- show help instead of crashing on `omp setup` with no args
- show runtime-discovered MCP servers in `/mcp list`
- remove deprecated Anthropic model entries from models.json
- sort models by recency in model selector
The Z.AI login flow was hardcoded to validate API keys using the 'glm-5'
model. However, the Z.AI Coding Plan, which this login is for, currently
provides access to 'glm-4.7', with 'glm-5' being rolled out incrementally.
This caused login to fail for users who did not yet have 'glm-5' access.
This change updates the validation model to 'glm-4.7' to match the model
available in the Coding Plan.
Fixes#39
Detects known Bun.stringWidth mishandling of OSC sequences at startup
and exits with a clear upgrade message instead of crashing in the TUI
render path.
Fixes#31
- Fixed incorrect path resolution for CLI module by using Bun.fileURLToPath() to convert import.meta.resolve() result to a proper file path.
- Reordered constant declarations to ensure TMP_DIR is initialized before CLI_PATH which depends on it.
- Updated scripts/repro-stuck.ts to use consistent CLI path resolution with Bun.fileURLToPath().
- Replaced jsdom with linkedom for HTML parsing in browser tool, improving performance and reducing memory footprint.
- Removed @types/jsdom from devDependencies as jsdom is no longer used.
- Simplified HTML document parsing by removing VirtualConsole error handling and direct window cleanup.
- Removed unused yaml dependency from swarm-extension package.
- Updated pi-coding-agent peer dependency constraint from wildcard to ^11.
- Removed pi-coding-agent from devDependencies, keeping only in peerDependencies.
- Exported renderPromptTemplate function from config/prompt-templates module for programmatic prompt template rendering.
- Exported computeLineHash function from patch/hashline module for external patch utility access.
- Added ./cli export path in package.json for direct CLI module access.
Adds a standalone extension that orchestrates multi-agent workflows
defined in YAML. Supports pipeline (iterative), parallel (fan-in/out),
sequential, and arbitrary DAG execution patterns.
Each agent runs as a full oh-my-pi subagent with complete tool access.
Agents communicate through the shared workspace filesystem. The
orchestrator handles lifecycle, dependency ordering, and state tracking.
Key components:
- YAML schema parser with validation and cycle detection
- DAG builder with topological sort into execution waves
- Pipeline controller with iteration loop and wave execution
- Filesystem state persistence for monitoring and resumability
- Standalone runner (run-pipeline.ts) for long-running unattended work
- TUI integration via /swarm command
Designed for any task type: research, coding, data processing, content
creation, analysis workflows, or any multi-step objective that benefits
from specialized agents working in coordination.
- Updated Claude Code version header to 2.1.39 and runtime version to v24.13.1.
- Increased request timeout from 60s to 600s for improved reliability on slower connections.
- Reordered Accept-Encoding header values to prioritize Brotli compression.
- Updated OAuth endpoints to use platform.claude.com and expanded scopes to include user:sessions:claude_code and user:mcp_servers.
- Removed deprecated beta features claude-code-20250219 and fine-grained-tool-streaming-2025-05-14.
- Improved Bun binary detection and update process with version verification and environment variable checking.
- Added SwiftLint linter client with JSON reporter support for Swift file linting.
- Changed SwiftLint configuration to use 'lint' command with JSON reporter for structured output.
- Changed bash.virtualTerminal default setting from 'on' to 'off' for standard non-interactive execution.
- Implemented SwiftLintClient with lint() method that executes swiftlint with JSON output and converts violations to LSP diagnostics.
- Added bash.virtualTerminal settings option with 'on' (PTY-backed interactive) and 'off' (standard non-interactive) values.
- Extracted tool arguments cloning logic into dedicated `cloneToolArgs` helper function to improve code reusability.
- Moved preview diff injection logic from `computeHashlineDiff` callback into new `#getCallArgsForRender` method to centralize argument transformation for rendering.
- Applied `cloneToolArgs` to constructor and `updateArgs` method to ensure tool arguments are properly cloned on assignment.
- Added PTY (pseudo-terminal) backed interactive command execution with streaming output support via PtySession class in pi-natives.
- Added PtyStartOptions and PtyRunResult types to configure and report PTY session execution status.
- Added write(), resize(), and kill() methods to PtySession for interactive control of running commands.
- Added interactive bash execution via PTY with real-time terminal rendering and input forwarding in bash-interactive tool.
- Added --no-pty CLI flag and PI_NO_PTY environment variable to disable PTY-based interactive bash execution.
- Added bash.virtualTerminal setting to control PTY-backed interactive execution behavior.
- Sorted Antigravity models alphabetically in generated models.json and generation script.
- Updated Antigravity model configurations with corrected model IDs, names, and parameters.
- Fixed qwen/qwen3-max-thinking maxTokens from 4096 to 202752 to match context window.
- Added /move slash command to allow users to relocate sessions to different working directories.
- Added moveTo() method to SessionManager for programmatic session relocation with file system operations and rollback support.
- Added refreshSlashCommandState() method to InteractiveMode for dynamic slash command reloading when changing directories.
- Added setSlashCommands() method to AgentSession to allow external management of file-based slash commands.
- Added handleMoveCommand() handler in command-controller to process /move commands with validation and error handling.
