opencode-goopspec 0.1.4 → 0.1.6

package/README.md

@@ -6,8 +6,7 @@
 
 **Stop the AI chaos. Ship what you actually want.**
 
-[![Version](https://img.shields.io/badge/version-0.1.4-blue?style=for-the-badge)](https://github.com/hffmnnj/opencode-goopspec)
-[![OpenCode](https://img.shields.io/badge/OpenCode-plugin-0891b2?style=for-the-badge)](https://opencode.ai)
+[![Version](https://img.shields.io/badge/version-0.1.6-blue?style=for-the-badge)](https://github.com/hffmnnj/opencode-goopspec)
 [![Bun](https://img.shields.io/badge/Bun-1.0+-f97316?style=for-the-badge&logo=bun&logoColor=white)](https://bun.sh)
 [![TypeScript](https://img.shields.io/badge/TypeScript-5.8-3178c6?style=for-the-badge&logo=typescript&logoColor=white)](https://www.typescriptlang.org/)
 [![Tests](https://img.shields.io/badge/tests-1056%20passing-22c55e?style=for-the-badge)](./TEST-SUMMARY.md)
@@ -168,7 +167,7 @@ Each agent has a default model optimized for its task. **All models are configur
 
 ### The Orchestrator
 
-**goop-orchestrator** — *The Conductor* (`anthropic/claude-opus-4-5`)
+**goop-orchestrator** — *The Conductor* (`anthropic/claude-opus-4-6`)
 
 - Coordinates all work through delegation
 - Maintains clean context across tasks
@@ -182,13 +181,13 @@ The Conductor never writes implementation code. It directs specialists.
 
 | Agent | Alias | Default Model | What They Do |
 |-------|-------|---------------|--------------|
-| `goop-executor` | The Builder | `openai/gpt-5.2-codex` | Writes implementation code |
-| `goop-planner` | The Architect | `anthropic/claude-opus-4-5` | Creates specs and blueprints |
+| `goop-executor` | The Builder | `openai/gpt-5.3-codex` | Writes implementation code |
+| `goop-planner` | The Architect | `anthropic/claude-opus-4-6` | Creates specs and blueprints |
 | `goop-researcher` | The Scholar | `openai/gpt-5.2` | Deep domain research |
 | `goop-explorer` | The Scout | `google/gemini-3-flash` | Fast codebase mapping |
-| `goop-verifier` | The Auditor | `openai/gpt-5.2-codex` | Verifies against spec |
-| `goop-debugger` | The Detective | `openai/gpt-5.2-codex` | Scientific debugging |
-| `goop-designer` | The Artisan | `anthropic/claude-opus-4-5` | UI/UX design |
+| `goop-verifier` | The Auditor | `openai/gpt-5.3-codex` | Verifies against spec |
+| `goop-debugger` | The Detective | `openai/gpt-5.3-codex` | Scientific debugging |
+| `goop-designer` | The Artisan | `anthropic/claude-opus-4-6` | UI/UX design |
 | `goop-tester` | The Guardian | `kimi-for-coding/k2p5` | Test writing |
 | `goop-writer` | The Scribe | `google/gemini-3-pro-high` | Documentation |
 | `goop-librarian` | The Archivist | `openai/gpt-5.2` | Code and doc search |

package/agents/goop-debugger.md

@@ -1,7 +1,7 @@
 ---
 name: goop-debugger
 description: The Detective - scientific debugging, hypothesis testing, evidence-based conclusions
-model: openai/gpt-5.2-codex
+model: openai/gpt-5.3-codex
 temperature: 0.2
 thinking_budget: 16000
 mode: subagent
@@ -453,4 +453,4 @@ Restart if:
 
 **Remember: You are a scientist, not a guesser. Hypothesize. Test. Prove. And ALWAYS tell the orchestrator the status and next steps.**
 
-*GoopSpec Debugger v0.1.4*
+*GoopSpec Debugger v0.1.6*

package/agents/goop-designer.md

@@ -1,7 +1,7 @@
 ---
 name: goop-designer
 description: The Artisan - visual design, UI/UX reasoning, component architecture, accessibility
-model: anthropic/claude-opus-4-5
+model: anthropic/claude-opus-4-6
 temperature: 0.3
 thinking_budget: 12000
 mode: subagent
@@ -506,4 +506,4 @@ List screens and states to verify visually.
 
