- Added task.simple to settings and schema with default, schema-free, and independent modes.
- Added mode-aware task schema and validation to enforce context/schema rules per simple mode.
- Updated prompts and template rendering to tailor headers and guidance for each simple mode.
- Added simple-mode capabilities and updated TaskTool execution for mode-aware context behavior.
- Added tests for independent rendering and mode-specific rejection of invalid context or schema inputs.
- Added `toolStrictMode` support with `all_strict`/`none`/`mixed` options to OpenAI compatibility.
- Fixed OpenAI-completion strict-mode flows by capturing failed HTTP responses and retrying once as non-strict.
- Fixed completion error reporting by surfacing captured status, headers, and JSON `type`/`param`/`code` details.
- Improved strict-schema enforcement with WeakMap memoization and circular-schema detection in sanitization.
- Fixed OpenRouter provider lookup by resolving fallback model IDs for suffix and date variants in registry resolution.
- Refactored benchmark tooling and added async RPC error-window tracking for scheduled run execution.
- Replaced deprecated `await` tool wiring with `poll` in tool exports and built-in tool registry.
- Updated bash and task prompts plus start-result messaging to direct users to the `poll` tool.
- Added effective timeout metadata to Bash tool results and rendered output to show that effective timeout.
- Implemented bounded auto-background wait logic that backgrounds jobs when the timeout window is exhausted.
- Added gh-renderer.ts module with GitHub Actions workflow visualization for terminal output.
- Enhanced gh_pr_view tool to fetch and display inline review comments alongside pull request reviews.
- Improved gh_run_watch tool output with structured job state tracking and failed log formatting.
- Fixed gh_run_watch to resolve explicit branch watches against GitHub API branch head instead of local HEAD.
- Fixed gh_* tool outputs to spill full large results to artifacts instead of pre-truncating.
- Standardized trailing newlines across 100+ prompt template files for POSIX compliance.
- Added background model discovery with provider status tracking and 24-hour model caching to improve startup performance.
- Changed model discovery timeout from 3000ms to 250ms and deferred blocking refresh to background operations for faster initialization.
- Fixed model discovery to preserve cached models when providers are unavailable or unauthenticated.
- Added validation in Kitty key formatting to reject unsupported modifiers and improved error handling.
- Reorganized CI workflow to run TypeScript linting in check job and conditional Rust checks in native job matrix.
- Updated system prompt and tool guidance to recommend combined investigation-and-edit workflows and maintainable code practices.
- Removed per-task skill pinning feature; subagents now inherit session skill set instead of per-task selection.
- Removed `preloadedSkills` option from `CreateAgentSessionOptions` interface and related system prompt plumbing.
- Removed `skills` field from Task schema; task execution now passes available session skills directly to subagents.
- Simplified task execution pipeline by removing skill resolution logic and preloaded skills handling from system prompt templates.
- Restructured task tool documentation for clarity, moving parameter definitions into a dedicated section and consolidating guidance on context, assignments, and parallelization.
- Reformatted system prompt template to use markdown headings instead of XML tags for skills, preloaded skills, and rules sections.
- Renamed the `notes://` protocol to `local://` for better clarity.
- Updated all internal references, prompts, and tool documentation.
- Migrated plan storage paths to use the new `local://` scheme.
- Replaced plan:// protocol with notes:// for session-scoped artifact storage and plan finalization.
- Added title parameter to exit_plan_mode tool to enable plan file renaming during approval workflow.
- Implemented NotesProtocolHandler for notes:// URL scheme with path traversal protection and session fallback.
- Added renameApprovedPlanFile function to handle plan artifact finalization with validation and error handling.
- Updated system prompt documentation to reference notes:// protocol and internal URL schemes for artifact access.
- Standardized XML tag naming from snake_case to kebab-case across 50+ prompt files for consistency.
- Replaced imperative language with RFC 2119 keywords (MUST/SHOULD/MAY/MUST NOT) throughout system and tool prompts for clarity.
- Removed artifactsDir parameter from Python executor and simplified environment variable handling to use PI_SESSION_FILE only.
