diff --git a/.gitignore b/.gitignore index 30a2d0428..845833b88 100644 --- a/.gitignore +++ b/.gitignore @@ -39,5 +39,15 @@ docs/.vitepress/dist/ # Claude Code worktrees .claude/worktrees/ +# Claude Code local settings +.claude/settings.local.json + # Managed worktree MCP fallback config /.mcp.json + +# Local Cargo cache config +.cargo/config.local.toml + +# Test result artifacts with hardcoded local paths +TEST_RESULTS*.md +/crates/llm-cli-wrapper/TEST_RESULTS*.md diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md new file mode 100644 index 000000000..908489564 --- /dev/null +++ b/CONTRIBUTING.md @@ -0,0 +1,181 @@ +# Contributing to Animus + +Thank you for your interest in contributing to Animus! This document provides guidelines and instructions for contributing. + +## Getting Started + +### Prerequisites + +- **Rust**: Animus is a Rust-only project. Install Rust via [rustup](https://rustup.rs/). +- **Node.js** (optional, for web UI work): v18+ +- **At least one AI coding CLI** for testing agent integration: + - `@anthropic-ai/claude-code` (recommended) + - `@openai/codex` + - `@google/gemini-cli` + +### Building Locally + +```bash +# Clone the repository +git clone https://github.com/samishukri/animus.git +cd animus + +# Build the project +cargo build + +# Run tests +cargo test + +# Format and lint +cargo fmt +cargo clippy +``` + +## Development Workflow + +### Before You Start + +1. **Check existing issues and PRs** to avoid duplicate work +2. **Fork the repository** on GitHub +3. **Create a feature branch** from `main`: + ```bash + git checkout -b feature/your-feature-name + ``` + +### Making Changes + +- Keep changes focused and minimal — one feature per PR +- Follow the existing code style (Rust conventions) +- Add tests for new functionality +- Update documentation if your changes affect user-facing behavior +- Reference the [CLAUDE.md](./CLAUDE.md) file for architecture landmarks and verification checks + +### Code Organization + +The workspace is organized into functional crates: + +- **Core orchestration**: `orchestrator-cli`, `orchestrator-core`, `orchestrator-config`, `orchestrator-store` +- **Runtime & agents**: `agent-runner`, `llm-cli-wrapper`, `workflow-runner-v2`, `orchestrator-daemon-runtime` +- **Web & API**: `orchestrator-web-server`, `orchestrator-web-api` +- **Utilities**: `orchestrator-providers`, `orchestrator-git-ops`, `orchestrator-notifications`, `protocol` + +### Running Tests + +```bash +# Test a specific crate +cargo test -p crate-name + +# Run all tests +cargo test --workspace + +# Run web UI tests +cd crates/orchestrator-web-server/web-ui +npm test +``` + +### Verifying Your Changes + +Before submitting a PR, verify your changes don't break anything: + +```bash +# Format check +cargo fmt --check + +# Lint check +cargo clippy --all-targets + +# Test all crates +cargo test --workspace + +# Web UI checks (if relevant) +cd crates/orchestrator-web-server/web-ui +npm run typecheck +npm run build +``` + +## Submitting Changes + +### Pull Request Process + +1. **Push your branch** to your fork: + ```bash + git push origin feature/your-feature-name + ``` + +2. **Create a Pull Request** with a clear title and description: + - Link any related issues using `Closes #123` + - Explain *why* the change is needed, not just *what* changed + - Include testing notes if applicable + +3. **Respond to feedback** from reviewers and update the PR as needed + +4. **Ensure CI passes** — all automated checks must pass before merging + +### Commit Guidelines + +- Write clear, descriptive commit messages +- Use conventional commit format when possible: `type(scope): description` + - `feat`: new feature + - `fix`: bug fix + - `refactor`: code restructuring + - `test`: test additions or updates + - `docs`: documentation updates + - `chore`: maintenance tasks + +Example: +``` +feat(cli): add --dry-run flag to task create command + +Allows users to preview task creation without persisting state. +``` + +## Documentation + +- **CLI changes**: Update `docs/reference/cli/index.md` +- **MCP tools**: Update `docs/reference/mcp-tools.md` and `docs/guides/agents.md` +- **Configuration**: Update relevant docs in `docs/reference/` +- **README**: Keep it current with major feature additions + +## Code Standards + +### Rust Style + +- Use `cargo fmt` for formatting — this is enforced in CI +- Follow Clippy suggestions — address warnings before submitting +- Write idiomatic Rust code + +### Web UI Standards (TypeScript/React) + +- Use React 18 best practices +- Follow component patterns established in the codebase +- Test changes with responsive and accessibility checks + +### State Management + +- Treat AO-managed state (in `~/.ao/`) as immutable except through CLI commands +- Use service APIs rather than direct file manipulation +- Preserve backward compatibility where possible + +## Reporting Bugs + +When reporting bugs, please include: + +1. **Environment**: OS, Rust version, Animus version +2. **Steps to reproduce**: Clear, minimal example +3. **Expected behavior**: What should happen +4. **Actual behavior**: What actually happens +5. **Logs**: Output of `animus doctor` and relevant error messages + +## Questions or Feedback? + +- **Discussions**: Use GitHub Discussions for questions +- **Issues**: File issues for bugs or feature requests +- **Security**: For security issues, email security@example.com (do not open public issues) + +## License + +By contributing to Animus, you agree that your contributions will be licensed under the same [Elastic License 2.0 (ELv2)](LICENSE) as the project. + +--- + +Thank you for contributing to Animus! diff --git a/README.md b/README.md index e5556921d..5a67261a5 100644 --- a/README.md +++ b/README.md @@ -1,17 +1,17 @@
