Loading ToolBench
Browse, search, and analyze AI coding-assistant conversation history from GitHub Copilot, Claude Code, Cline, Roo Code, Cursor, Windsurf, Aider, and Google Antigravity. Provides semantic search, keyword search, session retrieval, and context-aware tools for accessing chat history.
ChatWizard demonstrates solid definition quality with 11 tools, all properly named with action verbs (find_, get_, list_, search_, sessions_for_). Tool descriptions are consistently well-written (avg ~180 chars), explaining WHAT the tool does, WHEN to use it, and prerequisites. Input schemas are comprehensive with type definitions and parameter descriptions present. However, output schemas are NOT documented in the provided source, a critical gap. All tools are read-only (no state mutation), which simplifies error handling but leaves recovery guidance sparse. Parameter naming is consistent and tool composition is sound (each tool does one thing), but the lack of output schema documentation and missing error recovery patterns prevent a higher score.
Semantic similarity search across all indexed chat sessions. Returns sessions whose content is semantically related to the supplied topic or question. More powerful than keyword search for conceptual queries. Requires semantic search to be enabled and the model to be initialized.
Smart context retrieval: finds the most relevant past sessions for a topic by combining semantic similarity search (when available) with keyword search. Deduplicates results and returns top passages with full session attribution. Preferred over individual search tools when you want the best relevant context in one call.
Retrieve the content of a specific chat session by ID, truncated to avoid context overflow. Defaults to 4000 characters. Use chatwizard_get_session_full for the complete untruncated content.
Retrieve the complete, untruncated content of a specific chat session by ID. Prefer chatwizard_get_session for large sessions to avoid context window overflow.
List recently updated chat sessions, sorted newest first. Optionally filter by source (AI tool) or a minimum date. Returns lightweight summaries suitable for orientation and triage.
Output schemas not documented. The tool definitions show input schemas but do not explicitly specify what fields and types are returned. LLMs cannot plan downstream tool calls or extract structured data without knowing response structure.
No error recovery guidance. Tool descriptions do not explain what the LLM should do if a query returns no results, if semantic search is unavailable, or if a session ID is invalid. Error responses should guide the agent's next step.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 66 | 2026-07-28+ | v2 |
List which AI coding tools are indexed by ChatWizard, along with session counts and the most recent session date per source. Use this to understand what data is available before issuing other queries.
Full-text keyword search across all indexed chat sessions. Returns sessions whose messages contain all supplied keywords, with a snippet of the matching content. Optionally filter by source (e.g. "copilot", "claude", "cursor"), workspaceId, or entity type/value (e.g. file paths, function names, errors, decisions).
Return ChatWizard MCP server metadata and health information. Includes extension version, total session count, indexed AI tool sources, semantic search availability, and server uptime. Use this to verify connectivity and understand current index state.
Returns sessions associated with a git branch name. Matches branch names against session titles and first user messages.
Returns sessions that referenced or modified a specific source file. Uses Chronicle checkpoint data (important_files) and extracted entity file paths. Provide either an absolute path or a workspace-relative path.
Returns sessions associated with a work-item ID such as a JIRA ticket (ABC-123), GitHub issue (#123), or Azure DevOps item (AB#12345). Matches against session titles and all messages (up to a character budget).
Parameter constraint documentation inconsistent. 'limit' parameters vary in description detail (some specify '1 - 50', others just say 'maximum'). Enums like 'source' in chatwizard_search and chatwizard_list_recent should have explicit enum constraints in schema, not just comma-separated descriptions.
chatwizard_get_session vs chatwizard_get_session_full naming is slightly unclear. While the names do distinguish 'truncated' from 'full', the tool descriptions could emphasize when each should be used (e.g. 'prefer get_session for large sessions to avoid context overflow').