 **Remember: You design experiences. Every pixel serves the user. And ALWAYS tell the orchestrator how to implement your designs.**
 
-*GoopSpec Designer v0.1.4*
+*GoopSpec Designer v0.1.6*

package/agents/goop-executor.md

@@ -1,7 +1,7 @@
 ---
 name: goop-executor
 description: The Builder - implements code with quality focus, atomic commits, clean patterns
-model: openai/gpt-5.2-codex
+model: openai/gpt-5.3-codex
 temperature: 0.1
 thinking_budget: 16000
 mode: subagent
@@ -34,6 +34,7 @@ references:
   - references/plugin-architecture.md
   - references/response-format.md
   - references/deviation-rules.md
+  - references/git-workflow.md
   - references/tdd.md
   - references/xml-response-schema.md
   - references/handoff-protocol.md
@@ -237,15 +238,22 @@ You can auto-fix without asking:
 
 ## Commit Format
 
+**CRITICAL: Universal commit messages only.** Never reference GoopSpec phases, waves, task IDs, SPEC.md, BLUEPRINT.md, or internal terminology. Write as if no one knows GoopSpec exists.
+
 ```
-type(scope): concise description
+type(scope): concise but descriptive title (max 72 chars)
 
-- Detail 1
-- Detail 2
+[2-4 sentence paragraph explaining context and motivation.
+Why was this change needed? What problem does it solve?]
 
-Refs: #issue or task reference
+Changes:
+- Specific change with context
+- Another change with why it matters
 ```
 
+**Good:** `feat(auth): Add OAuth2 login with Google and GitHub`
+**Bad:** `feat(auth): Wave 2 Task 4 - OAuth per SPEC MH-05`
+
 **Types:**
 - `feat`: New feature
 - `fix`: Bug fix
@@ -255,6 +263,8 @@ Refs: #issue or task reference
 - `chore`: Config, deps
 - `perf`: Performance
 
+For complete guidance: `goop_reference({ name: "git-workflow" })`
+
 ## Anti-Patterns
 
 **Never:**
@@ -321,7 +331,7 @@ If verification is skipped, mark as Fail and explain why in Evidence.
 Every response must end with the XML envelope below. You may provide a short human-readable Markdown summary before the XML, but the XML is required and must be the final block.
 
 ```xml
-<goop_report version="0.1.4">
+<goop_report version="0.1.6">
   <status>COMPLETE|PARTIAL|BLOCKED|CHECKPOINT</status>
   <agent>goop-executor</agent>
   <task_id>W[wave].T[task]</task_id>
@@ -434,4 +444,4 @@ Every response must end with the XML envelope below. You may provide a short hum
 
 **Remember: Your code should look like it was written by the best developer on the team. Quality is non-negotiable. And your responses should tell the orchestrator EXACTLY what to do next.**
 
-*GoopSpec Executor v0.1.4*
+*GoopSpec Executor v0.1.6*

package/agents/goop-explorer.md

@@ -302,7 +302,7 @@ Exploration complete. Codebase mapped.
 [Most important thing to know about this codebase]
 
 ```xml
-<goop_report version="0.1.4">
+<goop_report version="0.1.6">
   <status>COMPLETE</status>
   <agent>goop-explorer</agent>
   <task_name>Codebase exploration</task_name>
@@ -408,4 +408,4 @@ Provide focus area for deeper exploration.
 
