Roam Research

# Changelog v2.4.0 - 2026-01-04 - ADDED: `roam status` CLI command to show available graphs and connection status - `roam status` - Display all configured graphs with default and write-protection indicators - `roam status --ping` - Test connectivity to each graph - `roam status --json` - Output as JSON for scripting - ENHANCED: `roam_markdown_cheatsheet` MCP tool now dynamically prepends graph configuration - In multi-graph mode, displays a table of available graphs with write protection status - Shows the exact `write_key` needed for protected graphs - Enables AI models to know graph requirements before making write calls - Single-graph mode shows no additional info (clean output) v2.3.0 - 2026-01-04 - ADDED: `roam rename` CLI command to rename pages - `roam rename "Old Title" "New Title"` - rename by current title - `roam rename --uid abc123def "New Title"` - rename by page UID - Supports multi-graph mode with `-g` and `--write-key` flags - ADDED: `roam_rename_page` MCP tool - Parameters: `old_title` OR `uid` to identify page, `new_title` (required) - Returns: `{ success, message }` - ADDED: `updatePage` SDK function type declaration for page operations - ENHANCED: `RoamBatchAction` type now includes page actions (`create-page`, `update-page`, `delete-page`) v2.2.0 - 2026-01-03 - ADDED: Datalog query support in `roam search` CLI command - `-q, --query <datalog>` - Execute raw Datalog queries directly from CLI - `--inputs <json>` - JSON array of inputs for parameterized queries - `--regex <pattern>` - Client-side regex filter on results - `--regex-flags <flags>` - Regex flags (e.g., "i" for case-insensitive) - Examples: `roam search -q '[:find ?title :where [?e :node/title ?title]]'` - ENHANCED: `roam batch` CLI command reliability - `--simulate` mode for offline validation (no API calls) - Upfront placeholder validation catches `{{ref}}` errors before execution - Partial results output on failure shows created pages for manual cleanup - Better error messages with action index and field details v2.1.0 - 2026-01-03 - ADDED: `roam batch` CLI command for executing multiple operations in a single API call - Reduces rate limit issues by batching operations - Supports 10 command types: `create`, `update`, `delete`, `move`, `todo`, `table`, `outline`, `remember`, `page`, `codeblock` - Placeholder references (`{{name}}`) for cross-command dependencies - Automatic page title resolution (with parallel lookups) - Daily page auto-resolution for `todo` and `remember` commands - Level-based hierarchy for `outline` command - Table expansion to nested Roam structure - `--dry-run` mode for validating without execution - `--debug` mode for troubleshooting - Full spec: [docs/batch-cli-spec.md](docs/batch-cli-spec.md) v2.0.2 - 2026-01-03 - CHANGED: `roam_create_page` now adds "Processed: [[date]]" as last block on the new page - Replaces the previous behavior of adding "Created page: [[title]]" to today's daily page - The "Processed: [[date]]" block naturally links back to today's daily page - Removed `skip_daily_page_link` parameter from MCP tool - Removed `--no-daily-page` flag from CLI `roam save` command v2.0.1 - 2026-01-03 - ADDED: `skip_daily_page_link` parameter to `roam_create_page` MCP tool (removed in v2.0.2) - When `true`, skips adding the "Created page: [[title]]" block to today's daily page - Defaults to `false` (preserves existing behavior) - Useful for programmatic page creation where daily page logging is unnecessary v1.9.1 - 2026-01-02 - Updated: Added --heading to `roam save` CLI v1.9.0 - 2026-01-02 - ADDED: `roam_move_block` MCP tool - Standalone tool for moving a block to a new parent or position - Parameters: `block_uid` (required), `parent_uid` (required), `order` (optional, defaults to "last") - Convenience wrapper around `roam_process_batch_actions` for single block moves - Validates block existence before attempting move - Returns: `{ success, block_uid, new_parent_uid, order }` v1.8.2 - 2026-01-02 feat(cli): add --no-daily-page flag to roam save command (removed in v2.0.2) Introduces the `--no-daily-page` flag to the save command, allowing users to create pages without automatically linking them on the current Daily Page. This is useful for programmatic generation