-![header](https://capsule-render.vercel.app/api?type=waving&color=0:0d1117,50:161b22,100:1f6feb&height=200§ion=header&text=AO&fontSize=90&fontColor=f0f6fc&animation=fadeIn&fontAlignY=35&desc=Autonomous%20Agent%20Orchestrator&descAlignY=55&descSize=22&descColor=8b949e) +![header](https://capsule-render.vercel.app/api?type=waving&color=0:0d1117,50:161b22,100:1f6feb&height=200§ion=header&text=Animus&fontSize=90&fontColor=f0f6fc&animation=fadeIn&fontAlignY=35&desc=Autonomous%20Agent%20Orchestrator&descAlignY=55&descSize=22&descColor=8b949e)
-[![Typing SVG](https://readme-typing-svg.demolab.com?font=JetBrains+Mono&weight=500&size=20&duration=3000&pause=1500&color=58A6FF¢er=true&vCenter=true&multiline=true&repeat=true&random=false&width=700&height=80&lines=Define+your+engineering+team+as+YAML.;Dispatch+tasks+to+AI+agents+across+isolated+worktrees.;Review%2C+merge%2C+and+ship+%E2%80%94+while+you+sleep.)](https://github.com/launchapp-dev/ao) +[![Typing SVG](https://readme-typing-svg.demolab.com?font=JetBrains+Mono&weight=500&size=20&duration=3000&pause=1500&color=58A6FF¢er=true&vCenter=true&multiline=true&repeat=true&random=false&width=700&height=80&lines=Define+your+engineering+team+as+YAML.;Dispatch+tasks+to+AI+agents+across+isolated+worktrees.;Review%2C+merge%2C+and+ship+%E2%80%94+while+you+sleep.)](https://github.com/samishukri/animus)


-Release +Release   Rust   @@ -24,7 +24,7 @@ ## Install ```bash -curl -fsSL https://raw.githubusercontent.com/launchapp-dev/ao/main/install.sh | bash +curl -fsSL https://raw.githubusercontent.com/samishukri/animus/main/install.sh | bash ``` The upstream installer currently targets macOS. On Linux and Windows, use a release archive or build from source. @@ -34,10 +34,10 @@ The upstream installer currently targets macOS. On Linux and Windows, use a rele ```bash # Specific version -AO_VERSION=v0.0.11 curl -fsSL https://raw.githubusercontent.com/launchapp-dev/ao/main/install.sh | bash +ANIMUS_VERSION=v0.0.11 curl -fsSL https://raw.githubusercontent.com/samishukri/animus/main/install.sh | bash # Custom directory -AO_INSTALL_DIR=/usr/local/bin curl -fsSL https://raw.githubusercontent.com/launchapp-dev/ao/main/install.sh | bash +ANIMUS_INSTALL_DIR=/usr/local/bin curl -fsSL https://raw.githubusercontent.com/samishukri/animus/main/install.sh | bash ``` @@ -57,15 +57,15 @@ npm install -g @google/gemini-cli # Gemini --- -## What is AO? +## What is Animus? -AO turns a single YAML file into an autonomous software delivery pipeline. +Animus turns a single YAML file into an autonomous software delivery pipeline. -You define agents, wire them into phases, compose phases into workflows, schedule everything with cron — and AO's daemon handles the rest: dispatching tasks to AI agents in isolated git worktrees, managing quality gates, and merging the results. +You define agents, wire them into phases, compose phases into workflows, schedule everything with cron — and Animus's daemon handles the rest: dispatching tasks to AI agents in isolated git worktrees, managing quality gates, and merging the results. ``` ┌──────────────────────────────────────────────────┐ - │ AO Daemon (Rust) │ + │ Animus Daemon (Rust) │ │ │ ┌────────┐ │ ┌───────────┐ ┌───────────┐ ┌────────┐ │ ┌────────┐ │ Tasks │───▶│───▶│ Dispatch │───▶│ Agents │───▶│ Phases │─│──▶│ PRs │ @@ -86,13 +86,13 @@ You define agents, wire them into phases, compose phases into workflows, schedul ```bash cd your-project # any git repo -ao doctor # check prerequisites -ao setup # initialize .ao/ +animus doctor # check prerequisites +animus setup # initialize .ao/ -ao task create --title "Add rate limiting" --task-type feature --priority high -ao workflow run --task-id TASK-001 # run once +animus task create --title "Add rate limiting" --task-type feature --priority high +animus workflow run --task-id TASK-001 # run once -ao daemon start --autonomous # or go fully autonomous +animus daemon start --autonomous # or go fully autonomous ``` --- @@ -112,7 +112,7 @@ agents: default: model: claude-sonnet-4-6 tool: claude - mcp_servers: ["ao", "context7"] + mcp_servers: ["animus", "context7"] work-planner: system_prompt: | @@ -264,11 +264,11 @@ Every task gets its own git worktree. Agents work in parallel on separate branch ## Claude Code Integration -Install [**AO Skills**](https://github.com/launchapp-dev/ao-skills) for deep AO integration inside Claude Code: +Install [**Animus Skills**](https://github.com/samishukri/animus-skills) for deep Animus integration inside Claude Code: ```bash -git clone https://github.com/launchapp-dev/ao-skills.git ~/ao-skills -claude --plugin-dir ~/ao-skills +git clone https://github.com/samishukri/animus-skills.git ~/animus-skills +claude --plugin-dir ~/animus-skills ``` @@ -279,7 +279,7 @@ claude --plugin-dir ~/ao-skills | Command | What it does | |:---|:---| -| `/setup-ao` | Initialize AO in your project | +| `/setup-animus` | Initialize Animus in your project | | `/getting-started` | Install, concepts, first task | | `/workflow-authoring` | Write custom YAML workflows | | `/pack-authoring` | Build workflow packs | @@ -298,7 +298,7 @@ claude --plugin-dir ~/ao-skills | `daemon-operations` | Daemon monitoring and troubleshooting | | `workflow-patterns` | Patterns from 150+ autonomous PRs | | `agent-personas` | PO, architect, auditor agents | -| `mcp-tools` | Complete `ao.*` tool reference | +| `mcp-tools` | Complete `animus.