opencode-starterkit 1.0.1

package/baseline/memory/project/tech-stack.md

@@ -0,0 +1,53 @@
+---
+purpose: Tech stack, constraints, and integrations for AI context injection
+updated: 2026-02-24
+---
+
+# Tech Stack
+
+This file is automatically injected into ALL AI prompts via `opencode.json` instructions[].
+
+## Framework & Language
+
+- **Framework:** CLI tool (cac for argument parsing)
+- **Language:** TypeScript (ESNext, strict mode, bundler moduleResolution)
+- **Runtime:** Node.js >= 20.19.0
+
+## Key Dependencies
+
+- **CLI Framework:** cac (^6.7.14) - Command-line argument parsing
+- **UI Prompts:** @clack/prompts (^0.7.0) - Interactive CLI prompts
+- **Validation:** zod (^3.25.76) - Schema validation
+- **Task Tracking:** beads-village (^1.3.3) - Git-backed task management
+- **AI SDK:** @ai-sdk/provider (^3.0.6) - AI provider integration
+
+## Build & Tools
+
+- **Build:** tsdown + rsync for template bundling
+- **Lint:** oxlint (^1.38.0) - Fast JavaScript/TypeScript linter
+- **Format:** oxfmt (^0.23.0) - Code formatter
+- **TypeCheck:** tsgo (@typescript/native-preview)
+- **Dev Runner:** tsx - TypeScript execution for development
+
+## Testing
+
+- **Unit Tests:** vitest
+- **Test Location:** src/\*_/_.test.ts (colocated)
+- **Run Single:** vitest src/commands/init.test.ts
+
+## Key Constraints
+
+- Node.js >= 20.19.0 required (engines.node in package.json)
+- pnpm for package management
+- Build copies .opencode/ to dist/template/ - don't edit dist/ directly
+- Keep .opencode/ structure minimal and focused
+
+## Active Integrations
+
+- **OpenCode AI:** @opencode-ai/plugin (^1.1.12) - OpenCode integration
+- **Beads CLI:** beads_rust (br) - Task tracking CLI
+
+---
+
+_Update this file when tech stack or constraints change._
+_AI will capture architecture, conventions, and gotchas via the `observation` tool as it works._

package/baseline/memory/project/user.md

@@ -0,0 +1,38 @@
+---
+purpose: User identity, preferences, communication style
+updated: 2025-01-06
+---
+
+# User Profile
+
+## Identity
+
+- Name: huynhgiabuu
+- Git email: buub1309120@gmail.com
+
+## Communication Preferences
+
+- Style: Detailed responses preferred
+- Explanations welcome when helpful
+
+## Workflow Preferences
+
+- Git commits: Ask first before committing
+- Beads updates: Ask first before modifying task state
+- Auto-actions: Disabled - always confirm before state changes
+
+## Technical Preferences
+
+- Runtime: Node.js
+- Language: TypeScript
+- Linter: oxlint
+
+## Rules to Always Follow
+
+- Run `npm run lint:fix` before any commit
+- Validate changes with `npm run typecheck`
+- Don't modify dist/ directly (it's built output)
+- Ask before adding new dependencies
+- Ask before changing .opencode/ structure
+- Never commit secrets or .env files
+- Never force push to main

package/baseline/memory/research/benchmark-framework.md

@@ -0,0 +1,162 @@
+---
+purpose: Scoring rubric for evaluating template agent effectiveness
+updated: 2026-03-08
+based-on: tilth research (measurable pattern adoption improvements)
+---
+
+# Agent Effectiveness Benchmark Framework
+
+## Purpose
+
+Evaluate whether skills, tools, and commands in the OpenCodeKit template actually help AI agents perform better. Based on tilth's methodology: they measured accuracy, cost/correct answer, and tool adoption rates to prove what works.
+
+## Scoring Dimensions
+
+7 dimensions, each scored 0–2. Max score: **14**.
+
+### 1. Trigger Clarity (WHEN/SKIP)
+
+Does the description clearly specify when to load AND when NOT to?
+
+| Score | Criteria                                    |
+| ----- | ------------------------------------------- |
+| 0     | Vague or missing trigger conditions         |
+| 1     | Has WHEN but not WHEN NOT (or vice versa)   |
+| 2     | Clear WHEN and WHEN NOT (SKIP) binary gates |
+
+**Why it matters:** tilth found explicit WHEN/SKIP gates are the single most effective pattern for correct tool routing. Without them, agents either over-load (waste tokens) or under-load (miss relevant skills).