or workflows where a daily log entry is unnecessary. (Note: This feature was removed in v2.0.2 - pages now add a "Processed: [[date]]" block at the end instead of linking from daily page.) Changes: - Update `save` command to pass `noDailyPage` option to page operations. - Refactor `createPage` to conditionally skip the daily page link logic. - Tweak `MemoryOperations` to place the memory tag at the end of the block content rather than the beginning. - Update CHANGELOG.md. v1.8.1 - 2026-01-02 - ADDED: `roam update` CLI command to update block content by UID - `roam update <uid> "New content"` - Update any block - Useful for marking TODOs as DONE: `roam update <uid> "{{[[DONE]]}} ..."` - ADDED: `--parent <uid>` option to `roam save -b` for nested block creation - Create blocks under a specific parent block UID - `roam save -b "Child content" --parent <parent-uid>` - ADDED: `--json` input mode for `roam save` with explicit nesting control - Input format: `[{text, level, heading?}]` - `echo '[{"text":"Block","level":1}]' | roam save --json --title "Page"` - Provides precise control over indentation levels - ADDED: `--no-daily-page` flag to `roam save` to skip "Created page" link (removed in v2.0.2) - Useful when linking to the page from another location (e.g., brainstorm workflows) - `roam save content.md --title "Page" --no-daily-page` - ENHANCED: CLI help text clarifies `-i`/`-e` filter on text content, not tags v1.8.0 - 2026-01-02 - ADDED: TODO/DONE support in CLI commands - `roam get --todo` - Fetch all TODO items across the graph - `roam get --done` - Fetch all DONE items across the graph - `roam get --todo -p "Page Title"` - Filter by page - `roam get --todo -i "term1,term2"` - Include filter (text content only, not tags) - `roam get --todo -e "term1,term2"` - Exclude filter (text content only, not tags) - `roam save --todo "Task text"` - Create TODO on daily page - `echo "Task" | roam save --todo` - Create TODO from stdin - Multiple TODOs supported via newline-separated input - FIXED: TODO search now finds both `{{[[TODO]]}}` and `{{TODO}}` formats - Roam API normalizes `{{[[TODO]]}}` to `{{TODO}}` in storage - Search query updated to match prefix `{{TODO` for compatibility - ENHANCED: TODO output formatting - Results grouped by page with markdown headers - Clean checkbox format: `- [ ] Task (uid)` / `- [x] Done (uid)` - Strips TODO/DONE markers from display for readability - JSON output available with `--json` flag v1.7.0 - 2026-01-01 - ADDED: Nested table support in `roam_create_page` - Tables with `level: 2` or higher are now created as children of preceding text blocks - Uses UID tracking to map each level to its most recent block - Tables at level N become children of the last block created at level N-1 - If no parent found at the expected level, tables fall back to page level v1.6.2 - 2026-01-01 - FIXED: `roam_create_page` table ordering bug - Tables in mixed content were being created at the end of the page instead of inline at their original positions - Root cause: Content processing separated text and tables into two sequential batches rather than preserving original order - Solution: Refactored to process content items in order, flushing pending text batches before each table insertion - Tables now appear correctly after their preceding headings when using the `content` array with mixed types v1.6.0 - 2025-12-31 - ADDED: `roam refs` CLI command to find blocks referencing a page or block - Accepts page titles, `#tags`, `[[Page Names]]`, or `((block UIDs))` - Three output formats: grouped by page (default), `--json` for LLM/programmatic use, `--raw` for piping - `-n, --limit` option to control number of results (default: 50) - ENHANCED: `roam_search_block_refs` MCP tool with new `title` parameter - Find blocks referencing a page title using Roam's `:block/refs` attribute - Captures both `[[page]]` links and `#tag` references semantically - Existing `block_uid` parameter continues to work for `((uid))` pattern searches v1.5.0 - 2025-12-31 - ADDED: Unified `roam` CLI with three subcommands - `roam get` - Fetch pages by title or blocks by UID - `--json` for machine-readable output, `--depth` for child levels, `--flat` for flattened hierarchy - Accepts both `((uid))` and