressive context cleanup",
-      "action": "prune",
-      "options": {
-        "aggressive": true,
-      },
-    },
-    "context": {
-      "description": "Show context usage stats",
-      "action": "stats",
-    },
-  },
   "tools": {
     "settings": {
       "nudgeEnabled": true,

package/dist/template/.opencode/memory/observations/2026-01-25-decision-agent-roles-build-orchestrates-general-e.md

@@ -0,0 +1,14 @@
+---
+type: decision
+created: 2026-01-25T05:33:03.130Z
+confidence: high
+valid_until: null
+superseded_by: null
+concepts: ["agents", "build", "general", "orchestrator", "executor", "workflow", "swarm"]
+---
+
+# 🎯 Agent roles: Build orchestrates, General executes
+
+🟢 **Confidence:** high
+
+User decided on agent role split: Build is the primary orchestrator/lead agent controlling the workflow; General is the executor/implementer for individual Beads tasks. Other agents (plan/explore/scout/review) support as needed.

package/dist/template/.opencode/memory/observations/2026-01-25-decision-simplified-swarm-helper-tool-to-fix-type.md

@@ -0,0 +1,20 @@
+---
+type: decision
+created: 2026-01-25T06:49:53.678Z
+confidence: high
+valid_until: null
+superseded_by: null
+files: [".opencode/tool/swarm-helper.ts"]
+---
+
+# 🎯 Simplified swarm-helper tool to fix TypeScript errors
+
+🟢 **Confidence:** high
+
+The swarm-helper tool had TypeScript errors because it tried to import and use a `Task` function that doesn't exist in the @opencode-ai/plugin package. The @opencode-ai/plugin package only exports `tool`, not a `Task` function.
+
+Solution: Simplified the tool to only provide coordination operations that don't require external tool calls:
+- Removed: spawnTeam, assignTask operations
+- Kept: getTeamStatus, sendTeamMessage operations
+
+Team spawning should be done directly via the task tool in the command workflows, not wrapped in this helper. This keeps the tool focused on mailbox coordination only.

package/dist/template/.opencode/memory/observations/2026-01-25-decision-use-beads-as-swarm-board-source-of-truth.md

@@ -0,0 +1,14 @@
+---
+type: decision
+created: 2026-01-25T04:48:21.517Z
+confidence: high
+valid_until: null
+superseded_by: null
+concepts: ["beads", "swarm-protocol", "task-board", "dependencies", "source-of-truth", "opencode"]
+---
+
+# 🎯 Use Beads as swarm board source of truth
+
+🟢 **Confidence:** high
+
+User preference/decision: use `.beads/` as the single source of truth for the swarm task board (tasks + dependencies). Avoid introducing a separate swarm board file; plugins/tools should read from and write through Beads workflows.

package/dist/template/.opencode/memory/observations/2026-01-25-learning-user-wants-real-swarm-coordination-guida.md

@@ -0,0 +1,15 @@
+---
+type: learning
+created: 2026-01-25T04:43:38.235Z
+confidence: high
+valid_until: null
+superseded_by: null
+concepts: ["swarms", "multi-agent", "orchestration", "delegation", "task-board", "dependencies", "coordination", "opencode"]
+files: [".opencode/AGENTS.md"]
+---
+
+# 📚 User wants real swarm coordination guidance
+
+🟢 **Confidence:** high
+
+User wants best-practice guidance for building/improving 'real swarms' (multi-agent coordination) beyond centralized orchestration. They asked me to review `.opencode/AGENTS.md` and all `.opencode/agent/*.md` specs and synthesize improvements: shared task board with dependencies, parallel specialists, and coordination/messaging patterns compatible with their constraints (security-first, no URL guessing, delegation thresholds, LSP-first before edits, read-only specialists).

package/dist/template/.opencode/memory/research/opencode-mcp-bug-report.md

@@ -0,0 +1,126 @@
+# OpenCode MCP Bug Report
+
+## Title
+
+MCP tool calls fail with `TypeError: undefined is not an object (evaluating 'output.output.includes')` despite successful connection
+
+## Environment
+
+- **OpenCode version**: 1.1.36
+- **OS**: macOS (darwin)
+- **Node version**: (run `node --version` to fill in)
+
+## MCP Servers Affected
+
+- `context7` (https://mcp.context7.com/mcp)
+- `gh_grep` (https://mcp.grep.app)
+
+## Description
+
+MCP tool calls from the AI agent fail with a TypeError, even though the MCP servers connect successfully and respond with HTTP 200 OK.
+
+## Steps to Reproduce
+
+1. Configure MCP servers in `opencode.json`:
+
+```json
+{
+  "mcp": {
+    "context7": {
+      "type": "remote",
+      "url": "https://mcp.context7.com/mcp",
+      "headers": {
+        "CONTEXT7_API_KEY": "{env:CONTEXT7_API_KEY}"
+      }
+    },
+    "gh_grep": {
+      "type": "remote",
+      "url": "https://mcp.grep.app"
+    }
+  }
+}
+```
+
+2. Set the `CONTEXT7_API_KEY` environment variable
+
+3. Verify MCP servers are connected:
+
+```bash
+$ opencode mcp list
+✓ context7 connected - https://mcp.context7.com/mcp
+✓ gh_grep connected - https://mcp.grep.app
+```
+
+4. Debug shows successful connection:
+
+```bash
+$ opencode mcp debug context7
+HTTP response: 200 OK
+Server info: name=Context7, version=2.0.2
+
+$ opencode mcp debug gh_grep
+HTTP response: 200 OK
+```
+
+5. Use the AI agent to call a tool:
+
+```
+context7_resolve-library-id(query="React hooks", libraryName="react")
+```
+
+## Expected Behavior
+
+The tool should return library information from Context7.
+
+## Actual Behavior
+
+```
+TypeError: undefined is not an object (evaluating 'output.output.includes')
+```
+
+## Analysis
+
+The error `output.output.includes` suggests there's code in OpenCode that does something like:
+
+```javascript
+if (output.output.includes(...)) { ... }
+```
+
+When `output` is undefined or doesn't have an `output` property, this crashes.
+
+The bug is in **OpenCode's tool response handler**, not the MCP connection layer, because:
+
+1. `opencode mcp list` shows both servers as "connected"
+2. `opencode mcp debug` shows HTTP 200 OK responses
+3. The error happens only when the AI agent calls the tool
+
+## Workaround
+
+Use alternative tools that don't go through MCP:
+
+- `codesearch` (Exa Code API) - works
+- `websearch` (Exa Web Search) - works
+
+## Related Issues
+
+- #6972 - May be related to Accept header issues
+- #834 - SSE transport issues
+
+## Additional Context
+
+Log entries show benign errors that don't cause the crash:
+
+```
+ERROR service=mcp clientName=context7 error=MCP error -32601: Method not found failed to get prompts
+```
+
+This is expected (prompts method not implemented), but the TypeError happens later in the response processing pipeline.
+
+---
+
+## How to File
+
+1. Go to: https://github.com/anomalyco/opencode/issues/new
+2. Title: `MCP tool calls fail with TypeError: output.output.includes despite successful connection`
+3. Copy the content above
+4. Add labels: `bug`, `mcp`