- Added zombie process prevention by tracking reap state and implementing Drop trait to ensure processes are properly cleaned up.
- Enabled kill_on_drop for tokio child processes to automatically terminate processes when handles are dropped.
- Improved process termination logic to check reap state before attempting kill operations and properly track completion.
- Added Perplexity OAuth authentication support with native macOS app extraction and email OTP login flows.
- Implemented loginPerplexity and refreshPerplexityToken functions for OAuth token management.
- Created PerplexitySocket class providing Socket.IO v4 client over Engine.IO v4 WebSocket transport with RPC support.
- Added OAuth authentication to Perplexity web search provider with automatic token refresh and SSE streaming support.
- Extended SearchResponse interface with authMode field to track authentication method (oauth or api_key).
- Changed PerplexityProvider.isAvailable() to async to support OAuth token availability checking.
- Removed support for `.pi` configuration directory alias in favor of `.omp` across all packages.
- Updated all configuration paths and references from `.pi` to `.omp` in user and project directories.
- Refactored configuration discovery in builtin.ts to simplify directory traversal logic and remove multi-alias support.
- Simplified Python module discovery to use single directory paths instead of arrays of aliases.
- Updated debug and crash log paths from `.pi/agent/` to `.omp/agent/` in TUI package.
- Extracted cross-platform URL and file path opening logic into a unified `openPath` utility function.
- Removed duplicate platform-specific browser opening code from five modules (stats-cli, debug, login-dialog, command-controller, mcp-command-controller) and replaced with calls to the centralized utility.
- Simplified error handling by delegating platform detection and command execution to the reusable utility function.
- Added SAFETY comments documenting unsafe block invariants in process child collection.
- Refactored nested if statement to use let-else guard clause pattern in OS distro detection.
The GitHub Copilot providers were unconditionally overriding the X-Initiator
header based on message role. This prevented the subagent fix from working
because subagents send the task as a user message.
Now the providers check if X-Initiator is already set in the model headers
before applying the default logic. This allows the sdk.ts fix to properly
mark subagent requests as agent-initiated.
Add support for MiniMax Coding Plan with OpenAI-compatible API:
- New providers: minimax-code (international) and minimax-code-cn (China)
- Environment variables: MINIMAX_CODE_API_KEY and MINIMAX_CODE_CN_API_KEY
- Uses thinkingFormat: 'zai' for reasoning compatibility
- Models: MiniMax-M2, MiniMax-M2.1, MiniMax-M2.1-lightning
- Base URLs: https://api.minimax.io/v1 (intl), https://api.minimaxi.com/v1 (CN)
The Coding Plan is a subscription-based service separate from regular MiniMax API.
Add support for z.ai's GLM Coding Plan as a provider accessible via `/login`.
Implements API key-based authentication similar to Kimi Code.
Changes:
- Create login flow in `utils/oauth/zai.ts` with browser redirect to API keys page
- Add API key storage methods (`saveApiKey`, `getApiKey`) to CliAuthStorage
- Register Z.AI in OAuth provider types, exports, and CLI handler
- Add GLM-5 model to Z.AI provider
Users can now authenticate with:
bunx @oh-my-pi/pi-ai login zai
- Changed hashline format separator from two spaces to pipe character (|) throughout codebase for consistency.
- Simplified ollama installation checks in test files by replacing execSync with Bun.which() and inverting PI_NO_LOCAL_LLM to PI_LOCAL_LLM environment variable.
- Added early return guards in test beforeAll hooks to skip tests when required tools (ollama, magick) are unavailable.
- Added debug instrumentation to system prompt building via PI_DEBUG_STARTUP environment variable for startup performance monitoring.
- Removed prompt cache retention assertion from openai-codex streaming test.
- Added macOS marketing name detection to system information collection for both Rust and TypeScript implementations.
- Changed system info functions to return optional values instead of fallback strings, improving data accuracy.
- Refactored GPU caching to be independent from other system info, reducing unnecessary cache invalidation.
- Updated environment info collection to filter out undefined values, providing cleaner system prompt output.
- Changed hashline display format separator from pipe to two spaces for improved readability.
- Removed `lines` and `hashes` parameters from read tool in favor of automatic file display mode resolution.
- Added `resolveFileDisplayMode` utility to centralize file display mode configuration logic.
- Integrated file display mode settings into grep and read tools for consistent output formatting.
- Consolidated tool parameter types to use schema-derived types via Typebox `Static` utility.
- Updated parseLineRef to handle both legacy pipe-separator and new two-space hashline formats.
- Removed outdated benchmark reports from previous test runs across multiple models and edit variants.
- Added new benchmark reports for Claude Haiku 4.5, Claude Sonnet 4.5, Deepseek V3.2, Devstral Medium, Gemini 3 Flash, GLM-4.5-Air, GPT-5.1-Codex-Mini, GPT-5.2-Codex, Grok-4-1-Fast, Grok-4-Fast-Non-Reasoning, Grok-Code-Fast-1, Kimi K2.5, Minimax M2.1, Qwen Turbo, and Zai-GLM-4.7 models across hashline, patch, and replace edit variants.
- Updated benchmark test results with improved task success rates and edit success metrics across all tested models.
can1357