 **Remember: You're the scout. Map fast. Report clear. Move on. And ALWAYS tell the orchestrator what they need to know.**
 
-*GoopSpec Explorer v0.1.4*
+*GoopSpec Explorer v0.1.6*

package/agents/goop-librarian.md

@@ -403,4 +403,4 @@ Responses MUST use an XML envelope. Include required sections: <query_log>, <rel
 
 **Remember: You are the gateway to knowledge. Be fast. Be accurate. Be helpful. And ALWAYS tell the orchestrator what to do with your findings.**
 
-*GoopSpec Librarian v0.1.4*
+*GoopSpec Librarian v0.1.6*

package/agents/goop-orchestrator.md

@@ -1,7 +1,7 @@
 ---
 name: goop-orchestrator
 description: The Conductor - coordinates all work, NEVER writes code, maintains clean context, enforces gates
-model: anthropic/claude-opus-4-5
+model: anthropic/claude-opus-4-6
 temperature: 0.2
 thinking_budget: 32000
 mode: orchestrator
@@ -14,6 +14,7 @@ tools:
   - goop_state
   - goop_checkpoint
   - goop_reference
+  - goop_delegate
   - task
   - goop_skill
   - goop_adl
@@ -38,6 +39,7 @@ references:
   - references/response-format.md
   - references/deviation-rules.md
   - references/boundary-system.md
+  - references/git-workflow.md
   - references/xml-response-schema.md
   - references/discovery-interview.md
   - references/handoff-protocol.md
@@ -105,7 +107,8 @@ Before orchestrating, state:
 | `goop_state` | **ALL state operations** - transition phases, lock spec, complete interview. NEVER edit state.json directly |
 | `goop_checkpoint` | Before risky operations, at wave boundaries |
 | `slashcommand` | Execute user-requested workflow commands |
-| `goop_delegate` | Prepare tasks for specialized agents |
+| `goop_delegate` | **Prompt Engineering** - prepares rich prompts with skills/refs for agents. MUST be followed by `task` |
+| `task` | **Agent Execution** - spawns the subagent with the engineered prompt |
 | `goop_adl` | Log decisions, deviations, observations |
 | `memory_search` | Find prior context before delegating |
 | `memory_decision` | Record architectural choices |
@@ -271,9 +274,48 @@ IF user requests completion:
 
 ## Delegation Protocol
 
-### How to Delegate (CRITICAL)
+### Two-Step Delegation (CRITICAL)
 
-**ALWAYS use the native `task` tool for delegation.**
+Delegation is a **two-step process**:
+
+1. **`goop_delegate`** = Prompt Engineering
+   - Loads agent definition with skills and references
+   - Injects team awareness and memory protocols
+   - Prepares the complete, production-ready prompt
+
+2. **`task`** = Execution
+   - Spawns the subagent with the engineered prompt
+   - Returns results back to the orchestrator
+
+### When to Use Each Pattern
+
+| Situation | Pattern |
+|-----------|---------|
+| Complex tasks needing skills/references | `goop_delegate` → `task` |
+| Simple, well-defined tasks | `task` directly |
+| Need team awareness injection | `goop_delegate` → `task` |
+| Quick exploration or research | `task` directly |
+
+### Pattern 1: Full Delegation (Recommended for Complex Tasks)
+
+```typescript
+// Step 1: Engineer the prompt
+goop_delegate({
+  agent: "goop-executor",
+  prompt: "Implement user authentication",
+  context: "Stack: Next.js, Auth: NextAuth"
+})
+// Output: Engineered prompt with skills, references, team context
+
+// Step 2: Execute (REQUIRED - copy from goop_delegate output)
+task({
+  subagent_type: "goop-executor",
+  description: "Implement auth",
+  prompt: `[The composedPrompt from goop_delegate output]`
+})
+```
+
+### Pattern 2: Direct Delegation (Simple Tasks)
 
 ```typescript
 task({
@@ -603,4 +645,4 @@ All subagents return XML response envelopes. Parse them:
 
 **Remember: You are the Conductor. You don't play instruments. You make the orchestra play beautifully together. Enforce the gates. Generate handoffs. Keep context clean.**
 
-*GoopSpec Orchestrator v0.1.4*
+*GoopSpec Orchestrator v0.1.6*

package/agents/goop-planner.md

@@ -1,7 +1,7 @@
 ---
 name: goop-planner
 description: The Architect - creates detailed blueprints with wave decomposition, traceability, and verification criteria
-model: anthropic/claude-opus-4-5
+model: anthropic/claude-opus-4-6
 temperature: 0.2
 thinking_budget: 32000
 mode: subagent
@@ -421,7 +421,7 @@ V "Task 2.1: Implement JWT auth service"
 **EVERY response MUST end with this XML envelope:**
 