- Renamed read_path.md to read-path.md and updated memory guidance with hierarchy rules and conflict resolution workflow.
- Added noEscape option to bash URL expansion and extracted cwd parameter from leading cd commands for improved path handling.
- Exported NO_PAGER_ENV constant from bash-interactive module for centralized environment variable management.
- Added poll_jobs tool for blocking until background jobs complete without manual polling loops.
- Added task.maxConcurrency setting to limit concurrent subagent task execution with semaphore-based control.
- Enhanced task progress tracking to report per-task status with individual timing and token metrics.
- Improved parallel task execution to schedule multiple background jobs independently for true concurrent execution.
- Updated bash and task tool documentation to recommend poll_jobs instead of polling read jobs:// in loops.
- Added async background job execution for bash and task tools with configurable concurrency limits and automatic result delivery.
- Added cancel_job tool and /jobs slash command to manage and inspect running background jobs with status display.
- Added jobs:// internal protocol handler for querying job status and retrieving job execution details.
- Added async.enabled and async.maxJobs settings to control background job execution behavior.
- Enhanced status line to display count of running background jobs with visual indicator.
- Implemented AsyncJobManager with exponential backoff retry delivery, job lifecycle tracking, and automatic eviction.
Fixes#56.
- Clarified that subagents automatically inherit full system prompt including AGENTS.md, context files, and skills -- do not repeat project rules or conventions in task context.
- Updated task context guidance to focus on session-specific information subagents lack, eliminating redundant documentation of project constraints already available to them.
- Refined constraints template to emphasize task-specific rules and session decisions rather than global project conventions.
- Expanded anti-patterns section to explicitly flag redundant context that wastes tokens by repeating AGENTS.md rules, project constraints, and tool preferences.
- Fixed `newSession()` to properly await session manager operations for full initialization.
- Changed session formatting to use XML structure instead of YAML for tools and tool invocations.
- Added formatArgsAsXml() helper function to serialize objects as XML parameter elements.
- Removed trailing whitespace from task.md prompt file.
- Added regression test for AgentSession.newSession() to verify proper session switching and artifact isolation.
- Fixed task executor to properly handle agents calling `submit_result` with null data by treating it as missing and attempting to extract output from conversation text rather than silently failing.
- Added documentation caution in task.md about schema vs agent mismatch causing null output, with guidance on using `schema` parameter to override built-in schemas.
- Added warning message when subagent calls submit_result with null data to help diagnose schema mismatch issues.
- Added task isolation setting to enable running subagents in isolated git worktrees.
- Made task schema dynamically include isolated parameter only when isolation is enabled.
- Added validation to reject isolated parameter when task isolation is disabled.
- Updated task tool to conditionally render isolation documentation based on feature flag.
- Refactored task schema into factory function to support dynamic isolation configuration.
- Refactored task API to use 'assignment' field instead of 'args' for per-task instructions, enabling clearer separation between shared context and task-specific work.
- Introduced structured context/assignment separation pattern with '<swarm_context>' wrapper for template rendering, replacing placeholder-based substitution.
- Changed agent frontmatter field from 'thinkingLevel' to 'thinking-level' (kebab-case) for consistency with YAML conventions.
- Removed 'context' parameter from ExecutorOptions as context is now prepended at template level rather than executor level.
- Removed 'args' field from AgentProgress and SingleResult interfaces, simplifying task result tracking.
- Updated task rendering to display full task text instead of formatted args, improving clarity in progress output.
- Refactored ExecutablePathSearch to remove generic type parameter N and use pre-computed filename candidates instead of single filename.
- Extracted platform-specific candidate_filenames() functions to generate executable filename variants on Windows and Unix-like systems.
- Replaced manual PATH string splitting with std::env::split_paths() for cross-platform path parsing.
- Simplified find_in_path() method to delegate to pathsearch::search_for_executable() instead of custom path searching logic.
- Implemented executable_extensions() function with lazy initialization to parse PATHEXT environment variable on Windows.
- Enhanced executable() method to validate file extensions against known executable extensions instead of returning hardcoded true.