*` tool reference | @@ -309,25 +309,25 @@ claude --plugin-dir ~/ao-skills ## CLI ``` -ao task Create, list, update, prioritize tasks -ao workflow Run and manage multi-phase workflows -ao daemon Start/stop the autonomous scheduler -ao queue Inspect and manage the dispatch queue -ao agent Control agent runner processes -ao output Stream and inspect agent output -ao doctor Health checks and auto-remediation -ao setup Interactive project initialization -ao requirements Manage product requirements -ao mcp Start AO as an MCP server -ao web Launch the embedded web dashboard -ao status Project overview at a glance +animus task Create, list, update, prioritize tasks +animus workflow Run and manage multi-phase workflows +animus daemon Start/stop the autonomous scheduler +animus queue Inspect and manage the dispatch queue +animus agent Control agent runner processes +animus output Stream and inspect agent output +animus doctor Health checks and auto-remediation +animus setup Interactive project initialization +animus requirements Manage product requirements +animus mcp Start Animus as an MCP server +animus web Launch the embedded web dashboard +animus status Project overview at a glance ``` --- ## Architecture -AO is a Rust-only workspace with 17 crates. The major crates are: +Animus is a Rust-only workspace with 17 crates. The major crates are: - `orchestrator-cli` - CLI commands and dispatch - `orchestrator-core` - services, state, and workflow lifecycle @@ -386,17 +386,17 @@ This project is licensed under the [Elastic License 2.0 (ELv2)](LICENSE). You ma **Update** ```bash -curl -fsSL https://raw.githubusercontent.com/launchapp-dev/ao/main/install.sh | bash +curl -fsSL https://raw.githubusercontent.com/samishukri/animus/main/install.sh | bash ``` **Uninstall** ```bash -rm -f ~/.local/bin/ao \ +rm -f ~/.local/bin/animus \ ~/.local/bin/agent-runner \ ~/.local/bin/llm-cli-wrapper \ - ~/.local/bin/ao-oai-runner \ - ~/.local/bin/ao-workflow-runner + ~/.local/bin/animus-oai-runner \ + ~/.local/bin/animus-workflow-runner ```
diff --git a/crates/agent-runner/src/runner/mcp_policy/tests.rs b/crates/agent-runner/src/runner/mcp_policy/tests.rs index 59f7c8984..a47acf1b0 100644 --- a/crates/agent-runner/src/runner/mcp_policy/tests.rs +++ b/crates/agent-runner/src/runner/mcp_policy/tests.rs @@ -47,10 +47,10 @@ fn native_mcp_policy_rejects_unknown_cli_when_enforced() { enabled: true, endpoint: None, stdio: Some(McpStdioConfig { - command: "/Users/samishukri/ao-cli/target/debug/ao".to_string(), + command: "/path/to/ao/target/debug/ao".to_string(), args: vec![ "--project-root".to_string(), - "/Users/samishukri/ao-cli".to_string(), + "/path/to/project".to_string(), "mcp".to_string(), "serve".to_string(), ], @@ -173,10 +173,10 @@ fn native_mcp_policy_preserves_primary_server_when_additional_server_name_collid enabled: true, endpoint: None, stdio: Some(McpStdioConfig { - command: "/Users/samishukri/ao-cli/target/debug/ao".to_string(), + command: "/path/to/ao/target/debug/ao".to_string(), args: vec![ "--project-root".to_string(), - "/Users/samishukri/ao-cli".to_string(), + "/path/to/project".to_string(), "mcp".to_string(), "serve".to_string(), ], @@ -209,13 +209,13 @@ fn native_mcp_policy_preserves_primary_server_when_additional_server_name_collid assert_eq!( parsed.pointer("/mcpServers/ao/command").and_then(serde_json::Value::as_str), - Some("/Users/samishukri/ao-cli/target/debug/ao") + Some("/path/to/ao/target/debug/ao") ); assert_eq!( parsed.pointer("/mcpServers/ao/args").and_then(serde_json::Value::as_array).cloned(), Some(vec![ serde_json::Value::String("--project-root".to_string()), - serde_json::Value::String("/Users/samishukri/ao-cli".to_string()), + serde_json::Value::String("/path/to/project".to_string()), serde_json::Value::String("mcp".to_string()), serde_json::Value::String("serve".to_string()), ]) @@ -261,10 +261,10 @@ fn codex_native_lockdown_sets_stdio_transport_when_configured() { apply_codex_native_mcp_lockdown( &mut args, McpServerTransport::Stdio { - command: "/Users/samishukri/ao-cli/target/debug/ao", + command: "/path/to/ao/target/debug/ao", args: &[ "--project-root".to_string(), - "/Users/samishukri/ao-cli".to_string(), + "/path/to/project".to_string(), "mcp".to_string(), "serve".to_string(), ], @@ -275,9 +275,9 @@ fn codex_native_lockdown_sets_stdio_transport_when_configured() { ); let joined = args.join(" "); - assert!(joined.contains("mcp_servers.ao.command=\"/Users/samishukri/ao-cli/target/debug/ao\"")); + assert!(joined.contains("mcp_servers.ao.command=\"/path/to/ao/target/debug/ao\"")); assert!( - joined.contains("mcp_servers.ao.args=[\"--project-root\", \"/Users/samishukri/ao-cli\", \"mcp\", \"serve\"]") + joined.contains("mcp_servers.ao.args=[\"--project-root\", \"/path/to/project\", \"mcp\", \"serve\"]") ); assert!(joined.contains("mcp_servers.ao.enabled=true")); } @@ -294,10 +294,10 @@ fn native_mcp_policy_sets_gemini_system_settings_path_for_stdio_transport() { enabled: true, endpoint: None, stdio: Some(McpStdioConfig { - command: "/Users/samishukri/ao-cli/target/debug/ao".to_string(), + command: "/path/to/ao/target/debug/ao".to_string(), args: vec![ "--project-root".to_string(), - "/Users/samishukri/ao-cli".to_string(), + "/path/to/project".to_string(), "mcp".to_string(), "serve".to_string(), ], @@ -377,10 +377,10 @@ fn native_mcp_policy_sets_opencode_local_mcp_command_array() { enabled: true, endpoint: None, stdio: Some(McpStdioConfig { - command: "/Users/samishukri/ao-cli/target/debug/ao".to_string(), + command: "/path/to/ao/target/debug/ao".to_string(), args: vec![ "--project-root".to_string(), - "/Users/samishukri/ao-cli".to_string(), + "/path/to/project".to_string(), "mcp".to_string(), "serve".to_string(), ], @@ -403,7 +403,7 @@ fn native_mcp_policy_sets_opencode_local_mcp_command_array() { assert_eq!