@ainative/cody-cli 0.7.1 → 0.7.2

package/.ainative/AINATIVE.md

@@ -1,49 +1,84 @@
 # Project Context
 
-<!-- Replace with your project name and description -->
-**Project**: My Project
-**Compatible With**: Gemini CLI, Claude Code, and other AI coding assistants
+**Project**: Cody CLI — AINative Studio's AI-powered coding assistant
+**Compatible With**: Claude Code, Gemini CLI, and other AI coding assistants
 
 ---
 
 ## Project Overview
 
-<!-- Replace with your project description -->
-<!-- Example: This is a Next.js web application for... -->
+Cody CLI is a fork/rebrand of Claude Code, building AINative Studio's terminal-based AI coding assistant. It connects to AINative's API gateway (Kong) which routes to multiple LLM providers. The CLI supports OAuth 2.1 + PKCE authentication, MCP (Model Context Protocol) servers, streaming chat completions, tool use, sub-agents, and a rich terminal UI built with Ink/React.
+
+**Current Version**: 0.7.1
+**Status**: Active Development — SDK works, chat completions API confirmed working
 
 ---
 
 ## Repository Paths
 
 ### Primary Repository
-<!-- Replace with your repository path -->
-**Path**: `/path/to/your/project`
-**Framework**: <!-- e.g., Next.js 16, FastAPI, etc. -->
+**Path**: `/Users/aideveloper/Desktop/cody-cli`
+**GitHub**: `https://github.com/AINative-Studio/cody-cli`
+**Framework**: Bun + TypeScript + React (Ink terminal UI)
 **Status**: Active Development
 
 ---
 
 ## Tech Stack
 
-<!-- Replace with your actual tech stack -->
-- **Framework**: <!-- e.g., Next.js 16 (App Router, TypeScript) -->
-- **Styling**: <!-- e.g., Tailwind CSS + shadcn/ui -->
-- **State Management**: <!-- e.g., React Query -->
-- **Deployment**: <!-- e.g., Railway, Vercel, AWS -->
-- **Package Manager**: <!-- e.g., npm, bun, pnpm -->
+- **Runtime**: Bun >= 1.1.0 (also supports Node >= 18 via thin wrapper)
+- **Language**: TypeScript (ES2022, JSX, bundler module resolution)
+- **UI Framework**: Ink 5.0 + React 19.1.0 (terminal UI)
+- **CLI Framework**: Commander.js 12.0
+- **AI SDK**: @anthropic-ai/sdk 0.82.0 (Beta Messages API)
+- **MCP**: @modelcontextprotocol/sdk 1.29.0
+- **Observability**: OpenTelemetry (traces, metrics, logs)
+- **Build**: Bun bundler (TypeScript -> JS bundle -> optional native binary)
+- **Testing**: bun:test (built-in test runner)
+- **Deployment**: npm registry (@ainative/cody-cli), Homebrew tap
+- **Package Manager**: Bun (bun.lock)
 
 ---
 
 ## Architecture
 
-<!-- Replace with your architecture description -->
+### System Flow
+```
+Developer -> Cody CLI -> Kong API Gateway -> Backend LLMs
+                |
+                +-> MCP Servers (ZeroDB, Chrome, etc.)
+                +-> OAuth 2.1 + PKCE (api.ainative.studio)
+                +-> Local tools (Bash, FileRead, FileWrite, etc.)
+```
 
-### Page/Module Structure
+### Two CLI Modes
+1. **Thin Wrapper** (`bin/cody.js`): Node.js launcher, auto-installs Bun
+2. **Full Binary** (`cody-bin`): ~80MB self-contained Bun-compiled binary
+
+### Source Directory Structure
 ```
-<!-- Replace with your directory structure -->
-app/
-  layout.tsx
-  page.tsx
+src/
+  entrypoints/       # CLI bootstrap (cli.tsx, init.ts, mcp.ts, sdk/)
+  commands/           # 100+ slash commands
+  tools/              # LLM-callable tools (Bash, FileRead, FileWrite, Glob, Grep, Agent, MCP)
+  services/           # Business logic (oauth, mcp, api, analytics, compact, memory)
+  components/         # 140+ Ink/React UI components
+  hooks/              # 85+ React hooks
+  utils/              # 330+ utility modules
+  bridge/             # Remote control / bridge mode
+  skills/             # Loadable skill system
+  tasks/              # Task tracking subsystem
+  context/            # Context window management
+  state/              # Application state management
+  main.tsx            # Main application loop & Ink render tree
+  commands.ts         # Command registry & dispatch
+```
+
+### Build Pipeline
+```
+TypeScript (src/) -> bun build -> dist/cli.js (JS bundle)
+                                      |
+                                      +-> bun build --compile -> cody-bin (native binary)
 ```
 
