- Changed `HashlineEdit.src` from string format (e.g., `"5:ab"`, `"5:ab..9:ef"`) to structured `SrcSpec` object with discriminated union types (`{ kind: "single", ref: "..." }`, `{ kind: "range", start: "...", end: "..." }`, `{ kind: "insertAfter", after: "..." }`, `{ kind: "insertBefore", before: "..." }`, `{ kind: "substring", needle: "..." }`).
- Refactored `parseSrc` function to `parseSrcSpec` to handle structured `SrcSpec` objects instead of string parsing, replacing string validation and delimiter parsing logic with explicit case handling for each operation kind.
- Updated tool schema in `index.ts` to define `srcSpecSchema` as a discriminated union type with five variants, replacing the previous simple string-based `src` parameter definition.
- Added substring-based source matching for hashline edits to support matching lines by content when hash references are unavailable.
- Added automatic detection and repair of single-line merges where models incorrectly combine multiple lines into one.
- Added normalization of Unicode-confusable hyphens to ASCII hyphens to handle model-generated variations.
- Added heuristics to restore indentation and preserve wrapped line formatting during edits.
- Relaxed comma validation in src to allow commas while rejecting inputs with multiple line references.
- Enhanced parseLineRef to accept shorter hash prefixes instead of requiring exact hash matches.
- Improved error messages for hash mismatches to provide more actionable guidance.
- Replaced HashlineEdit API from old/new/after fields to src/dst fields with unified range syntax supporting single lines, ranges, insert-after, and insert-before operations.
- Added parseSrc() function to parse structured line references from src strings supporting formats like '5:ab', '5:ab..9:ef', '5:ab..', and '..5:ab'.
- Added heuristics to strip anchor line echoes and range boundary echoes from model-generated replacement content.
- Implemented preserveWhitespaceOnlyLinesLoose() for improved whitespace preservation using loose matching strategy when replacement line counts don't match.
- Added comprehensive benchmark reports for Claude Sonnet 4.5, Gemini 2.5 Flash Lite, and GPT-5.1 Codex Mini models evaluating hashline edit performance across 60 tasks.
- Added abort signal support to MCP server connection and tool listing operations, enabling cancellation via Escape key during testing.
- Enhanced MCP connection timeout handling with improved abort signal integration in the withTimeout function to prevent race conditions.
- Improved MCP test command UI to display '(esc to cancel)' indicator and handle cancellation gracefully.
- Updated HTTP transport session termination to include timeout mechanism preventing indefinite hangs.
- Fixed MCP test command cleanup to prevent resource leaks when operations are cancelled or aborted.
- Added abort event listener registration in bash executor to properly handle abort signals and clean up resources in finally block.
- Improved bash tool error handling to distinguish between user-initiated aborts via AbortSignal and other cancellations, throwing ToolAbortError for aborted requests.
- Wrapped bash executor command execution in try-finally block to ensure abort event listeners are properly removed after execution completes.
- Added MCPRequestOptions interface with signal property for request cancellation via AbortSignal.
- Added abort signal support to MCP tool execution enabling request cancellation via Escape-to-interrupt or other abort mechanisms.
- Enhanced MCP request handling with abort signal propagation through HTTP, SSE, and stdio transports with proper cleanup.
- Improved stdio transport request handling to use Promise.withResolvers for cleaner async flow and better abort signal integration.
- Updated HTTP transport to combine operation abort signals with timeout signals using AbortSignal.any() for unified cancellation.
- Modified SSE response parsing to support abort signals and distinguish between timeout and user-initiated cancellation.
- Added automatic stripping of `LINE:HASH|` display prefixes and unified-diff `+` markers from replacement content in hashline edits.
- Enhanced replace edits to preserve original whitespace on lines where only whitespace differs, preventing spurious formatting diffs when models reformat code.
- Added helper functions `stripNewLinePrefixes()`, `preserveWhitespaceOnlyLines()`, and `equalsIgnoringWhitespace()` to improve edit robustness.
- Enhanced system prompt to display tool descriptions alongside tool names for improved clarity on available capabilities.
- Added toolDescriptions data structure to system prompt rendering context, mapping tool names to their descriptions from the tools registry.
- Updated system prompt template to render tools in structured XML format with name and description elements instead of comma-separated list.
- Added `toArray()` helper function to normalize `string | string[]` inputs for hashline edit operations with comma-separated line reference support.
- Added `HashlineMismatchError` class that displays grep-style output with `>>>` markers for hash validation failures.
- Added `HashMismatch` type for representing individual hash mismatches with line number, expected hash, and actual hash.
- Updated `HashlineEdit` type to accept `string | string[]` for `old` and `new` fields, allowing both singular and plural input formats.
- Reduced hash length from 4 to 2 hex characters (16-bit hashes) for more concise line references.
- Enhanced `parseLineRef()` to strip display-format suffixes and `validateLineRef()` to throw `HashlineMismatchError` with context lines for better error reporting.
- Optimized line hash computation by precomputing hex lookup table and using bitwise operations instead of string formatting.
- Simplified conditional logic in read tool for hash and line number configuration.