+
+### 2. "Replaces X" Framing
+
+Does it explicitly state what behavior, tool, or workflow it replaces?
+
+| Score | Criteria                                       |
+| ----- | ---------------------------------------------- |
+| 0     | No replacement framing                         |
+| 1     | Implied replacement or "better than X"         |
+| 2     | Explicit "Replaces X" statement in description |
+
+**Why it matters:** tilth measured +36 percentage points adoption on Haiku when tool descriptions included "Replaces X" framing. Models route better when they know what's superseded.
+
+### 3. Concrete Examples
+
+Does it provide working code with actual tool calls, not just prose?
+
+| Score | Criteria                                                                |
+| ----- | ----------------------------------------------------------------------- |
+| 0     | No examples                                                             |
+| 1     | Prose descriptions or generic prompt templates                          |
+| 2     | Working code examples with actual tool calls / before-after comparisons |
+
+**Why it matters:** Models follow examples more reliably than instructions. Prompt templates ("Analyze this image: [attach]") score 1, not 2 — they lack tool integration.
+
+### 4. Anti-Patterns
+
+Does it show what NOT to do?
+
+| Score | Criteria                                                       |
+| ----- | -------------------------------------------------------------- |
+| 0     | No anti-patterns section                                       |
+| 1     | Brief "don't do X" mentions                                    |
+| 2     | Wrong/right comparison table or detailed anti-patterns section |
+
+**Why it matters:** Failure prevention is as valuable as success instruction. tilth's evidence-based feature removal (disabling `--map` because 62% of losing tasks used it) proves tracking what fails matters.
+
+### 5. Verification Integration
+
+Does it reference or require verification steps?
+
+| Score | Criteria                                                       |
+| ----- | -------------------------------------------------------------- |
+| 0     | No mention of verification                                     |
+| 1     | Mentions verification in passing                               |
+| 2     | Integrates verification steps into workflow (run X, confirm Y) |
+
+**Why it matters:** Skills that don't include verification produce unverified outputs. The build loop is perceive → create → **verify** → ship.
+
+### 6. Token Efficiency
+
+Is the token cost proportional to value delivered?
+
+| Score | Criteria                                                                       |
+| ----- | ------------------------------------------------------------------------------ |
+| 0     | >2500 tokens with low value density (filler, repetition, obvious instructions) |
+| 1     | Reasonable size OR moderate value density                                      |
+| 2     | <1500 tokens with high value density, OR larger with proportional density      |
+
+**Why it matters:** Every loaded skill consumes context budget. A 4000-token skill that could be 1500 tokens is actively harmful — it displaces working memory.
+
+### 7. Cross-References
+
+Does it link to related skills for next steps?
+
+| Score | Criteria                                                                     |
+| ----- | ---------------------------------------------------------------------------- |
+| 0     | No references to other skills                                                |
+| 1     | Mentions related skills in text                                              |
+| 2     | Clear "Related Skills" table or "Next Phase" with skill loading instructions |
+
+**Why it matters:** Skills that exist in isolation force agents to discover connections. Explicit connections reduce routing failures.