(parsed.pointer("/mcp/ao/type").and_then(serde_json::Value::as_str), Some("local")); assert_eq!( parsed.pointer("/mcp/ao/command/0").and_then(serde_json::Value::as_str), - Some("/Users/samishukri/ao-cli/target/debug/ao") + Some("/path/to/ao/target/debug/ao") ); assert_eq!(parsed.pointer("/mcp/ao/command/4").and_then(serde_json::Value::as_str), Some("serve")); assert!(parsed.pointer("/mcp/ao/args").is_none()); @@ -428,12 +428,12 @@ fn native_mcp_policy_inserts_oai_runner_mcp_config_after_run_subcommand() { enabled: true, endpoint: None, stdio: Some(McpStdioConfig { - command: "/Users/samishukri/ao-cli/target/debug/ao".to_string(), + command: "/path/to/ao/target/debug/ao".to_string(), args: vec![ "mcp".to_string(), "serve".to_string(), "--project-root".to_string(), - "/Users/samishukri/ao-cli".to_string(), + "/path/to/project".to_string(), ], }), agent_id: "ao".to_string(), diff --git a/crates/llm-cli-wrapper/TEST_RESULTS.md b/crates/llm-cli-wrapper/TEST_RESULTS.md deleted file mode 100644 index 7bc7499b7..000000000 --- a/crates/llm-cli-wrapper/TEST_RESULTS.md +++ /dev/null @@ -1,109 +0,0 @@ -# CLI Wrapper Test Results - -## ✅ Test Run: Success! - -### Discovery Test -**Command**: `./target/release/llm-cli-wrapper discover` - -**Result**: ✅ PASS -``` -✓ Found 3 CLI(s) -``` - -**CLIs Discovered**: -- ✅ Claude Code at `/Users/samishukri/.local/bin/claude` -- ✅ OpenAI Codex at `/Users/samishukri/.bun/bin/codex` -- ✅ Google Gemini CLI at `/Users/samishukri/.nvm/versions/node/v22.17.0/bin/gemini` -- ⚠️ Aider not found in PATH - ---- - -### List Test -**Command**: `./target/release/llm-cli-wrapper list` - -**Result**: ✅ PASS -``` -Installed CLIs: -──────────────────────────────────────────────────────────── -Claude Code ⚠ Not Authenticated -OpenAI Codex ⚠ Not Authenticated -Google Gemini CLI ⚠ Not Authenticated -``` - -**Note**: CLIs are detected but not authenticated (no API keys set) - ---- - -### Health Check Test -**Command**: `./target/release/llm-cli-wrapper health` - -**Result**: ✅ PASS (Detection works, auth needs setup) -``` -Running health checks... -──────────────────────────────────────────────────────────── -✗ UNHEALTHY OpenAI Codex (0ms) - CLI is not authenticated -✗ UNHEALTHY Claude Code (0ms) - CLI is not authenticated -✗ UNHEALTHY Google Gemini CLI (0ms) - CLI is not authenticated -``` - -**Note**: Health checks correctly identify missing authentication - ---- - -## Test Summary - -| Test | Status | Details | -|------|--------|---------| -| CLI Discovery | ✅ PASS | Found 3 CLIs successfully | -| CLI List | ✅ PASS | Lists all discovered CLIs | -| Health Check | ✅ PASS | Correctly detects auth status | -| Info Command | ✅ PASS | Shows CLI capabilities | - -## Features Verified - -✅ **Auto-discovery**: Automatically finds CLIs in PATH -✅ **Multi-CLI support**: Works with Claude, Codex, Gemini -✅ **Status detection**: Identifies authentication state -✅ **Logging**: Clear, colored output -✅ **Error handling**: Graceful handling of missing CLIs - -## Authentication Setup Needed - -To make CLIs fully functional, set these environment variables: - -```bash -# For Claude -export ANTHROPIC_API_KEY="your-key-here" - -# For Codex -codex login -# OR -export OPENAI_API_KEY="your-key-here" - -# For Gemini -export GEMINI_API_KEY="your-key-here" -# OR -export GOOGLE_APPLICATION_CREDENTIALS="/path/to/credentials.json" -``` - -## Next Steps - -1. ✅ Discovery works - Can find installed CLIs -2. ✅ Listing works - Can show all CLIs with status -3. ✅ Health checks work - Can verify CLI state -4. ⚠️ Auth needed - Set up API keys to test full functionality -5. 🚧 Run full test suite - `./target/release/llm-cli-wrapper test` - -## Conclusion - -The CLI wrapper is **fully functional** and successfully: -- Discovers installed CLIs automatically -- Detects authentication status -- Provides detailed CLI information -- Shows clear, colored output -- Handles missing CLIs gracefully - -**Status**: Ready for use! 🎉 diff --git a/crates/llm-cli-wrapper/TEST_RESULTS_COMPLETE.md b/crates/llm-cli-wrapper/TEST_RESULTS_COMPLETE.md deleted file mode 100644 index d33591353..000000000 --- a/crates/llm-cli-wrapper/TEST_RESULTS_COMPLETE.md +++ /dev/null @@ -1,305 +0,0 @@ -# CLI Wrapper - Complete Test Results - -## ✅ All CLIs Working - Production Ready! - -**Date**: 2026-02-01 -**Status**: 🎉 4 CLIs Discovered and Tested Successfully - ---- - -## Executive Summary - -The CLI wrapper successfully discovers, authenticates, and executes commands on **4 different AI coding assistants**: - -✅ **Claude Code** - Anthropic's official CLI -✅ **OpenAI Codex** - OpenAI's coding assistant -✅ **Google Gemini CLI** - Google's AI assistant -✅ **OpenCode** - Open-source multi-model CLI *(newly added)* - -All CLIs pass health checks and basic verification tests. - ---- - -## Discovered CLIs - -| CLI | Version | Path | Health | Tests | -|-----|---------|------|--------|-------| -| **Claude Code** | 2.1.29 | `/Users/samishukri/.local/bin/claude` | ✅ Healthy (851ms) | 2/2 PASS | -| **OpenAI Codex** | 0.92.0 | `/Users/samishukri/.bun/bin/codex` | ✅ Healthy (133ms) | 2/2 PASS | -| **Google Gemini** | 0.26.0 | `/Users/samishukri/.nvm/versions/node/v22.17.0/bin/gemini` | ✅ Healthy (4144ms) | 2/2 PASS | -| **OpenCode** | Latest | `/Users/samishukri/.opencode/bin/opencode` | ✅ Healthy (1582ms) | 2/2 PASS | - ---- - -## Test Results by CLI - -### ✅ Claude Code -**Command Format**: `claude "prompt"` - -**Health Check**: ✅ PASS (851ms) -- Version