- Refactored react-edit-benchmark model/provider parsing to use concise ternary operator.
- Added HashlineMismatchError class with formatted error messages showing mismatched hashes with context lines.
- Added HashMismatch type to represent hash validation errors with line number, expected, and actual hash values.
- Enhanced hash validation in applyHashlineEdits to collect all hash mismatches before applying edits instead of failing on first mismatch.
- Renamed HashlineEdit fields from 'src'/'dst' to 'old'/'new' for improved naming clarity and consistency.
- Added hashline edit mode for line-addressed edits using hash-verified line references with xxHash64 integrity verification.
- Replaced `edit.patchMode` boolean setting with `edit.mode` enum supporting 'replace', 'patch', and 'hashline' modes.
- Added `readHashLines` configuration setting to include line hashes in read output for hashline edit mode.
- Implemented `computeLineHash`, `formatHashLines`, `parseLineRef`, `validateLineRef`, and `applyHashlineEdits` utility functions for hashline operations.
- Updated read tool to support optional `hashes` parameter and prioritize hash lines over line numbers when both are requested.
- Changed `getEditModelVariants()` return type from `Record<string, 'patch' | 'replace'>` to `Record<string, EditMode | null>` and removed hardcoded model-specific defaults.
- Added inline hint support for slash commands with getInlineHint() method on SlashCommand and AutocompleteProvider interfaces.
- Added SubcommandDef interface for declarative subcommand definitions with name, description, and usage properties.
- Added subcommands and inlineHint properties to BuiltinSlashCommand interface for enhanced command configuration.
- Added getArgumentCompletions and getInlineHint functions to materialized slash commands for browser and mcp subcommand completion.
- Converted browser command to support subcommands (headless, visible) with inline hints.
- Converted mcp command to support 10 subcommands (add, list, remove, test, reauth, unauth, enable, disable, reload, help) with descriptions and usage hints.
- Added missing type annotation for `settings` parameter in `resolvePrimaryModel` function.
- Added missing type annotation for `settings` parameter in `resolveSmolModel` function.
- Added missing type annotation for `settings` parameter in `ModelSelectorComponent` constructor.
- Added `temperature` configuration setting to control LLM sampling temperature with values from 0 (deterministic) to 1 (creative) to -1 (provider default).
- Added temperature option selector in settings UI with preset values: Default, 0, 0.2, 0.5, 0.7, 1.
- Renamed `settingsInstance` parameter to `settings` in `CreateAgentSessionOptions` for consistency.
- Updated all internal references from `settingsInstance` to `settings` throughout SDK and components.
- Integrated temperature setting into agent configuration and selector controller.
- Removed unsafe OAuth endpoint extraction from error message text
- Fixed PKCE verifier storage with typed #codeVerifier field
- Fixed refresh token fallback using access token as refresh token
- Enforced restrictive file permissions (0o700/0o600) for MCP configs
- Fixed wizard buildConfig() to respect user-chosen env var and header names
- Fixed reauth endpoint discovery for non-OAuth servers
- Stored original config on connection, resolved config only for transport
- Added runtime type validation for enabled/timeout in config loaders
- Converted all TS private keywords to ES # private fields
- Wrapped uncaught throws in /mcp add with try/catch error handling
- Replaced new Promise with Promise.withResolvers() pattern
- Sanitized TUI output with replaceTabs/truncateToWidth
- Enforced http/https URL validation in add wizard
- Fixed greedy /mcp prefix match in input controller
- Corrected config filename references in MCP guide
- Added server name validation to updateMCPServer
- Fixed timeout timer leak in stdio transport
* + /mcp
- Reloads MCP manager in runtime state (no restart needed) and syncs with mcp.json.
- Handles OAuth discovery/auth flow automatically for auth-required servers.
- Validates server names and config shape before saving.
- Persists OAuth credentials in auth storage and links them to MCP config.
- Provides immediate connection checks and clear status messages.
- Supports enable/disable, reauth, and unauth flows that are easy to get wrong by hand.
* active agent tool registry runtime reload + token support for bearer auth http based transport
* +session rebind on succesful connection
* fix(coding-agent): address /mcp check failures
---------
Co-authored-by: can1357 <me@can.ac>
- Improved browser script evaluation to handle both expression and statement forms by adding fallback evaluation logic, fixing failures for certain script types.
- Migrated console.error() calls to structured logger.warn() and logger.error() throughout codebase.
- Updated os.tmpdir() import style in browser tool from destructured to namespace import.
- Enhanced skipped file messages to include file size information for better user visibility.
- Updated auto-read skip messages to display actual file size when files are too large to process.
- Inlined MAX_CLI_TEXT_BYTES calculation for improved code clarity.
- Fixed bash tool hanging with background jobs by implementing timeout-based cancellation and draining stdout/stderr pipes instead of waiting indefinitely for EOF.
- Fixed crash when auto-reading large @mentions by adding file size validation (5MB for text, 25MB for images) before reading file contents.
- Prevented out-of-memory issues by skipping files exceeding size limits and displaying file size information in UI for skipped files.
- Added GB file size formatting support to handle very large files in human-readable format.