bare 9-character UIDs - `roam search` - Full-text and tag-based search - `--tag` for tag filtering, `--page` for page scope, `-i` for case-insensitive, `-n` for result limit - `roam save` - Import markdown to Roam (replaces `roam-import`) - `--title` for explicit page title, `--update` for smart diff mode preserving block UIDs - Supports both file input and stdin piping - REMOVED: `roam-import` standalone CLI (functionality merged into `roam save`) - ADDED: `commander.js` dependency for robust CLI argument parsing - UPDATED: Package binary from `roam-import` to `roam` v1.4.0 - 2025-12-30 - ADDED: `roam_update_page_markdown` tool - Updates existing pages with new markdown content using smart diff algorithm - Preserves block UIDs where possible, keeping references intact across the graph - Three-phase block matching: exact text → normalized (removes list prefixes) → position-based fallback - Generates minimal batch operations: only creates/updates/moves/deletes what changed - Supports `dry_run` parameter to preview changes without executing them - Returns detailed stats: creates, updates, moves, deletes, and preserved UIDs - Ideal for: syncing external markdown to Roam, AI-assisted content updates, batch modifications - ADDED: Unit test infrastructure with Vitest - Added `npm run test` and `npm run test:watch` scripts - 71 tests covering diff module: matcher, parser, diff computation, action generation - Tests document expected behavior of three-phase matching algorithm v1.3.2 - 2025-12-27 - FIXED: Content duplication bug in `roam_create_page` - Added idempotency check to prevent duplicate content when tool is called multiple times - Root cause: Duplicate tool invocations (e.g., SSE transport retries, client timeouts) caused content to be created twice - If page already has child blocks, subsequent calls return success without adding duplicate content - Added test script: `scripts/test-create-page-duplication.ts` v1.3.1 - 2025-12-27 - ENHANCED: `Roam_Markdown_Cheatsheet.md` attribute usage guidance - Added clear rules for when to use `::` attribute syntax vs bold formatting - Attributes are for queryable metadata across the graph (Type::, Author::, Status::) - Bold formatting (`**Label:**`) should be used for page-specific labels (Step 1:, Summary:) - Added "The Test" decision guide: "Will I ever query for this across my graph?" - Added attribute anti-pattern to the DON'T DO THIS section v1.3.0 - 2025-12-26 - ADDED: `roam_create_table` tool - Abstracts Roam's complex nested table structure into simple headers/rows input - Validates row/column consistency before execution - Converts empty cells to spaces (required by Roam) - Returns table_uid on success - ADDED: Pre-validation for batch actions - Validates all actions before API execution, catching errors early - Checks action types, UIDs, strings, locations, and placeholder references - Returns structured error details with action index and field info - FIXED: Transaction reporting accuracy - `uid_map` only returned on successful transactions (was incorrectly returned on failure) - Prevents LLMs from using invalid UIDs for failed operations - ENHANCED: Structured error responses - New error codes: VALIDATION_ERROR, RATE_LIMIT, TRANSACTION_FAILED - Includes recovery suggestions and retry timing for rate limits - ADDED: Rate limit retry with exponential backoff - Up to 3 retries with configurable delays (1s initial, 2x multiplier, 60s max) - Automatic handling of 429 responses from Roam API - ENHANCED: `roam_create_page` now supports mixed content types - Content array can now include both text blocks and tables - Tables use `{type: "table", level, headers, rows}` format - Reduces MCP calls when creating pages with mixed content (2+ calls → 1 call) - Text blocks remain default (no type field needed for backward compatibility) v1.2.2 - 2025-12-24 - ENHANCED: CORS support for HTTP streaming endpoint - Now supports multiple CORS origins (comma-separated in `CORS_ORIGIN` env var) - Default origins include `http://localhost:5678` and `https://roamresearch.com` - Enables browser-based MCP clients (like Roam extensions) to connect to the server - Added `Access-Control-Allow-Credentials` header for authenticated requests v1.2.1 - 2025-12-20 - ENHANCED: `roam_create_outline` tool to not create empty blocks v1.2.0 - 2025-12-20 - ADDED: Server-side UID placeholder system for batch operations - Use `{{uid:name}}` syntax in batch actions; server generates proper random UIDs - Returns `uid_map` in response showing placeholder → generated UID mappings - Solves LLM random generation unreliability (LLMs can't generate truly random strings) - Exported `generateBlockUid()` function for use across modules - ADDED: Page UID cache to reduce redundant API queries - Server-side in-memory cache for page title → UID mappings - Eliminates repeated lookups for the same pages across operations - OPTIMIZED: Reduced API calls during verification after batch operations - Conditional verification based on batch size (threshold: 5 items) - For large batches, skips recursive child fetching to minimize queries - Reduced retry attempts in block lookup from 15 to 2 per block - ENHANCED: Tool descriptions with rate-limit efficiency guidance - `roam_process_batch_actions`: Marked as most API-efficient for multiple operations - `roam_create_outline` / `roam_import_markdown`: Added API usage notes - `roam_create_page`: Added efficiency tips - ENHANCED: `Roam_Markdown_Cheatsheet.md` with API Efficiency Guidelines section - Tool efficiency ranking (best to worst) - Best practices for intensive operations and page revisions - UID caching tips for LLM usage v1.1.0 - 2025-12-19 - ADDED: `roam-import` standalone CLI tool for importing markdown to Roam - Reads markdown from stdin and creates a new page with the specified title - Automatically links the new page from today's daily page - Usage: `cat document.md | roam-import "Page Title"` or `pbpaste | roam-import "Ideas"` - If page exists, content is appended to it - FIXED: Markdown parser now handles variable indentation (2 spaces, 4 spaces, tabs) - Previously assumed 2-space indentation, causing nested items to flatten to root - Now dynamically detects indentation levels from the document v1.0.0 - 2025-12-16 - OPTIMIZED: Server performance and reliability - Removed deprecated and non-functional SSE (Server-Sent Events) server implementation. - Implemented async file reading and caching for Roam Markdown Cheatsheet and custom instructions to avoid blocking the event loop. - Optimized `roam_fetch_page_by_title` to use a single Datomic query for all title variations (original, capitalized, lowercase) instead of sequential queries. - Replaced insecure `Math.random` with `crypto.randomBytes` for more robust block UID generation. - Added error logging to stderr for better debugging of server startup failures. - Refactored `RoamServer` to reduce code duplication in server initialization. v0.36.4 - 2025-10-03 - FIXED: SSE server implementation on port 8087 - Added missing SSE server setup in `src/server/roam-server.ts` that was previously imported but never instantiated - SSE server now properly creates its own MCP server instance with full tool capabilities - Configured CORS headers and preflight OPTIONS request handling for SSE endpoint - SSE server listens on port 8087 (or next available port) with proper error handling v0.36.3 - 2025-08-30 - FEATURE: Implemented `prompts/list` method for MCP server, returning an empty array of prompts. - FIXED: Removed `roam-markdown-cheatsheet.md` from advertised resources in MCP server capabilities to align with its tool-only access. v0.36.2 - 2025-08-28 - ENHANCED: `roam_datomic_query` tool - Added `regexFilter`, `regexFlags`, and `regexTargetField` parameters for client-side regex filtering of results. - Updated description to reflect enhanced filtering capabilities. v0.36.1 - 2025-08-28 - ENHANCED: `roam_find_pages_modified_today` tool - Added `limit`, `offset`, and `sort_order` parameters for pagination and sorting. v1.36.0 - 2025-08-28 - ENHANCED: `roam_search_for_tag` and `roam_search_by_text` tools - Added `offset` parameter for pagination support. - ENHANCED: `roam_search_for_tag` tool - Implemented `near_tag` and `exclude_tag` parameters for more precise tag-based filtering. - ENHANCED: `roam_datomic_query` tool - Updated description to clarify optimal use cases (Regex, Complex Boolean Logic, Arbitrary Sorting, Proximity Search). v.0.35.1 - 2025-08-23 9:33 - ENHANCED: `roam_create_page` and `roam_create_outline` tool descriptions in `src/tools/schemas.ts` for improved clarity and to guide users toward the most efficient workflow. v.0.35.0 - 2025-08-23 - ENHANCED: `roam_import_markdown` tool - Now returns a nested object structure for `created_uids`, reflecting the hierarchy of the imported content, including `uid`, `text`, `order`, and `children`. - If a `parent_string` is provided and the block does not exist, it will be created automatically. - FIXED: Block ordering issue in `roam_import_markdown` and `roam_create_outline`. Nested outlines are now created in the correct order. - FIXED: Duplication issue in the response of `roam_fetch_block_with_children`. v.0.32.4 - FIXED: Memory allocation issue (`FATAL ERROR: invalid array length Allocation failed - JavaScript heap out of memory`) - Removed `console.log` statements from `src/tools/operations/outline.ts` to adhere to MCP server stdio communication rules. - Optimized `parseMarkdown` function in `src/markdown-utils.ts` to avoid inefficient `lines.splice()` operations when handling mid-line code blocks, improving memory usage and performance. - ENHANCED: `roam_create_outline` tool - Successfully created outlines with nested code blocks, confirming the fix for memory allocation issues. v.0.32.1 - ENHANCED: `roam_create_outline` tool - The tool now returns a nested structure of UIDs (`NestedBlock[]`) for all created blocks, including children, accurately reflecting the outline hierarchy. - Implemented a recursive fetching mechanism (`fetchBlockWithChildren` helper) to retrieve all nested block UIDs and their content after creation. - Fixed an issue where the `created_uids` array was only returning top-level block UIDs. - Corrected the Datomic query used for fetching children to ensure only direct children are retrieved, resolving previous duplication and incorrect nesting issues. - Removed `console.log` and `console.warn` statements from `src/tools/operations/outline.ts` to adhere to MCP server stdio communication rules. - ADDED: `NestedBlock` interface in `src/tools/types/index.ts` to represent the hierarchical structure of created blocks. v.0.32.3 - ENHANCED: `roam_create_page` tool - Now creates a block on the daily page linking to the newly created page, formatted as `Create [[Page Title]]`. v.0.32.2 - FIXED: `roam_create_outline` now correctly respects the order of top-level blocks. - Changed the default insertion order for batch actions from 'first' to 'last' in `src/tools/operations/outline.ts` to ensure blocks are added in the intended sequence. v.0.30.10 - ENHANCED: `roam_markdown_cheatsheet` tool - The tool now reads the `Roam_Markdown_Cheatsheet.md` and concatenates it with custom instructions from the path specified by the `CUSTOM_INSTRUCTIONS_PATH` environment variable, if the file exists. If the custom instructions file is not found, only the cheatsheet content is returned. - UPDATED: The description of `roam_markdown_cheatsheet` in `src/tools/schemas.ts` to reflect the new functionality. v.0.30.9 - FIXED: `roam_fetch_block_with_children` tool to use a more efficient batched recursive approach, avoiding "Too many requests" and other API errors. - The tool now fetches all children of a block in a single query per level of depth, significantly reducing the number of API calls. v.0.30.8 - ADDED: `roam_fetch_block_with_children` tool - Fetches a block by its UID along with its hierarchical children down to a specified depth. - Automatically handles Roam's `((UID))` formatting, extracting the raw UID for lookup. - This tool provides a direct and structured way to retrieve specific block content and its nested hierarchy. v.0.30.7 - FIXED: `roam_create_outline` now prevents errors from invalid outline structures by enforcing that outlines must start at level 1 and that subsequent levels cannot increase by more than 1 at a time. - Updated the tool's schema in `src/tools/schemas.ts` with more explicit instructions to guide the LLM in generating valid hierarchical structures. - Added stricter validation in `src/tools/operations/outline.ts` to reject outlines that do not start at level 1, providing a clearer error message. - Optimized page creation v.0.30.6 - FIXED: `roam_create_page` now