 ---
@@ -51,9 +86,19 @@ app/
 ## Services & APIs
 
 ### Backend API
-<!-- Replace with your API details -->
-**Base URL**: `https://your-api.example.com`
-**Auth**: <!-- e.g., JWT tokens, API keys -->
+**Base URL**: `https://api.ainative.studio`
+**Auth**: OAuth 2.1 + PKCE (Client ID: `ainative-cody-cli-001`)
+**Scopes**: `user:profile`, `user:inference`, `user:sessions:claude_code`, `user:mcp_servers`, `user:file_upload`, `org:create_api_key`
+
+### MCP Servers
+- **ZeroDB**: Vector database and storage (`https://api.ainative.studio/v1/mcp/zerodb-mcp/messages`)
+
+### Available Models
+- `qwen-coder-32b` (default)
+- `qwen-coder-7b`
+- `nouscoder-14b`
+- `gemma-9b`
+- `deepseek-r1-distill-qwen-7b`
 
 ---
 
@@ -61,33 +106,34 @@ app/
 
 ### Local Development
 ```bash
-# Replace with your actual commands
-npm run dev          # Start dev server
-npm run lint         # Linting
-npm run type-check   # Type checking
-npm run build        # Build verification
-npm test             # Run tests
+bun install             # Install dependencies
+bun run build           # Build JS bundle (dist/cli.js)
+bun run start           # Run built CLI
+bash compile.sh         # Build native binary (optional)
+bun test                # Run all tests
+bun test --coverage     # Run tests with coverage report
 ```
 
 ### Pre-Commit Checklist
-1. Lint - must pass
-2. Type-check - must pass
-3. Build - must succeed
-4. Tests - all tests pass
+1. Build — must succeed (`bun run build`)
+2. Tests — all tests pass with >= 80% coverage (`bun test --coverage`)
+3. No secrets, API keys, or tokens in code
+4. No AI tool attribution in commit messages
 
 ---
 
 ## Git Workflow
 
 ### Branch Strategy
-- `main` - production branch
+- `main` — production branch
 - Feature branches: `feature/[name]`
 - Bug fixes: `fix/[name]`
 
 ### Commit Rules (ZERO TOLERANCE)
 **NEVER include in commits, PRs, or GitHub activity:**
-- AI tool attribution of any kind
+- AI tool attribution of any kind ("Claude", "Anthropic", "ChatGPT", "Copilot")
 - "Generated with", "Co-Authored-By" referencing AI tools
+- "AI-generated" or "AI-assisted"
 
 See `.ainative/git-rules.md` for complete rules.
 
@@ -96,33 +142,45 @@ See `.ainative/git-rules.md` for complete rules.
 ## AI Agent Configuration
 
 ### Claude Code
-- Primary agent configuration: `.claude/CLAUDE.md`
-- Custom rules: `.claude/rules/`
+- Primary agent configuration: `CLAUDE.md` (project root)
+- Settings: `.claude/settings.json`
+- Skills: `.claude/skills/`
+- Hooks: `.claude/hooks/`
 
-### Gemini CLI (This File)
+### Gemini CLI / Other AI Tools
 - Project context: `.ainative/AINATIVE.md` (this file)
 - Settings: `.ainative/settings.json`
-- Custom rules: `.ainative/rules/`
+- Skills: `.ainative/skills/`
 
 ---
 
 ## MCP Servers (Model Context Protocol)
 
-<!-- Replace with your MCP server configuration -->
-See `.claude/mcp.json` for MCP configuration.
+See `.mcp.json` for MCP configuration.
+See `.claude/settings.local.json` for enabled MCP servers.
 