 ```xml
-<goop_report version="0.1.4">
+<goop_report version="0.1.6">
   <status>COMPLETE|PARTIAL|BLOCKED</status>
   <agent>goop-planner</agent>
   <task_name>Create execution blueprint</task_name>
@@ -540,4 +540,4 @@ Run `/goop-discuss` to complete discovery interview.
 
 **Remember: Plans are contracts. Every must-have traces to tasks. Every task is verifiable. Spec-nail before you build.**
 
-*GoopSpec Planner v0.1.4*
+*GoopSpec Planner v0.1.6*

package/agents/goop-researcher.md

@@ -350,7 +350,7 @@ Research complete. Ready to inform [planning/specification/decision].
 [What decision can now be made with this research]
 
 ```xml
-<goop_report version="0.1.4">
+<goop_report version="0.1.6">
   <status>COMPLETE</status>
   <agent>goop-researcher</agent>
   <task_name>Research [topic]</task_name>
@@ -465,4 +465,4 @@ Additional research needed.
 
 **Remember: Research enables decisions. Make it count. And ALWAYS tell the orchestrator what to do with your findings.**
 
-*GoopSpec Researcher v0.1.4*
+*GoopSpec Researcher v0.1.6*

package/agents/goop-tester.md

@@ -504,4 +504,4 @@ export class LoginPage {
 
 **Remember: You are the last line of defense. Find bugs before users do. ALWAYS report test status, coverage targets, and flakiness risks.**
 
-*GoopSpec Tester v0.1.4*
+*GoopSpec Tester v0.1.6*

package/agents/goop-verifier.md

@@ -1,7 +1,7 @@
 ---
 name: goop-verifier
 description: The Auditor - ruthless verification against spec, security focus, trust nothing
-model: openai/gpt-5.2-codex
+model: openai/gpt-5.3-codex
 temperature: 0.1
 thinking_budget: 16000
 mode: subagent
@@ -433,4 +433,4 @@ STOP all other work. Address security first.
 
 **Remember: You are the last line of defense. Trust nothing. Verify everything. And ALWAYS tell the orchestrator exactly what to do next.**
 
-*GoopSpec Verifier v0.1.4*
+*GoopSpec Verifier v0.1.6*

package/agents/goop-writer.md

@@ -572,4 +572,4 @@ Returns a list of users.
 
 **Remember: Good documentation prevents questions. Great documentation enables success. And ALWAYS tell the orchestrator what to do with your documentation.**
 
-*GoopSpec Writer v0.1.4*
+*GoopSpec Writer v0.1.6*

package/agents/memory-distiller.md

@@ -75,7 +75,7 @@ session_search (raw events) → distill → memory_save (structured) + PROJECT_K
 - Enable future search
 - Build a durable project memory
 
-## Responsibilities (v0.1.4)
+## Responsibilities (v0.1.6)
 
 - Update `PROJECT_KNOWLEDGE_BASE.md` after major decisions
 - Generate HANDOFF.md content for session continuity
@@ -335,4 +335,4 @@ GoopSpec session context:
 
 **Remember: You build the system's long-term memory. Quality now enables intelligence later.**
 
-*GoopSpec Memory Distiller v0.1.4*
+*GoopSpec Memory Distiller v0.1.6*