correctly strips heading markers (`#`) from block content before creation. - FIXED: Block creation order is now correct. Removed the incorrect `.reverse()` call in `convertToRoamActions` and the corresponding workaround in `createBlock`. - UPDATED: the cheat sheet for ordinal dates. v.0.30.5 - FIXED: `roam_search_for_tag` now correctly scopes searches to a specific page when `page_title_uid` is provided. - The Datalog query in `src/search/tag-search.ts` was updated to include the `targetPageUid` in the `where` clause. v.0.30.4 - FIXED: Tools not loading properly in Gemini CLI - Clarified outline description - FIXED: `roam_process_batch_actions` `heading` enum type in `schemas.ts` for Gemini CLI compatibility. v.0.30.3 - ADDED: `roam_markdown_cheatsheet` tool - Provides the content of the Roam Markdown Cheatsheet directly via a tool call. - The content is now read dynamically from `Roam_Markdown_Cheatsheet.md` on the filesystem. - **Reason for Tool Creation:** While Cline can access local resources provided by an MCP server, other AI models (suchs as Claude AI) may not have this capability. By exposing the cheatsheet as a tool, it ensures broader accessibility and utility for all connected AI models, allowing them to programmatically request and receive the cheatsheet content when needed. - REMOVED: Roam Markdown Cheatsheet as a direct resource - The cheatsheet is no longer exposed as a static resource; it is now accessed programmatically through the new `roam_markdown_cheatsheet` tool. - ADDED: package.json new utilty scripts v.0.30.2 - ADDED: 4x4 table creation example - Created a 4x4 table with random data on the "Testing Tables" page, demonstrating proper Roam table structure. - ENHANCED: `Roam_Markdown_Cheatsheet.md` - Updated the "Roam Tables" section with a more detailed explanation of table structure, including proper indentation levels for headers and data cells. - ENHANCED: `src/tools/schemas.ts` - Clarified the distinction between `roam_create_outline` and `roam_process_batch_actions` in their respective descriptions, providing guidance on their best use cases. v.0.30.1 - ENHANCED: `roam_process_batch_actions` tool description - Clarified that Roam-flavored markdown, including block embedding with `((UID))` syntax, is supported within the `string` property for `create-block` and `update-block` actions. - Added a note advising users to obtain valid page or block UIDs using `roam_fetch_page_by_title` or other search tools for actions on existing blocks or within a specific page context. - Clarified the `block_text_uid` description for `roam_create_outline` to explicitly mention defaulting to the daily page. - Simplified the top-level description for `roam_datomic_query`. - Refined the introductory sentence for `roam_datomic_query`. - ADDED: "Example Prompts" section in `README.md` - Provided 2-3 examples demonstrating how to prompt an LLM to use the Roam tool, specifically leveraging `roam_process_batch_actions` for creative use cases. v.0.30.0 - DEPRECATED: **Generic Block Manipulation Tools**: - `roam_create_block`: Deprecated in favor of `roam_process_batch_actions` (action: `create-block`). - `roam_update_block`: Deprecated in favor of `roam_process_batch_actions` (action: `update-block`). - `roam_update_multiple_blocks`: Deprecated in favor of `roam_process_batch_actions` for batch updates. Users are encouraged to use `roam_process_batch_actions` for all direct, generic block manipulations due to its enhanced flexibility and batch processing capabilities. - REFACTORED: `roam_add_todo` to internally use `roam_process_batch_actions` for all block creations, enhancing efficiency and consistency. - REFACTORED: `roam_remember` to internally use `roam_process_batch_actions` for all block creations, enhancing efficiency and consistency. - ENHANCED: `roam_create_outline` - Refactored to internally use `roam_process_batch_actions` for all block creations, including parent blocks. - Added support for `children_view_type` in outline items, allowing users to specify the display format (bullet, document, numbered) for nested blocks. - REFACTORED: `roam_import_markdown` to internally use `roam_process_batch_actions` for all content imports, enhancing efficiency and consistency. v.0.29.0 - ADDED: **Batch Processing Tool**: Introduced `roam_process_batch_actions`, a powerful new tool for executing a sequence of low-level block actions (create, update, move, delete) in a single API call. This enables complex, multi-step workflows, programmatic content reorganization, and high-performance data imports. - ENHANCED: **Schema Clarity**: Updated the descriptions for multiple tool parameters in `src/tools/schemas.ts` to explicitly state that using a block or page UID is preferred over text-based identifiers for improved accuracy and reliability. - NOTE: **Heading Removal Limitation**: Discovered that directly removing heading formatting (e.g., setting `heading` to `0` or `null`) via `update-block` action in `roam_process_batch_actions` is not supported by the Roam API. The `heading` attribute persists its value. v.0.28.0 - ADDED: **Configurable HTTP and SSE Ports**: The HTTP and SSE server ports can now be configured via environment variables (`HTTP_STREAM_PORT` and `SSE_PORT`). - ADDED: **Automatic Port Conflict Resolution**: The server now automatically checks if the desired ports are in use and finds the next available ports, preventing startup errors due to port conflicts. v.0.27.0 - ADDED: SSE (Server-Sent Events) transport support for legacy clients. - REFACTORED: `src/server/roam-server.ts` to use separate MCP `Server` instances for each transport (Stdio, HTTP Stream, and SSE) to ensure they can run concurrently without conflicts. - ENHANCED: Each transport now runs on its own isolated `Server` instance, improving stability and preventing cross-transport interference. - UPDATED: `src/config/environment.ts` to include `SSE_PORT` for configurable SSE endpoint (defaults to `8087`). v.0.26.0 - ENHANCED: Added HTTP Stream Transport support - Implemented dual transport support for Stdio and HTTP Stream, allowing communication via both local processes and network connections. - Updated `src/config/environment.ts` to include `HTTP_STREAM_PORT` for configurable HTTP Stream endpoint. - Modified `src/server/roam-server.ts` to initialize and connect `StreamableHTTPServerTransport` alongside `StdioServerTransport`. - Configured HTTP server to listen on `HTTP_STREAM_PORT` and handle requests via `StreamableHTTPServerTransport`. v.0.25.7 - FIXED: `roam_fetch_page_by_title` schema definition - Corrected missing `name` property and proper nesting of `inputSchema` in `src/tools/schemas.ts`. - ENHANCED: Dynamic tool loading and error reporting - Implemented dynamic loading of tool capabilities from `toolSchemas` in `src/server/roam-server.ts` to ensure consistency. - Added robust error handling during server initialization (graph, tool handlers) and connection attempts in `src/server/roam-server.ts` to provide more specific feedback on startup issues. - CENTRALIZED: Versioning in `src/server/roam-server.ts` - Modified `src/server/roam-server.ts` to dynamically read the version from `package.json`, ensuring a single source of truth for the project version. v.0.25.6 - ADDED: Docker support - Created a `Dockerfile` for containerization. - Added an `npm start` script to `package.json` for running the application within the Docker container. v.0.25.5 - ENHANCED: `roam_create_outline` tool for better heading and nesting support - Reverted previous change in `src/tools/operations/outline.ts` to preserve original indentation for outline items. - Refined `parseMarkdown` in `src/markdown-utils.ts` to correctly parse markdown heading syntax (`#`, `##`, `###`) while maintaining the block's hierarchical level based on indentation. - Updated `block_text_uid` description in `roam_create_outline` schema (`src/tools/schemas.ts`) to clarify its use for specifying a parent block by text or UID. - Clarified that `roam_create_block` creates blocks directly on a page and does not support nesting under existing blocks. `roam_create_outline` should be used for this purpose. v.0.25.4 - ADDED: `format` parameter to `roam_fetch_page_by_title` tool - Allows fetching page content as raw JSON data (blocks with UIDs) or markdown. - Updated `fetchPageByTitle` in `src/tools/operations/pages.ts` to return stringified JSON for raw format. - Updated `roam_fetch_page_by_title` schema in `src/tools/schemas.ts` to