 ---
 
 ## Key Files Reference
 
 ### Configuration
-<!-- Replace with your key config files -->
-- `package.json` - Dependencies
-- `tsconfig.json` - TypeScript configuration
+- `package.json` — Dependencies, scripts, npm publishing config
+- `tsconfig.json` — TypeScript config (ES2022, bundler, JSX)
+- `bunfig.toml` — Bun test runner config (30s timeout)
+- `.mcp.json` — MCP server URLs
+- `globals.d.ts` — Build-time MACRO type declarations
+
+### Build
+- `build.sh` — TypeScript -> JS bundle
+- `compile.sh` — JS -> native binary
+- `postbuild.sh` — Add shebang + chmod
+- `install.sh` — End-user installer (curl | sh)
 
 ### Documentation
-- `.ainative/AINATIVE.md` - This file
-- `.claude/CLAUDE.md` - Claude Code context
+- `docs/ARCHITECTURE.md` — System design
+- `docs/DEVELOPMENT.md` — Setup & contribution guide
+- `docs/TESTING.md` — Test strategy & running tests
+- `docs/MODELS.md` — Available models & selection
 
 ---
 
@@ -130,15 +188,17 @@ See `.claude/mcp.json` for MCP configuration.
 
 ### Required
 ```bash
-# Replace with your required env vars
-# EXAMPLE:
-# NEXT_PUBLIC_SITE_URL=https://www.example.com
-# NEXT_PUBLIC_API_URL=https://api.example.com
+ANTHROPIC_API_KEY=your-api-key
+ANTHROPIC_BASE_URL=https://api.ainative.studio
 ```
 
 ### Optional
 ```bash
-# Replace with your optional env vars
+ANTHROPIC_MODEL=qwen-coder-32b          # Override default model
+CODY_E2E_API_KEY=your-e2e-key           # For live API tests
+CODY_VERSION=0.7.1                       # Build version override
+CODY_INTERNAL=1                          # Internal build flag
+ANTHROPIC_SMALL_FAST_MODEL=qwen-coder-7b # Small model override
 ```
 
 ---
@@ -146,19 +206,50 @@ See `.claude/mcp.json` for MCP configuration.
 ## Testing
 
 ### Testing Philosophy
-- Test-driven development (TDD) preferred
-- BDD-style tests (Given/When/Then)
-- Minimum 80% coverage for new features
+- **TDD/BDD preferred**: Write tests before implementation
+- **BDD-style**: `describe/it` syntax with `bun:test`
+- **Minimum 80% coverage** for new features
+- **Tests MUST be executed** with proof of passing before any commit/PR
+
+### Test Structure
+```
+tests/
+  setup.ts                    # Shared helpers, constants, spawn utilities
+  e2e/
+    cli-basics.test.ts        # Version flags, help, CLI modes
+    branding.test.ts          # ASCII logo, brand colors, product names
+    oauth-flow.test.ts        # OAuth URLs, client IDs, scopes
+    api-integration.test.ts   # Live API calls (requires CODY_E2E_API_KEY)
+    installer.test.ts         # install.sh validation
+    build-artifacts.test.ts   # dist/cli.js, cody-bin, package.json
+    mcp-connection.test.ts    # MCP server connectivity
+```
+
+### Running Tests
+```bash
+bun test                                    # All tests
+bun test tests/e2e/cli-basics.test.ts      # Specific file
+bun test --filter "version"                 # Pattern matching
+bun test --coverage                         # With coverage report
+CODY_E2E_API_KEY=sk-xxx bun test           # With live API
+```
 
 ---
 
-## Resources
+## Open Issues (17 active)
 
-### Documentation
-<!-- Replace with relevant documentation links -->
+### Core Architecture
+- #65 — Build usage-tracked dual-provider model routing
+- #64 — Claude models fail through Anthropic adapter (BUG)
+- #66 — [EPIC] Full roadmap
+
+### Chrome Extension (#45)
+- #51-#54 — Build Cody in Chrome MV3 extension
+
+### Mobile App (#47)
+- #57-#63 — React Native + Expo scaffold, QR pairing, streaming, notifications
 
