- Port pi-ai from upstream with Bun-first approach (source files, no dist)
- Add cli_ prefix to tool names in OAuth mode to avoid collisions
- Improve beta header handling with proper deduplication
- Convert codex-instructions.md to Bun text embed
- Strip .js extensions from all imports
- Added caching for system environment detection to improve startup performance.
- Changed GPU detection on Linux to prioritize discrete GPUs (NVIDIA, AMD) over integrated graphics and skip server management adapters.
- Implemented SystemInfoCache interface and file-based persistence in ~/.omp/system_info.json.
- Added disk usage information to automatic environment detection.
- Added versioned SSH host info cache format for automatic schema refresh on version mismatch.
- Added SSH compat shell detection to probe for bash/sh on Windows hosts.
- Expanded SSH tool documentation with shell-specific command examples for Unix, PowerShell, and CMD.
- Refactored buildCompatCommand to sync function with explicit shell type parameter.
- Added automatic remote host OS and shell detection with disk-based caching.
- Added compat mode option for SSH hosts to use POSIX shell compatibility wrapping.
- Added Windows shell support with PowerShell and CMD path escaping functions.
- Updated SSH tool description to display detected shell type per host.
- Simplified SSH tool documentation by removing OS-specific command examples.
- Added automatic environment detection to system prompt including OS, distro, kernel, CPU, GPU, shell, terminal, desktop environment, and window manager information.
- Implemented cross-platform detection functions for Windows, macOS, and Linux environments.
- Added SSH tool for executing commands on remote hosts with project config discovery from ssh.json files.
- Added persistent SSH connection manager with master socket support and automatic cleanup.
- Added SSHFS mount utilities for mounting remote filesystems with automatic unmount on session end.
- Added SSH capability definition with host validation and discovery provider registration.
- Added environment variable management via settings.json with applyEnvironmentVariables() and related methods.
- Enhanced file mentions to support directories, validate boundaries, and truncate large files.
- Added fallback model candidate logic in title generator to retry on model failures.
- Fixed bash output preview to minify JSON content and skip structural-only lines.
- Added Cerebras provider support with thinking block filtering and zai-glm-4.6 model priority.
- Refactored tree rendering to use consistent theme.tree properties instead of boxSharp.
Long whitespace tokens (e.g., lines with many spaces) were not being
truncated when they exceeded the line width, causing rendered lines
to exceed terminal width and crash the TUI.
- Added progress indicator callback when starting LSP servers at session startup.
- Added bundled /init slash command available by default without requiring user configuration.
- Added 5-second timeout for LSP server warmup with lazy initialization fallback.
- Added configurable timeout parameters to LSP client initialization and request methods.
- Refactored slash command loading to use parseCommandTemplate helper and merge embedded commands.
- Added upfront diff parsing and filtering in code review command with automatic reviewer agent count recommendation.
- Added fsync call before closing session writer to ensure data durability.
- Replaced Bun.Glob-based gitignore scanning with fd command execution in find tool.
- Fixed hasOverrides logic to only set true when parsed.servers has entries in LSP config.
- Replaced console.log/error/warn calls with structured logger across session manager and LSP modules.
- Added in-memory session storage abstraction with SessionStorage interface and MemorySessionStorage implementation.
- Added configurable bash interceptor rules with pattern-based command interception settings.
- Added AbortSignal support to 80+ web scrapers for request cancellation.
- Changed task tool to require explicit task `id` field and single `agent` parameter at top level.
- Removed `submit_review` tool - reviewers now use `complete` tool with structured output.
- Replaced Date.now() and crypto.randomUUID() with nanoid() for unique ID generation throughout codebase.
- Added 25+ new specialized web scrapers for sites including MusicBrainz, Discourse, Lemmy, Sourcegraph, ORCID, and package registries.
- Refactored web-fetch-handlers module to web-scrapers with modular architecture and centralized handler exports.
- Added AbortSignal support throughout web-fetch tool and scraper utilities for improved request cancellation.
- Fixed npm license parsing to handle object format with type property.
- Fixed raw GitHub URL construction by removing refs/heads prefix from ref path.
- Added graceful termination with cleanup and timeout handling to task executor.
- Fixed task worker error handling to prevent hanging on crashes with global error handlers.
- Fixed double-stringified JSON output parsing from subagents.
- Fixed markitdown tool to use automatic installer via ensureTool() instead of Bun.which().
- Added abort reason tracking and centralized timer/listener cleanup in executor.
- Added 50+ specialized web-fetch handlers for platform-specific content extraction including GitHub, GitLab, npm, PyPI, crates.io, and many more.
- Added automatic yt-dlp installation for YouTube transcript extraction with fallback support.
- Added handlers for social platforms including Twitter/X via Nitter, Reddit, Mastodon, Bluesky, and Hacker News.