include `format` parameter with 'raw' as default. - Updated `fetchPageByTitle` handler in `src/tools/tool-handlers.ts` to pass `format` parameter. - Updated `roam_fetch_page_by_title` case in `src/server/roam-server.ts` to extract and pass `format` parameter. v.0.25.3 - FIXED: roam_create_block multiline content ordering issue - Root cause: Simple newline-separated content was being created in reverse order - Solution: Added logic to detect simple newline-separated content and reverse the nodes array to maintain original order - Fix is specific to simple multiline content without markdown formatting, preserving existing behavior for complex markdown v.0.25.2 - FIXED: roam_create_block heading formatting issue - Root cause: Missing heading parameter extraction in server request handler - Solution: Added heading parameter to roam_create_block handler in roam-server.ts - Also removed problematic default: 0 from heading schema definition - Heading formatting now works correctly for both single and multi-line blocks - roam_create_block now properly applies H1, H2, and H3 formatting when heading parameter is provided v.0.25.1 - Investigated heading formatting issue in roam_create_block tool - Attempted multiple fixes: direct createBlock API → batchActions → convertToRoamActions → direct batch action creation - Confirmed roam_create_page works correctly for heading formatting - Identified that heading formatting fails specifically for single block creation via roam_create_block - Issue remains unresolved despite extensive troubleshooting and multiple implementation approaches - Current status: roam_create_block does not apply heading formatting, investigation ongoing v.0.25.0 - Updated roam_create_page to use batchActions v.0.24.6 - Updated roam_create_page to use explicit levels v.0.24.5 - Enhanced createOutline to properly handle block_text_uid as either a 9-character UID or string title - Added proper detection and use of existing blocks when given a valid block UID - Improved error messages to be more specific about block operations v.0.24.4 - Clarified roam_search_by_date and roam_fetch_page_by_title when it comes to searching for daily pages vs. blocks by date v.0.24.3 - Clarified roam_update_multiple_blocks - Added a variable to roam_find_pages_modified_today v.0.24.2 - Added sort_by and filter_tag to roam_recall v.0.24.1 - Fixed searchByStatus for TODO checks - Added resolution of references to various tools v.0.23.2 - Fixed create_page tool as first-level blocks were created in reversed order v.0.23.1 - Fixed roam_outline tool not writing properly v.0.23.0 - Added advanced, more flexible datomic query v.0.22.1 - Important description change in roam_remember v0.22.0 - Restructured search functionality into dedicated directory with proper TypeScript support - Fixed TypeScript errors and import paths throughout the codebase - Improved outline creation to maintain exact input array order - Enhanced recall() method to fetch memories from both tag searches and dedicated memories page - Maintained backward compatibility while improving code organization v0.21.0 - Added roam_recall tool to recall memories from all tags and the page itself. v0.20.0 - Added roam_remember tool to remember specific memories as created on the daily page. Can be used throughout the graph. Tag set in environmental vars in config. v0.19.0 - Changed default case-sensitivity behavior in search tools to match Roam's native behavior (now defaults to true) - Updated case-sensitivity handling in findBlockWithRetry, searchByStatus, searchForTag, and searchByDate tools v0.18.0 - Added roam_search_by_date tool to search for blocks and pages based on creation or modification dates - Added support for date range filtering and content inclusion options v0.17.0 - Enhanced roam_update_block tool with transform pattern support, allowing regex-based content transformations - Added ability to update blocks with either direct content or pattern-based transformations v0.16.0 - Added roam_search_by_text tool to search for blocks containing specific text, with optional page scope and case sensitivity - Fixed roam_search_by_tag v.0.15.0 - Added roam_find_pages_modified_today tool to search for pages modified since midnight today v.0.14