detection working -- Authentication verified -- Ready for use - -**Basic Tests**: ✅ 2/2 PASS -- Simple greeting: ✓ PASS (3651ms) -- Simple math: ✓ PASS (3493ms) - -**Notes**: Fastest response times, excellent for interactive use - ---- - -### ✅ OpenAI Codex -**Command Format**: `codex exec --skip-git-repo-check "prompt"` - -**Health Check**: ✅ PASS (133ms) -- Version detection working -- Authentication verified -- Ready for use - -**Basic Tests**: ✅ 2/2 PASS -- Simple greeting: ✓ PASS (5762ms) -- Simple math: ✓ PASS (4652ms) - -**Notes**: -- Requires `--skip-git-repo-check` flag for non-repo directories -- Supports advanced reasoning with high effort mode -- Excellent for complex coding tasks - ---- - -### ✅ Google Gemini CLI -**Command Format**: `gemini -p "prompt"` - -**Health Check**: ✅ PASS (4144ms) -- Version detection working -- Authentication verified -- Ready for use - -**Basic Tests**: ✅ 2/2 PASS -- Simple greeting: ✓ PASS (8454ms) -- Simple math: ✓ PASS (8151ms) - -**Notes**: -- Longest response times (high reasoning effort) -- Massive context window (1M tokens) -- Best for large codebases - ---- - -### ✅ OpenCode (New!) -**Command Format**: `opencode run "prompt"` - -**Health Check**: ✅ PASS (1582ms) -- Version detection working -- Authentication verified -- Ready for use - -**Basic Tests**: ✅ 2/2 PASS -- Simple greeting: ✓ PASS (3724ms) -- Simple math: ✓ PASS (5971ms) - -**Notes**: -- Open-source, privacy-focused -- Supports 75+ model providers -- Can use local models via Ollama -- Great for sensitive projects - -**Why OpenCode?** -OpenCode is a standout addition because it: -- Doesn't store your code or context (privacy-first) -- Supports multiple providers (OpenAI, Anthropic, local models) -- Switch models mid-session without losing context -- Terminal-native with polished TUI -- Integrates with language servers for code intelligence - ---- - -## Issues Fixed in This Session - -### 1. ✅ Codex Not Working in Test Directories -**Problem**: Codex failed with "Not inside a trusted directory" error -**Solution**: Added `--skip-git-repo-check` flag to exec command -**File**: `src/cli/codex.rs:53` - -### 2. ✅ OpenCode Discovery and Integration -**Added**: -- OpenCode CLI type to enum (src/cli/types.rs) -- OpenCode implementation (src/cli/opencode.rs) -- Registry integration (src/cli/registry.rs) -- Command parser support (src/main.rs) - -**Command Format**: Uses `opencode run "message"` for execution - ---- - -## Performance Comparison - -| CLI | Health Check | Greeting Test | Math Test | Avg Response | -|-----|-------------|---------------|-----------|--------------| -| Claude | 851ms | 3651ms | 3493ms | **3572ms** ⚡ | -| OpenCode | 1582ms | 3724ms | 5971ms | **4848ms** | -| Codex | 133ms | 5762ms | 4652ms | **5207ms** | -| Gemini | 4144ms | 8454ms | 8151ms | **8303ms** | - -**Fastest**: Claude Code (3.6s average) -**Slowest**: Gemini (8.3s average - but handles largest context) - ---- - -## CLI Capabilities Matrix - -| Capability | Claude | Codex | Gemini | OpenCode | -|-----------|--------|-------|--------|----------| -| File Editing | ✅ | ✅ | ✅ | ✅ | -| Streaming | ✅ | ✅ | ✅ | ✅ | -| Tool Use | ✅ | ✅ | ✅ | ✅ | -| Vision | ✅ | ❌ | ✅ | ❌ | -| Long Context | ✅ 200K | ✅ 128K | ✅ 1M | ✅ 200K+ | -| Local Models | ❌ | ❌ | ❌ | ✅ | -| Multi-Provider | ❌ | ❌ | ❌ | ✅ | - ---- - -## Integration Readiness - -### ✅ Ready for Agent Orchestrator - -The CLI wrapper can now: - -1. **CLI Selection** - - Automatically discover all 4 installed CLIs - - Check health before task assignment - - Select best CLI based on task requirements - -2. **Health Monitoring** - - Pre-execution health checks - - Performance tracking - - Automatic fallback to alternative CLIs - -3. **Task Routing** - - Vision tasks → Claude or Gemini - - Large context → Gemini (1M tokens) - - Privacy-sensitive → OpenCode (local models) - - General coding → Claude (fastest) - - Complex reasoning → Codex - -4. **Quality Gates** - - Verify CLI availability before workflow - - Validate output quality - - Track execution times - ---- - -## Commands Reference - -### Discovery -```bash -./target/release/llm-cli-wrapper discover -# Output: ✓ Found 4 CLI(s) -``` - -### List All CLIs -```bash -./target/release/llm-cli-wrapper list -# Shows all CLIs with authentication status -``` - -### Health Checks -```bash -# All CLIs -./target/release/llm-cli-wrapper health - -# Specific CLI -./target/release/llm-cli-wrapper health opencode -``` - -### Run Tests -```bash -# Test specific CLI -./target/release/llm-cli-wrapper test claude --suite basic -./target/release/llm-cli-wrapper test codex --suite basic -./target/release/llm-cli-wrapper test gemini --suite basic -./target/release/llm-cli-wrapper test opencode --suite basic - -# All CLIs -./target/release/llm-cli-wrapper test --suite basic -``` - -### CLI Information -```bash -./target/release/llm-cli-wrapper info opencode -# Shows version, capabilities, and status -``` - ---- - -## Next Steps - -### Immediate -- ✅ All 4 CLIs discovered and tested -- ✅ Health checks passing -- ✅ Basic verification complete -- ✅ Ready for integration - -### Future Enhancements -1. **Additional Test Suites** - - File operations (read/write/edit) - - Code generation - - Multi-file refactoring - -2. **Advanced Features** - - Performance benchmarking - - Cost tracking (API usage) - - Automatic model selection based on task - - Parallel execution tests - -3. **Integration** - - Import into agent-runner daemon - - Add to workflow executor - - Implement in PM/EM loops - ---- - -## OpenCode Resources - -Based on web research, OpenCode is a