+
+## Score Interpretation
+
+| Range | Tier       | Meaning                                                     |
+| ----- | ---------- | ----------------------------------------------------------- |
+| 12–14 | Exemplary  | Ready to ship — high adoption, measurable value             |
+| 8–11  | Adequate   | Functional but missing patterns that would improve adoption |
+| 4–7   | Needs Work | Significant gaps — may load but produce suboptimal results  |
+| 0–3   | Poor       | Should be rewritten or merged into another skill            |
+
+## Category Assessment
+
+Beyond individual scoring, evaluate each skill's **category fit**:
+
+| Category             | Expected Traits                                                                  |
+| -------------------- | -------------------------------------------------------------------------------- |
+| Core Workflow        | Loaded frequently, high token ROI, tight integration with other core skills      |
+| Planning & Lifecycle | Clear phase transitions, handoff points between skills                           |
+| Debugging & Quality  | Real examples from actual debugging sessions, measurable impact                  |
+| Code Review          | Severity levels, actionable findings format                                      |
+| Design & UI          | Visual reference integration, component breakdown                                |
+| Agent Orchestration  | Parallelism rules, coordination protocols                                        |
+| External Integration | API examples, auth handling, error patterns                                      |
+| Platform Specific    | Version-pinned APIs, migration guidance                                          |
+| Meta Skills          | Self-referential consistency (does the skill-about-skills follow its own rules?) |
+
+## Audit Process
+
+1. **Inventory** — List all skills with token size
+2. **Sample** — Read representative skills from each category
+3. **Score** — Apply 7 dimensions to each sampled skill
+4. **Classify** — Assign tier and category
+5. **Identify** — Flag overlaps, dead weight, and upgrade candidates
+6. **Prioritize** — Rank improvements by impact (core skills first)
+
+## Effectiveness Signals (Observable)
+
+Beyond the rubric, track these runtime signals when possible:
+
+| Signal                                     | Indicates                                                    |
+| ------------------------------------------ | ------------------------------------------------------------ |
+| Skill loaded but instructions not followed | Trigger too broad OR instructions too vague                  |
+| Skill never loaded despite relevant tasks  | Trigger too narrow OR description doesn't match task framing |
+| Agent re-reads files after skill search    | Skill examples insufficient — agent needs more context       |
+| Verification skipped after skill workflow  | Skill doesn't integrate verification                         |
+| Agent loads 5+ skills simultaneously       | Skills too granular — should be merged                       |
+
+## Template-Level Metrics
+
+For the overall template (all skills + tools + commands):
+
+| Metric                        | Target | Current |
+| ----------------------------- | ------ | ------- |
+| Core skills at Exemplary tier | 100%   | (audit) |
+| No skills at Poor tier        | 0      | (audit) |
+| Average token cost per skill  | <1500  | (audit) |
+| Skills with WHEN/SKIP gates   | 100%   | (audit) |
+| Skills with anti-patterns     | >75%   | (audit) |
+| Overlap/redundancy pairs      | 0      | (audit) |
+
+---
+
+_Apply this framework during effectiveness audits. Update scoring criteria as new evidence emerges._

package/baseline/memory/research/ccpm-analysis.md

@@ -0,0 +1,334 @@
+# CCPM Research Analysis
+
+**Source:** https://github.com/automazeio/ccpm  
+**Research Date:** 2026-02-01  
+**Purpose:** Learn and adapt patterns into OpenCodeKit
+
+---
+
+## Executive Summary
+
+CCPM (Claude Code PM) is a comprehensive workflow system for spec-driven development using GitHub Issues as the task database. Key innovations include:
+
+1. **PRD → Epic → Task → Issue pipeline** - Full traceability from idea to code
+2. **Parallel agent execution** - Multiple agents working simultaneously via git worktrees
+3. **Context preservation** - Agents as "context firewalls" shielding main thread
+4. **GitHub-native integration** - Issues as source of truth, no external databases
+
+---
+
+## Architecture Overview
+
+```
+ccpm/
+├── commands/
+│   ├── pm/                    # 40+ project management commands
+│   │   ├── prd-new.md         # Create PRD via brainstorming
+│   │   ├── prd-parse.md       # PRD → Epic conversion
+│   │   ├── epic-decompose.md  # Epic → Tasks breakdown
+│   │   ├── epic-sync.md       # Push to GitHub Issues
+│   │   ├── issue-start.md     # Begin parallel work
+│   │   ├── next.md            # Get prioritized task
+│   │   └── status.md          # Project dashboard
+│   ├── context/               # Context management
+│   │   ├── create.md          # Initialize project context
+│   │   ├── prime.md           # Load context for session
+│   │   └── update.md          # Refresh context
+│   └── testing/               # Test execution
+├── agents/                    # Specialized workers
+│   ├── parallel-worker.md     # Coordinates parallel streams
+│   ├── code-analyzer.md       # Bug hunting specialist
+│   ├── test-runner.md         # Test execution + analysis
+│   └── file-analyzer.md       # Verbose file summarization
+├── rules/                     # Operational guidelines
+│   ├── agent-coordination.md  # Parallel work rules
+│   ├── standard-patterns.md   # Command patterns
+│   └── worktree-operations.md # Git worktree usage
+├── context/                   # Project context files
+├── prds/                      # Product requirement docs
+└── epics/                     # Local workspace
+    └── {epic-name}/
+        ├── epic.md            # Implementation plan
+        ├── {issue}.md         # Individual tasks
+        └── updates/           # Progress tracking
+```
+
+---
+
+## Key Patterns to Adopt
+
+### 1. PRD-First Development Flow
+
+**CCPM Flow:**
+
+```
+/pm:prd-new feature    → Brainstorm + create PRD
+/pm:prd-parse feature  → Convert PRD to Epic
+/pm:epic-decompose     → Break into tasks
+/pm:epic-sync          → Push to GitHub Issues
+/pm:issue-start 123    → Begin parallel work
+```
+
+**OpenCodeKit Equivalent:**
+
+```
+/create <bead> --prd   → Create PRD artifact
+/plan <bead>           → Generate implementation plan
+/start <bead>          → Begin work with task allocation
+```
+
+**Gap:** We don't have `/pm:epic-decompose` equivalent - our `/plan` command does both epic creation AND task decomposition. Consider separating.