- Added file size display in UI for skipped files to inform users why files were not included.
- Added test case verifying bash executor returns correctly when command spawns background jobs.
- 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.
- Improved error handling in Python cell execution to preserve and display combined output from previous cells when an error occurs.
- Fixed tab character rendering in Python tool output display to properly format whitespace in cell output and status events.
- Enhanced error messages for failed Python cells to include full combined output context instead of just the error message.
- Updated error cell output styling to use error color theme instead of standard tool output color for better visual distinction.
- Added guidance on formatting as a batch operation in system prompt.
- Expanded patch tool documentation to clarify that formatting should be done once at the end with a single command, not through individual edits.
- Ensured todoItems is cleared when no session file exists or todo file cannot be loaded.
- Added explicit initialization of todoItems to empty array in error and no-session-file cases to prevent stale todo data from persisting.
- Added null check for details.results before accessing length property to prevent potential crash when rendering results with undefined details.results.
- Apply replaceTabs to all text rendering paths in task tool (output,
task section, context, review explanations, finding bodies)
- Strip <swarm_context> block from expanded task view since it is the
shared context duplicated across every task
todo_write renderResult used truncateToWidth with Ellipsis.Omit which
silently drops content, and the empty-todos fallback returned a raw
component that bypassed width handling entirely. Switched all paths
to return Text(text, 0, 0) which applies wrapTextWithAnsi at render
time, preserving full todo content on narrow terminals.
Ask tool renderResult returned raw line arrays that bypassed the Text
component's word wrapping, causing a TUI crash when user input exceeded
terminal width. Switched all renderResult code paths to return
Text(text, 0, 0) which applies wrapTextWithAnsi at render time.
- Fixed schema normalization in Google provider to handle array-valued type fields by converting them to single type with nullable flag.
- Refactored version parsing in install script to use bash parameter expansion instead of IFS field splitting.
- Changed variable declarations from let to var in theme module global scope.
- Added terminal breadcrumb tracking to remember the last session per terminal, enabling `--continue` to work correctly with concurrent sessions in different terminals.
- Implemented `getTerminalId()` function to identify the current terminal using TTY device path or environment variables (Kitty, tmux, macOS Terminal, Windows Terminal).
- Implemented `writeTerminalBreadcrumb()` function to persist terminal-to-session mappings in the agent directory.
- Implemented `readTerminalBreadcrumb()` function to retrieve the last session for the current terminal, scoped to the current working directory.
- Modified session initialization to write breadcrumbs when loading or creating sessions, ensuring terminal context is preserved.
- Removed unused import of resolveToCwd from path-utils module.
- Removed unused resolveArtifactsDir function that was not called elsewhere.
- Prefixed unused ctx parameter with underscore to indicate intentional non-usage.
- Changed screenshot format to always use PNG instead of supporting JPEG with quality parameter.
- Changed default extract_readable format from text to markdown.
- Changed screenshot storage to use temporary directory with Snowflake IDs instead of artifacts directory.
- Changed ResizedImage interface to return buffer as Uint8Array with lazy-loaded base64 data getter for improved memory efficiency.
- Removed JPEG quality parameter from screenshot options.
- Removed ability to save screenshots to custom paths or artifacts directory.
- Fixed tab character display in error messages and bash tool output by applying replaceTabs() sanitization.
- Added TUI rendering sanitization documentation to AGENTS.md with guidelines for sanitizing text in tool renderers.
- Applied replaceTabs() function to error text in patch shared module to ensure proper tab-to-space conversion.
- Applied replaceTabs() function to bash tool output lines in both expanded and collapsed display modes.
- Renamed XML tag `<important>` to `<caution>` across system prompts and tool documentation for improved semantic clarity.
- Collapsed multi-line function calls to single lines in diff component formatting for improved code conciseness.
- Consolidated emoji guideline into critical section in write tool documentation.
- Fixed whitespace visualization in CHANGELOG diff output and corrected incomplete text.
- Fixed patch applicator to preserve indentation in context-only hunks by skipping unnecessary adjustments when pattern equals actual content.
- Implemented linear regression-based tab width inference to correctly convert space-indented diffs to tab-indented files using ax+b model.
- Added regression tests for context-only hunk handling and space-to-tab indentation conversion with offset models.
- Updated patch tool documentation to emphasize never using edit operations for indentation or formatting fixes.
- Added visualization of leading whitespace (indentation) in diff output with dim glyphs--tabs display as ` -> ` and spaces as `·` for improved readability.
- Implemented visualizeIndent() function to normalize and display indentation patterns, collapsing 3-space groups into tab arrows and marking individual spaces with middle dots.
- Fixed tab character rendering in tool output previews and code cell displays by converting tabs to spaces for consistent terminal display.
- Applied replaceTabs() function across patch rendering, write tool output, and code cell output to ensure uniform tab handling.
- 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 background color stability in output blocks when inner content contains SGR reset sequences, preventing background color from being cleared mid-line.
- Fixed spurious ellipsis appended to output lines that were already padded to terminal width by trimming trailing spaces before truncation check.
- Improved hash function to include string length in hash computation for better collision resistance.
can1357