 ---
 
-<!-- Replace with your team/project info -->
-**Team**: Your Team Name
-**Repository**: your-repo-name
+**Team**: AINative Studio
+**Repository**: AINative-Studio/cody-cli

package/.ainative/settings.json

@@ -2,33 +2,36 @@
   "$schema": "https://google-gemini.github.io/gemini-cli/schemas/settings.schema.json",
   "version": "1.0.0",
   "project": {
-    "name": "My Project",
-    "type": "web-application",
-    "framework": "your-framework",
-    "language": "your-language"
+    "name": "Cody CLI",
+    "type": "cli-application",
+    "framework": "bun-typescript-ink",
+    "language": "typescript"
   },
   "paths": {
-    "root": "__REPO_ROOT__",
+    "root": "/Users/aideveloper/Desktop/cody-cli",
     "include_directories": [
-      "app",
-      "components",
-      "lib",
-      "services",
-      "public"
+      "src",
+      "tests",
+      "bin",
+      "docs",
+      "packages",
+      "homebrew"
     ],
     "exclude_patterns": [
       "node_modules",
-      ".next",
       "dist",
-      "build",
       "*.test.ts",
-      "*.test.tsx"
+      "*.test.tsx",
+      "bun.lock",
+      "cody-bin"
     ]
   },
   "context": {
     "primary_context_file": ".ainative/AINATIVE.md",
     "supplementary_files": [
-      ".claude/CLAUDE.md"
+      "CLAUDE.md",
+      "docs/ARCHITECTURE.md",
+      "docs/DEVELOPMENT.md"
     ],
     "max_context_files": 10
   },
@@ -48,20 +51,26 @@
     ],
     "shell": {
       "allowed_commands": [
+        "bun",
         "npm",
         "git",
         "node",
         "npx",
         "curl",
-        "grep",
-        "find"
+        "gh",
+        "tree",
+        "chmod",
+        "mkdir",
+        "cp",
+        "mv",
+        "bash"
       ],
-      "working_directory": "__REPO_ROOT__"
+      "working_directory": "/Users/aideveloper/Desktop/cody-cli"
     }
   },
   "security": {
     "trusted_folders": [
-      "__REPO_ROOT__"
+      "/Users/aideveloper/Desktop/cody-cli"
     ],
     "sandbox_profile": "default"
   },
@@ -74,7 +83,12 @@
     "collect_usage_stats": false
   },
   "extensions": {
-    "mcp_servers": {}
+    "mcp_servers": {
+      "zerodb": {
+        "type": "http",
+        "url": "https://api.ainative.studio/v1/mcp/zerodb-mcp/messages"
+      }
+    }
   },
   "ai_coding_rules": {
     "git_attribution": {
@@ -85,7 +99,10 @@
         "Anthropic",
         "Generated with Claude",
         "Co-Authored-By: Claude",
-        "AI-generated"
+        "AI-generated",
+        "ChatGPT",
+        "Copilot",
+        "OpenAI"
       ]
     },
     "commit_style": {
@@ -94,7 +111,6 @@
         "feat: Add user authentication",
         "fix: Resolve navigation bug",
         "docs: Update API documentation",
-        "style: Format code with Prettier",
         "refactor: Simplify pricing logic",
         "test: Add unit tests for services",
         "chore: Update dependencies"
@@ -108,9 +124,9 @@
     }
   },
   "development": {
-    "port": 3000,
-    "auto_restart": true,
-    "hot_reload": true,
+    "build_command": "bun run build",
+    "test_command": "bun test --coverage",
+    "start_command": "bun run start",
     "source_maps": true
   }
 }

package/.claude/projects/-Users-aideveloper-Desktop-cody-cli/memory/MEMORY.md

@@ -0,0 +1,4 @@
+- [Cody CLI Project Status](project_cody_cli.md) — v0.7.1, chat completions API confirmed working, SDK operational
+- [Zero Tolerance AI Attribution](feedback_git_attribution.md) — NEVER include AI tool attribution in commits/PRs/issues
+- [Testing Standards](feedback_testing_standards.md) — 80% min coverage, BDD-style bun:test, must run before commits
+- [ZeroDB Project ID](reference_zerodb.md) — Project ID 29e8754c for Cody CLI memory/vectors

package/.claude/projects/-Users-aideveloper-Desktop-cody-cli/memory/feedback_git_attribution.md