+
+---
+
+### 2. Agent Philosophy: "Context Firewalls"
+
+**CCPM Quote:**
+
+> "Don't anthropomorphize subagents. Use them to organize your prompts and elide context. Subagents are best when they can do lots of work but then provide small amounts of information back."
+
+**Key Insight:** Agents aren't "experts" - they're task executors that protect main thread from information overload.
+
+**CCPM Agent Pattern:**
+
+```typescript
+// Agent reads 10 files → Main gets 1 summary
+Task({
+  subagent_type: "code-analyzer",
+  prompt: "Hunt bugs in auth module",
+  // Returns: "Found 3 issues: [concise list]"
+  // Main thread never sees: All the files examined
+});
+```
+
+**OpenCodeKit Gap:** Our subagents return too much detail. Should adopt concise summary pattern.
+
+---
+
+### 3. Parallel Work via Git Worktrees
+
+**CCPM Innovation:** Each epic gets its own worktree for isolated parallel development.
+
+```bash
+# Create worktree for epic
+git worktree add ../epic-{name} -b epic/{name}
+
+# Multiple agents work in SAME worktree on DIFFERENT files
+Agent-A: src/db/*     → No conflict
+Agent-B: src/api/*    → No conflict
+Agent-C: src/ui/*     → No conflict
+```
+
+**Coordination Rules:**
+
+1. File-level parallelism - different files = no conflict
+2. Explicit coordination for shared files
+3. Atomic commits with clear messages
+4. Human resolution for conflicts (never auto-merge)
+
+**OpenCodeKit Gap:** We have `using-git-worktrees` skill but don't integrate it into task execution flow.
+
+---
+
+### 4. Task File Format with Dependencies
+
+**CCPM Task Frontmatter:**
+
+```yaml
+---
+name: Implement user authentication
+status: open
+created: 2024-01-20T10:00:00Z
+updated: 2024-01-20T10:00:00Z
+github: https://github.com/org/repo/issues/1234
+depends_on: [001, 002] # Task dependencies
+parallel: true # Can run in parallel
+conflicts_with: [003] # Same files modified
+---
+```
+
+**Key Fields Missing in OpenCodeKit:**
+
+- `depends_on` - Task dependencies
+- `parallel` - Parallelization flag
+- `conflicts_with` - File conflict tracking
+
+---
+
+### 5. Context Management System
+
+**CCPM Context Files:**
+
+```
+.claude/context/
+├── project-brief.md       # What + why
+├── project-overview.md    # Features + capabilities
+├── project-vision.md      # Long-term direction
+├── tech-context.md        # Stack + dependencies
+├── project-structure.md   # Directory organization
+├── system-patterns.md     # Architecture patterns
+├── project-style-guide.md # Coding conventions
+├── product-context.md     # Users + requirements
+└── progress.md            # Current status
+```
+
+**Commands:**
+
+- `/context:create` - Initialize all context files
+- `/context:prime` - Load context at session start
+- `/context:update` - Refresh after changes
+
+**OpenCodeKit Comparison:**
+
+- We have `.opencode/memory/` but less structured
+- We lack `/context:prime` equivalent for session start
+- Our `memory-read` is more ad-hoc
+
+---
+
+### 6. Standard Output Patterns
+
+**CCPM Standard:**
+
+```markdown
+# Success
+
+✅ {Action} complete
+
+- {Key result 1}
+- {Key result 2}
+  Next: {Single suggested action}
+
+# Error
+
+❌ {What failed}: {Exact solution}
+
+# List
+
+{Count} {items} found:
+
+- {item 1}: {key detail}
+- {item 2}: {key detail}
+```
+
+**Principles:**
+
+- Fail fast - check essentials, trust the system
+- Minimal output - show what matters, skip decoration
+- Clear errors - say exactly what and how to fix
+- Smart defaults - only ask when destructive
+
+---
+
+### 7. Parallel Execution System
+
+**CCPM Insight:** One issue ≠ one task. One issue = multiple parallel work streams.