- Added handlers for academic sources including arXiv, PubMed, Semantic Scholar, bioRxiv, and IACR ePrint.
- Added Python tool installation support via uv or pip fallback for markitdown and html2text.
- Extracted URL handlers from monolithic web-fetch.ts into modular handler files with shared utilities.
- Added YouTube video support with automatic transcript extraction via yt-dlp.
- Added parseYouTubeUrl function to extract video and playlist IDs from various URL formats.
- Added cleanVttToText function to convert VTT subtitle files to plain text.
- Reformatted indentation from spaces to tabs in jtd-to-json-schema.ts and name-generator.ts.
- Updated DEVELOPMENT.md architecture diagram to reflect SDK wrapper, capability system, and git-tool dependency.
- Expanded directory structure documentation to cover 40+ new files including capability, discovery, and extension systems.
- Revised development workflow instructions to use bun commands instead of npm/tsx.
- Added documentation sections for new extension sources and capability types.
- Updated AGENTS.md with simplified issue labeling, corrected repository URLs, and streamlined release commands.
- Consolidated README.md package table and removed redundant development/publishing sections.
- Removed matrix strategy; all binaries built on ubuntu-latest
- Added cross-platform native binding installation for bun
- Simplified artifact handling by building directly in packages/coding-agent/binaries
- Added tag input for workflow_dispatch
- Fixed memory accumulation by streaming events directly to disk instead of storing in memory.
- Changed subprocess usage tracking to accumulate incrementally from message_end events.
- Fixed session persistence to exclude transient streaming data (partialJson, jsonlEvents).
- Removed parseSubagentUsage() function in favor of pre-accumulated usage data.
- Added sessionId option to StreamOptions for session-based prompt caching in supported providers.
- Changed reasoning configuration to only apply when explicitly specified.
- Replaced static CODEX_PI_BRIDGE export with dynamic buildCodexPiBridge function that formats available tools.
- Added buildCodexSystemPrompt function to construct system prompts from codex instructions and user prompts.
- Co-located tool renderers with their respective tool implementations across 14 tool modules.
- Added render-utils module with shared formatting helpers for diagnostics, diffs, paths, and metadata.
- Removed ~800 lines of inline renderer implementations from centralized renderers.ts file.
- Added Google Vertex AI provider with 11 Gemini model configurations.
- Added OpenAI Codex provider with 13 GPT-5 model configurations.
- Added webSearchProvider, imageProvider, and git.enabled settings for configurable provider selection and feature toggles.
- Added offset and limit parameters to Output tool for paginated reading of large outputs.
- Added provider fallback chain for web search with WebSearchProviderError class for graceful degradation.
- Changed generate_image tool to save images to temp files instead of inline base64 to reduce context bloat.
- Changed web search default model from claude-sonnet-4-5-20250514 to claude-haiku-4-5.
- Added git command interception in bash to redirect to dedicated git tool when enabled.
- Added `all` parameter to edit tool for replacing all occurrences with fuzzy matching support.
- Added OpenRouter provider support for image generation when `OPENROUTER_API_KEY` is set.
- Renamed image tool from `gemini_image` to `generate_image` with updated default model.
- Added test cases for edit tool's `all: true` option covering multiple scenarios.
- Added ImageMagick fallback for image processing when sharp is unavailable.
- Added image-magick utility module with detection, conversion, and resize functions.
- Changed web_fetch tool label from 'web_fetch' to 'Web Fetch'.
- Updated image-convert and image-resize to use ImageMagick fallback instead of failing or returning original.
- Added --ref flag to install specific tags, branches, or commits from source or binary releases.
- Enabled source installation from git refs by cloning and checking out the specified reference.
- Added git availability check for source installations with --ref flag.
- Updated README with usage examples for --ref flag on both shell and PowerShell installers.
Branch summary prompts when navigating away from a branch are now
controlled by a setting (disabled by default). Enable in Settings →
Config → "Branch summaries" to restore the previous prompting behavior.
- Fixed context file preview in inspector panel to read content from raw object instead of async file reads.
- Added slash commands to extensions inspector panel for visibility and management.
- Extracted getContextFileContent helper method for type-safe content extraction.
- Added file-based slash commands discovery from commands/ directories.
- Added $ARGUMENTS placeholder support for slash command argument substitution.
- Changed argument substitution order to process positional args before wildcards to prevent re-substitution bugs.
- Integrated slash command expansion into AgentSession initialization flow.
fd's --glob mode matches against filenames only, not paths. Patterns
like `reports/**` were failing because fd looked for entries literally
named `reports/**`. Now detects path separators and adds --full-path
flag with `**/` prefix to match correctly.
- Changed multi-task display to show task descriptions instead of agent names when available.
- Fixed indentation in worktree constants module for consistency.