significant addition: - -- **Installation**: `curl -fsSL https://opencode.ai/install | bash` -- **GitHub**: [opencode-ai/opencode](https://github.com/opencode-ai/opencode) -- **Documentation**: [opencode.ai/docs/cli](https://opencode.ai/docs/cli/) -- **Comparison**: [OpenCode vs Claude Code](https://www.builder.io/blog/opencode-vs-claude-code) - -**Key Differentiator**: Privacy-focused, open-source alternative with multi-provider support and local model capability via Ollama. - -### Sources: -- [CLI | OpenCode](https://opencode.ai/docs/cli/) -- [GitHub - opencode-ai/opencode](https://github.com/opencode-ai/opencode) -- [OpenCode CLI Guide 2026](https://yuv.ai/learn/opencode-cli) -- [OpenCode vs Claude Code](https://www.builder.io/blog/opencode-vs-claude-code) -- [Top 5 CLI Coding Agents in 2026](https://dev.to/lightningdev123/top-5-cli-coding-agents-in-2026-3pia) - ---- - -## Conclusion - -🎉 **CLI Wrapper Status: Production Ready** - -**Achievements**: -- ✅ 4 CLIs discovered automatically -- ✅ All health checks passing -- ✅ All basic tests passing (8/8) -- ✅ Codex fixed for non-repo directories -- ✅ OpenCode added with full support -- ✅ Ready for Agent Orchestrator integration - -**Total CLIs Supported**: 4 active + 3 planned (Aider, Cursor, Cline) - -The system can now intelligently route tasks to the most appropriate CLI based on: -- Task complexity -- Context window requirements -- Privacy needs -- Performance requirements -- Feature requirements (vision, tool use, etc.) - -Ready for production use in autonomous agent workflows! 🚀 diff --git a/crates/llm-cli-wrapper/TEST_RESULTS_FINAL.md b/crates/llm-cli-wrapper/TEST_RESULTS_FINAL.md deleted file mode 100644 index d0d7f7fb2..000000000 --- a/crates/llm-cli-wrapper/TEST_RESULTS_FINAL.md +++ /dev/null @@ -1,191 +0,0 @@ -# CLI Wrapper - Final Test Results - -## ✅ All Tests Passing! - -**Date**: 2026-02-01 -**Status**: Production Ready - ---- - -## Test Summary - -| Test Category | Status | Details | -|--------------|---------|---------| -| Discovery | ✅ PASS | Successfully found 3 CLIs | -| Health Checks | ✅ PASS | All CLIs healthy and authenticated | -| Basic Verification | ✅ PASS | Claude & Gemini pass all tests | -| CLI Execution | ✅ PASS | Successfully executes AI prompts | - ---- - -## Discovered CLIs - -### ✅ Claude Code -- **Path**: `/Users/samishukri/.local/bin/claude` -- **Version**: 2.1.29 (Claude Code) -- **Status**: Healthy (861ms response time) -- **Tests**: 2/2 passed - - ✓ Simple greeting test (3651ms) - - ✓ Simple math test (3493ms) - -### ✅ Google Gemini CLI -- **Path**: `/Users/samishukri/.nvm/versions/node/v22.17.0/bin/gemini` -- **Version**: 0.26.0 -- **Status**: Healthy (3581ms response time) -- **Tests**: 2/2 passed - - ✓ Simple greeting test (8454ms) - - ✓ Simple math test (8151ms) - -### ✅ OpenAI Codex -- **Path**: `/Users/samishukri/.bun/bin/codex` -- **Version**: codex-cli 0.92.0 -- **Status**: Healthy (125ms response time) -- **Tests**: Requires workspace context (expected behavior) - -### ⚠️ Aider -- **Status**: Not found in PATH -- **Note**: Not installed on this system - ---- - -## Issues Fixed - -### 1. ✅ Authentication Detection -**Problem**: CLIs were showing as "Not Authenticated" despite being logged in -**Solution**: Changed authentication check from environment variable validation to version command execution -**File**: `src/cli/{claude,codex,gemini}.rs` - -### 2. ✅ Incorrect CLI Commands -**Problem**: Wrong subcommands being used for Codex and Gemini -**Solution**: -- Codex: Changed from `codex run` to `codex exec` -- Gemini: Changed from `gemini chat` to `gemini -p` -**Files**: `src/cli/codex.rs`, `src/cli/gemini.rs` - -### 3. ✅ Working Directory Not Found -**Problem**: Test execution failing with "No such file or directory" error -**Root Cause**: Temp directory `/var/folders/.../llm-cli-wrapper-tests` didn't exist -**Solution**: Create test workspace directory before running tests -**File**: `src/main.rs:126-129` - -### 4. ✅ Case-Sensitive Output Validation -**Problem**: Tests failing because AI responses use "Hello" instead of "hello" -**Solution**: Made output validation case-insensitive -**File**: `src/tester/test_runner.rs:97-106` - ---- - -## Architecture Improvements - -### Error Handling -- Added executable path existence check before spawning -- Added working directory validation -- Improved error messages for debugging - -### Logging -- Added debug statements for command execution -- Track spawn success/failure -- Monitor working directory changes - -### Test Framework -- Case-insensitive output matching -- Proper working directory management -- Timeout handling for long-running AI operations - ---- - -## Commands Verified - -### Discovery -```bash -./target/release/llm-cli-wrapper discover -``` -✅ Successfully discovers all installed CLIs - -### List -```bash -./target/release/llm-cli-wrapper list -``` -✅ Shows all CLIs with authentication status - -### Health Checks -```bash -./target/release/llm-cli-wrapper health -``` -✅ All CLIs report healthy status - -### Individual CLI Tests -```bash -./target/release/llm-cli-wrapper test claude --suite basic -./target/release/llm-cli-wrapper test gemini --suite basic -``` -✅ Both pass all test cases - ---- - -## Performance Metrics - -| CLI | Health Check | Greeting Test | Math Test | -|-----|-------------|---------------|-----------| -| Claude | 861ms | 3651ms | 3493ms | -| Gemini | 3581ms | 8454ms | 8151ms | -| Codex | 125ms | N/A* | N/A* | - -*Codex requires workspace context for exec mode - ---- - -## Integration Points - -This CLI wrapper is ready for integration with the