+
+```bash
+# Issue #1234 explodes into parallel streams:
+Agent 1: Database tables
+Agent 2: Service layer
+Agent 3: API endpoints
+Agent 4: UI components
+Agent 5: Tests + docs
+```
+
+**Velocity Math:**
+
+- Traditional: 3 issues × 1 agent = 3 sequential tasks
+- CCPM: 3 issues × 4 streams each = 12 parallel agents
+
+**OpenCodeKit Gap:** Our `/start` command doesn't do stream analysis or parallel agent spawning.
+
+---
+
+## Recommended Adaptations for OpenCodeKit
+
+### Priority 1: Command Enhancements
+
+1. **`/context:prime`** - Load all memory files at session start
+2. **`/status`** - Project dashboard showing beads, progress, blockers
+3. **`/next`** - Get next priority task with full context
+4. **Enhanced `/start`** - Include parallel stream analysis
+
+### Priority 2: Agent Improvements
+
+1. **Concise Returns** - Agents should return summaries, not verbose output
+2. **Parallel Worker Pattern** - Coordinator agent that spawns sub-agents
+3. **Code Analyzer** - Dedicated bug-hunting agent with structured output
+
+### Priority 3: Task System Enhancements
+
+1. **Dependency Tracking** - `depends_on`, `parallel`, `conflicts_with` fields
+2. **Stream Analysis** - Break tasks into parallel work streams
+3. **Worktree Integration** - Auto-create worktrees for epics
+
+### Priority 4: Context System
+
+1. **Structured Context Files** - Adopt CCPM's context file categories
+2. **Auto-Context Loading** - Prime context automatically on session start
+3. **Context Staleness Detection** - Warn when context is outdated
+
+---
+
+## Implementation Recommendations
+
+### Immediate Actions
+
+1. **Create `/prime` command** - Equivalent to `/context:prime`
+   - Load `.opencode/memory/` files
+   - Show project status
+   - List active beads
+
+2. **Create `/status` command** - Project dashboard
+   - Bead count by status
+   - Recent activity
+   - Blockers
+
+3. **Enhance agent return format** - Adopt CCPM's concise summary pattern
+
+### Medium-Term Actions
+
+1. **Add task dependency fields** to bead artifacts
+2. **Create `parallel-worker` skill** for coordinated parallel execution
+3. **Integrate git worktrees** into `/start` command
+
+### Long-Term Actions
+
+1. **Full context system** with structured categories
+2. **GitHub Issues integration** (optional, for teams)
+3. **Progress visualization** with dashboard commands
+
+---
+
+## Key Quotes Worth Remembering
+
+> "Every line of code must trace back to a specification." - No vibe coding
+
+> "Simple is not simplistic - we still handle errors properly, we just don't try to prevent every possible edge case."
+
+> "Your main conversation becomes the conductor, not the orchestra."
+
+> "Context preservation: Use extremely concise language. Every word must earn its place."
+
+---
+
+## Files Worth Studying Further
+
+1. `ccpm/commands/pm/issue-analyze.md` - Stream analysis logic
+2. `ccpm/commands/pm/epic-start.md` - Worktree creation flow
+3. `ccpm/agents/parallel-worker.md` - Coordination patterns
+4. `ccpm/rules/agent-coordination.md` - Parallel work rules
+
+---
+
+## Conclusion
+
+CCPM provides a mature, battle-tested workflow for AI-assisted development. The key innovations are:
+
+1. **PRD-driven discipline** - Requirements before code
+2. **Parallel by default** - One task = many agents
+3. **Context as firewall** - Agents protect main thread
+4. **GitHub as database** - No external dependencies
+
+OpenCodeKit can adopt these patterns incrementally, starting with `/prime`, `/status`, and enhanced agent return formats.