- Updated task tool guidance to enforce small, well-defined task scope with maximum 3-5 files per task to prevent timeouts and improve parallel execution.
- Added critical section with task granularity requirements, red flags for overly broad tasks, and examples of bad vs. good task definitions.
- Condensed prompt documentation throughout by removing articles, auxiliary verbs, and redundant explanations for improved clarity and conciseness.
- Simplified system prompts by replacing multi-line explanations with terse, semicolon-separated fragments and removing verbose motivational language.
- Reformatted tool documentation to use abbreviated language, removing 'the', 'a', and 'is' throughout for more direct technical writing.
- Removed blank lines and consolidated spacing in prompt files to reduce vertical whitespace and improve document density.
- Removed redundant instruction about preferring delegation in coordinator prompt.
- Improved task context template formatting with consistent markdown code fence markers.
- Removed dependencies section from task template constraints as it was redundant.
- Enhanced readability of task examples with better spacing and formatting.
- Added native find module for high-performance filesystem discovery with glob pattern matching and .gitignore support.
- Made grep, html, image, and text native functions asynchronous to prevent blocking the event loop.
- Updated native function names to use camelCase convention for JavaScript interoperability.
- Extended grep and text functions to accept Uint8Array/Buffer input for zero-copy operations.
- Delegated file discovery logic from JavaScript to native Rust implementation for improved performance.
- Migrated native bindings from WebAssembly (wasm-bindgen) to Node.js N-API architecture for improved performance and platform support.
- Replaced worker pool-based async architecture with synchronous native addon calls, eliminating thread overhead and simplifying the API.
- Added new grep module with ripgrep-backed filesystem search supporting parallel processing, glob filtering, and context line extraction.
- Updated build system to compile Rust crates to platform-specific .node binaries instead of WASM modules.
- Removed WASI implementation and worker pool infrastructure no longer needed with N-API native bindings.
- Updated CI/CD pipeline to build native addons for multiple platforms (Linux x64/arm64, macOS x64/arm64, Windows x64) and include them in releases.
- Added phased execution guidance for migrations and refactors to prevent parallel work on dependent layers.
- Added example demonstrating phased execution pattern for porting WASM to N-API with sequential foundation, parallel implementation, integration, and dependent layer phases.
- Updated task tool documentation to clarify that subagents can access parent conversation context via a searchable file, reducing need to repeat information in context parameter.
- Revised critical instruction section to guide users toward leveraging searchable parent context file instead of requiring self-contained context.
- Clarified that user instructions about delegation override tool-use defaults in system prompt.
- Updated coordinator guidance to emphasize Task tool preference for substantial work with improved emphasis on context window limitations.
- Enhanced `context` parameter documentation to require self-contained information for subagents, including file contents and user requirements.
- Renamed 'output' parameter to 'schema' throughout task configuration and documentation for clarity.
- Updated task schema definition to use 'schema' field with improved description clarifying it defines expected response structure.
- Restructured task summary output format to consolidate metadata into tag attributes and simplify XML structure.
- Added truncation detection for task output exceeding 5000 character threshold with truncated flag.
- Removed charCount from task metadata, retaining only lineCount and charSize properties.
- Extracted tool choice mapping logic into dedicated utility module for provider-specific format conversions.
- Refactored task summary generation to use template-based rendering instead of string concatenation.
- Migrated system prompt generation to use Handlebars templates with structured data instead of hardcoded strings.
- Extracted JTD to TypeScript conversion logic into dedicated utility module for schema documentation.
- Consolidated prompt template rendering across multiple modules using unified renderPromptTemplate function.
- Simplified type assertions in stream module to use generic OptionsForApi type instead of provider-specific types.
- Renamed 'complete' tool to 'submit_result' throughout codebase for clarity and consistency.
- Updated all tool references, function names, and variable names from 'complete' to 'submit_result' in executor, render, and tools modules.
- Renamed CompleteTool class to SubmitResultTool and CompleteDetails interface to SubmitResultDetails.