Agent Orchestrator: - -### ✅ Workflow Executor -- Can verify CLI availability before task assignment -- Health check before executing workflow steps -- Detect and report CLI failures - -### ✅ Engineering Manager Loop -- Monitor CLI availability -- Track CLI performance metrics -- Automatic fallback to alternative CLIs - -### ✅ Product Manager Loop -- Assess CLI capabilities -- Match tasks to appropriate CLIs -- Evaluate CLI suitability for requirements - ---- - -## Next Steps - -1. **Production Deployment** - - ✅ All tests passing - - ✅ Error handling robust - - ✅ Authentication working - - Ready for use in daemon - -2. **Additional Test Suites** - - File operations test suite - - Code generation test suite - - Multi-file editing tests - -3. **Integration** - - Import into agent-runner daemon - - Use for CLI selection in workflows - - Add to quality gates - ---- - -## Conclusion - -The CLI wrapper is **production ready** with all core functionality working: - -- ✅ Auto-discovery of installed CLIs -- ✅ Health monitoring -- ✅ Authentication detection -- ✅ Command execution -- ✅ Output validation -- ✅ Error handling -- ✅ Performance tracking - -**Status**: Ready for integration with the Agent Orchestrator system. diff --git a/crates/llm-cli-wrapper/test_spawn.rs b/crates/llm-cli-wrapper/test_spawn.rs index 55b6665b6..fe7d30dd2 100644 --- a/crates/llm-cli-wrapper/test_spawn.rs +++ b/crates/llm-cli-wrapper/test_spawn.rs @@ -3,7 +3,7 @@ use std::process::Stdio; #[tokio::main] async fn main() { - let path = "/Users/samishukri/.local/bin/claude"; + let path = "/usr/local/bin/claude"; let args = vec!["Say hello"]; println!("Testing spawn with path: {:?}", path); diff --git a/crates/orchestrator-cli/src/cli_types/cloud_types.rs b/crates/orchestrator-cli/src/cli_types/cloud_types.rs index 3920a9f57..40a418c10 100644 --- a/crates/orchestrator-cli/src/cli_types/cloud_types.rs +++ b/crates/orchestrator-cli/src/cli_types/cloud_types.rs @@ -2,6 +2,8 @@ use clap::{Parser, Subcommand}; #[derive(Debug, Subcommand)] pub(crate) enum CloudCommand { + /// Authenticate with animus cloud using device auth flow. + Login(CloudLoginArgs), /// Configure the sync server connection for this project. Setup(CloudSetupArgs), /// Push local tasks and requirements to the sync server. @@ -19,6 +21,14 @@ pub(crate) enum CloudCommand { }, } +#[derive(Debug, Parser)] +pub(crate) struct CloudLoginArgs { + #[arg(long, help = "Animus cloud server URL (defaults to https://api.animus.cloud)")] + pub(crate) server: Option, + #[arg(long, help = "Skip opening browser (print URL instead)")] + pub(crate) no_browser: bool, +} + #[derive(Debug, Subcommand)] pub(crate) enum DeployCommand { /// Create a new deployment diff --git a/crates/orchestrator-cli/src/services/cloud.rs b/crates/orchestrator-cli/src/services/cloud.rs index 5a2558439..451378f53 100644 --- a/crates/orchestrator-cli/src/services/cloud.rs +++ b/crates/orchestrator-cli/src/services/cloud.rs @@ -1,4 +1,5 @@ use std::sync::Arc; +use std::time::Duration; use anyhow::{Context, Result}; use orchestrator_core::{FileServiceHub, ServiceHub}; @@ -8,8 +9,8 @@ use protocol::DeployConfig; use serde::{Deserialize, Serialize}; use crate::{ - print_value, CloudCommand, CloudLinkArgs, CloudSetupArgs, DeployCommand, DeployCreateArgs, DeployDestroyArgs, - DeployStartArgs, DeployStatusArgs, DeployStopArgs, + print_value, CloudCommand, CloudLinkArgs, CloudLoginArgs, CloudSetupArgs, DeployCommand, DeployCreateArgs, + DeployDestroyArgs, DeployStartArgs, DeployStatusArgs, DeployStopArgs, }; pub(crate) async fn handle_cloud( @@ -19,6 +20,7 @@ pub(crate) async fn handle_cloud( json: bool, ) -> Result<()> { match command { + CloudCommand::Login(args) => handle_login(args, json).await, CloudCommand::Setup(args) => handle_setup(args, project_root, json).await, CloudCommand::Link(args) => handle_link(args, project_root, json).await, CloudCommand::Push => handle_push(hub, project_root, json).await, @@ -28,6 +30,102 @@ pub(crate) async fn handle_cloud( } } +async fn handle_login(args: CloudLoginArgs, json: bool) -> Result<()> { + let server = args.server.unwrap_or_else(|| "https://api.animus.cloud".to_string()); + let server = server.trim_end_matches('/'); + + // Step 1: Initiate device auth flow + let client = reqwest::Client::new(); + let resp = client + .post(&format!("{}/api/cli/auth/initiate", server)) + .send() + .await + .context("Failed to connect to auth server")?; + + if !resp.status().is_success() { + let status = resp.status(); + let body = resp.text().await.unwrap_or_default(); + anyhow::bail!("Auth initiation failed ({status}): {body}"); + } + + let auth_response: AuthInitiateResponse = resp.json().await.context("Failed to parse auth response")?; + + // Step 2: Open browser or print URL + let auth_url = &auth_response.auth_url; + if args.no_browser { + if !json { + eprintln!("Open the following URL in your browser to authenticate:"); + eprintln!("{}", auth_url); + eprintln!("Device code: {}", auth_response.device_code); + } + } else { + // Attempt to open browser + let _ = open_browser(auth_url); + if !json { + eprintln!("Opening browser for authentication..."); + eprintln!("If browser did not open, visit: {}", auth_url); + } + } + + // Step 3: Poll for completion + let max_attempts = 120; // 2 minutes with 1 second polling + let poll_interval = Duration::from_secs(1); + + for attempt in 0..max_attempts { + tokio::time::sleep(poll_interval).await; + + let resp = client + .post(&format!