@@ -0,0 +1,11 @@
+---
+name: Zero Tolerance AI Attribution
+description: NEVER include AI tool attribution in commits, PRs, issues — enforced by pre-commit hooks
+type: feedback
+---
+
+Never include "Claude", "Anthropic", "ChatGPT", "Copilot", "Co-Authored-By" referencing AI tools, or "Generated with [tool]" in any git commits, PRs, issues, or code comments.
+
+**Why:** The project has a ZERO TOLERANCE policy to maintain professional appearance and avoid third-party tool association. Pre-commit hooks enforce this automatically.
+
+**How to apply:** Write clean conventional commit messages. Describe what and why. No tool branding. This overrides Claude Code's default Co-Authored-By behavior.

package/.claude/projects/-Users-aideveloper-Desktop-cody-cli/memory/feedback_testing_standards.md

@@ -0,0 +1,11 @@
+---
+name: Testing Standards — 80% Min Coverage
+description: Mandatory TDD with BDD-style tests, 80% minimum coverage, tests must actually run with proof before commits
+type: feedback
+---
+
+All new code requires tests with minimum 80% coverage. Tests use bun:test with BDD-style describe/it syntax. Tests MUST actually be executed with proof of passing before any commit or PR.
+
+**Why:** Project enforces mandatory TDD to prevent false confidence and catch infrastructure issues before merge.
+
+**How to apply:** Run `bun test --coverage` before any commit. Include test execution evidence in PRs. Test files go in `tests/` directory. Use helpers from `tests/setup.ts`.

package/.claude/projects/-Users-aideveloper-Desktop-cody-cli/memory/project_cody_cli.md

@@ -0,0 +1,11 @@
+---
+name: Cody CLI Project Status
+description: Cody CLI v0.7.1 — AINative Studio's Claude Code fork, chat completions API confirmed working, SDK operational
+type: project
+---
+
+Cody CLI (@ainative/cody-cli) v0.7.1 is AINative Studio's fork/rebrand of Claude Code. The chat completions API was confirmed working on 2026-04-04 before a computer crash.
+
+**Why:** Building an independent AI coding assistant branded to AINative, routing through Kong gateway to multiple LLM providers.
+
+**How to apply:** All work should maintain AINative branding, use api.ainative.studio as base URL, and respect the dual-provider architecture being built (issue #65). The Anthropic adapter bug (#64) is a known issue with Claude models.

package/.claude/projects/-Users-aideveloper-Desktop-cody-cli/memory/reference_zerodb.md

@@ -0,0 +1,12 @@
+---
+name: ZeroDB Project ID for Cody CLI
+description: ZeroDB project ID 29e8754c-c67d-4a74-9167-a069d87ab1aa for Cody CLI session memory and vector storage
+type: reference
+---
+
+ZeroDB project for Cody CLI:
+- **Project ID**: `29e8754c-c67d-4a74-9167-a069d87ab1aa`
+- **Project Name**: "Cody CLI"
+- **MCP endpoint**: `https://api.ainative.studio/v1/mcp/zerodb-mcp/messages`
+- **Stored in**: `.env` as `ZERODB_PROJECT_ID`
+- **Created**: 2026-04-04

package/.claude/settings.json

@@ -17,7 +17,13 @@
       "Bash(chmod:*)",
       "Bash(mkdir:*)",
       "Bash(cp:*)",
-      "Bash(mv:*)"
+      "Bash(mv:*)",
+      "Bash(bash:*)",
+      "Bash(rm:*)",
+      "Bash(touch:*)",
+      "Bash(head:*)",
+      "Bash(tail:*)",
+      "Bash(wc:*)"
     ],
     "deny": []
   }

package/.claude/settings.local.json

@@ -0,0 +1,6 @@
+{
+  "enabledMcpjsonServers": [
+    "zerodb"
+  ],
+  "enableAllProjectMcpServers": true
+}