- Updated agent prompts and documentation to reference the new 'submit_result' tool name.
- Reorganized task.md prompt to move critical guidance to the top and restructure instructions for better clarity.
- Added session compaction auto-continue feature that automatically sends a synthetic continuation prompt after compaction completes when enabled in settings.
- Added tool output pruning system that intelligently removes verbose tool outputs from session history during compaction to reduce token usage while preserving critical information.
- Added short summary generation for compaction results, providing concise PR-style summaries of conversation changes alongside full summaries.
- Added session.compacting extension hook that allows extensions to customize compaction prompts, add context, and preserve metadata during summarization.
- Added synthetic message support to mark system-injected messages (like auto-continue prompts) and prevent them from being persisted to editor history.
- Inline preloaded skill contents in system prompts and allow task-level skill selection.
- Record session init entries with system prompt, task, tools, and output schema.
- Fix session persistence to keep pre-assistant entries in subagent logs.
- Plan mode provides structured workflow where agents propose plans for user approval before execution.
- Added plan:// internal URL protocol for accessing plan files and injecting plan-mode context into subagent prompts.
- Added plan mode toggle shortcut and paused status indicator in status line.
- Fixed plan reference injection to properly pass workflow state to plan-mode system prompts.
- Improved autocomplete fuzzy matching to support subsequence matching for skill suggestions.
- Added ExistingChangelogEntries interface to pass existing unreleased changelog entries for deletion support.
- Agent prompts now display existing unreleased changelog entries organized by path for potential removal.
- Made analyze_files tool conditional on enableAnalyzeFiles flag when pre-computed observations are available.
- Fixed changelog application to process proposals with only deletions without new entries.
- Simplified Python gateway coordination by removing reference counting and idle timeout functionality.
- Updated Python kernel initialization to support environment variable setup and working directory configuration.
- Modified interactive status display to show Python and venv paths instead of client counts.
- Updated system prompt to clarify CHECKPOINT step 0 timing after user request completion.
- Fixed Python environment warming to use await for proper async error handling.
- Changed interactive mode shutdown to use postmortem.quit instead of process.exit.
- Enhanced bash and python executors to save full output as artifacts when truncated.
- Replaced manual error formatting with ToolError.render() for standardized handling.
- Fixed timeout parameter handling to auto-convert milliseconds to seconds.
- Renamed task parameter from vars to args throughout task tool interface.
- Enhanced task template rendering with built-in {{id}} and {{description}} placeholders.
- Simplified task template validation by removing strict placeholder requirements.
- Streamlined CHECKPOINT step requirements in system prompts.
- Removed internal URL handling from fetch tool to use read tool instead.
- Updated fetch and task tool documentation to reference read tool for agent:// URLs.
- Added timeout clamping to reasonable ranges across all tools.
- Fixed timeout parameter handling to auto-convert milliseconds to seconds.
- Simplified todo-write tool by removing active_form parameter requirement.
- Added internal URL routing system with agent:// and skill:// protocol handlers.
- Enhanced fetch tool to support internal URLs and JSON extraction from agent outputs.
- Extended read tool to handle internal URLs with pagination and query extraction.
- Removed Output tool and replaced functionality with agent:// URL integration.
- Replaced minimatch dependency with Bun.Glob for pattern matching across modules.
- Standardized documentation format across all 18 tools with consistent XML sections.
- Added `<output>` sections to clarify return values and behavior for all tools.
- Reorganized tool instructions with clearer `<instruction>`, `<critical>`, and `<avoid>` sections.
- Renamed Patch tool to Edit and Edit tool to Replace for better naming consistency.
- Added nonAbortable option to AgentTool interface for ignoring abort signals during tool execution.
- Added isolated option to task tool for running tasks in isolated git worktrees.
- Simplified tool parameter descriptions across all coding-agent tools for better readability.
- Updated find tool to always sort results by modification time and removed sortByMtime parameter.
- Changed web-fetch timeout default from 20s to 45s maximum and updated bash prompt to use cwd parameter.
- Removed model parameter from task and gemini-image tools to use session/provider defaults.