("{}/api/cli/auth/complete", server)) + .json(&AuthCompleteRequest { device_code: auth_response.device_code.clone() }) + .send() + .await; + + match resp { + Ok(r) if r.status().is_success() => { + let complete_response: AuthCompleteResponse = + r.json().await.context("Failed to parse completion response")?; + + // Step 4: Store token in SyncConfig + let mut config = SyncConfig::load_global(); + config.server = Some(server.to_string()); + config.token = Some(complete_response.token.clone()); + config.save_global()?; + + let result = LoginResult { + authenticated: true, + server: server.to_string(), + message: "Successfully authenticated with animus cloud".to_string(), + }; + + if !json { + eprintln!("✓ Authentication successful!"); + eprintln!("Server: {}", server); + } + + return print_value(result, json); + } + Ok(r) if r.status().as_u16() == 400 => { + // Not yet complete, continue polling + continue; + } + Ok(r) => { + let status = r.status(); + let body = r.text().await.unwrap_or_default(); + anyhow::bail!("Auth completion failed ({status}): {body}"); + } + Err(e) if attempt < max_attempts - 1 => { + // Network error, retry + continue; + } + Err(e) => { + anyhow::bail!("Auth completion request failed: {}", e); + } + } + } + + anyhow::bail!("Authentication timeout - user did not complete login within 2 minutes") +} + async fn handle_setup(args: CloudSetupArgs, project_root: &str, json: bool) -> Result<()> { let mut global_config = SyncConfig::load_global(); global_config.server = Some(args.server.clone()); @@ -490,3 +588,56 @@ struct DeployStatusDeployResult { machines: Vec, last_deployed_at: Option, } + +#[derive(Deserialize)] +struct AuthInitiateResponse { + device_code: String, + auth_url: String, +} + +#[derive(Serialize)] +struct AuthCompleteRequest { + device_code: String, +} + +#[derive(Deserialize)] +struct AuthCompleteResponse { + token: String, +} + +#[derive(Serialize)] +struct LoginResult { + authenticated: bool, + server: String, + message: String, +} + +fn open_browser(url: &str) -> Result<()> { + #[cfg(target_os = "macos")] + { + std::process::Command::new("open").arg(url).spawn()?; + } + + #[cfg(target_os = "linux")] + { + // Try xdg-open first, then firefox, then chromium + let _ = std::process::Command::new("xdg-open").arg(url).spawn().or_else(|_| { + std::process::Command::new("firefox") + .arg(url) + .spawn() + .or_else(|_| std::process::Command::new("chromium").arg(url).spawn()) + }); + } + + #[cfg(target_os = "windows")] + { + std::process::Command::new("cmd").args(&["/C", "start", url]).spawn()?; + } + + #[cfg(not(any(target_os = "macos", target_os = "linux", target_os = "windows")))] + { + // On other platforms, just return Ok (user will need to visit URL manually) + } + + Ok(()) +} diff --git a/crates/orchestrator-core/src/services/runner_helpers.rs b/crates/orchestrator-core/src/services/runner_helpers.rs index e05018062..a468d0755 100644 --- a/crates/orchestrator-core/src/services/runner_helpers.rs +++ b/crates/orchestrator-core/src/services/runner_helpers.rs @@ -782,7 +782,8 @@ mod tests { std::env::remove_var("AO_SKIP_RUNNER_START"); std::env::remove_var("AGENT_RUNNER_TOKEN"); - std::env::set_current_dir("/Users/samishukri/ao-cli").ok(); + // Preserve current directory for test environment + let test_cwd = std::env::current_dir().ok(); let expected_build_id = runner_binary_build_id(&_binary); let startup_result = ensure_agent_runner_running(&project_root).await; @@ -820,7 +821,8 @@ mod tests { std::env::remove_var("AO_SKIP_RUNNER_START"); std::env::remove_var("AGENT_RUNNER_TOKEN"); - std::env::set_current_dir("/Users/samishukri/ao-cli").ok(); + // Preserve current directory for test environment + let test_cwd2 = std::env::current_dir().ok(); let second_startup_result = ensure_agent_runner_running(&project_root).await; diff --git a/crates/workflow-runner-v2/src/runtime_contract.rs b/crates/workflow-runner-v2/src/runtime_contract.rs index 02bffef71..f92d65f5c 100644 --- a/crates/workflow-runner-v2/src/runtime_contract.rs +++ b/crates/workflow-runner-v2/src/runtime_contract.rs @@ -643,8 +643,8 @@ mod tests { "mcp": { "agent_id": "ao", "stdio": { - "command": "/Users/samishukri/ao-cli/target/debug/ao", - "args": ["--project-root", "/Users/samishukri/ao-cli", "mcp", "serve"] + "command": "/path/to/ao/target/debug/ao", + "args": ["--project-root", "/path/to/project", "mcp", "serve"] } } }); @@ -677,14 +677,14 @@ mod tests { "mcp": { "agent_id": "ao", "stdio": { - "command": "/Users/samishukri/ao-cli/target/debug/ao", - "args": ["--project-root", "/Users/samishukri/ao-cli", "mcp", "serve"] + "command": "/path/to/ao/target/debug/ao", + "args": ["--project-root", "/path/to/project", "mcp", "serve"] } } }); inject_named_mcp_servers( &mut runtime_contract, - "/Users/samishukri/ao-cli", + "/path/to/project", &ctx, "requirements", &["ao".to_string()], diff --git a/docs/reference/cli/index.md b/docs/reference/cli/index.md index 9ca94cf8c..556bb57af 100644 --- a/docs/reference/cli/index.md +++ b/docs/reference/cli/index.md @@ -242,11 +242,13 @@ ao │ ├── setup Guided onboarding and configuration wizard ├── cloud Sync tasks and requirements with a remote ao-sync server +│ ├── login Authenticate with animus cloud using device auth flow │ ├── setup Configure the sync server connection for this project │ ├── push Push local tasks and requirements to the sync server │ ├── pull Pull tasks and requirements from the sync server into local state │ ├── status Show sync configuration and last sync status -│ └── link Link this project to a specific remote project by ID +│ ├── link Link this project to a specific remote project by ID +│ └── deploy Manage deployments on ao-cloud │ └── doctor Run environment and configuration diagnostics ```