Mistri provides 7 file and memory management tools with basic schemas and descriptions. All tools have input schemas with typed properties and descriptions. However, descriptions are minimal (10-50 chars), lacking context on WHEN to use each tool, prerequisites, or recovery guidance. Parameter descriptions are present but terse. No output schemas documented. Error handling is absent from visible code. Tools follow verb_noun naming (read_file, write_file, edit_file) which is good, but lack guidance on mutual exclusivity (edit_file vs write_file), idempotency, or side effects. No tool annotations (readOnlyHint, destructiveHint) despite clear risk profiles (WRITE vs READ_ONLY).

• Expand tool descriptions to 100-200 chars. Include WHEN to use (e.g., 'Use read_file to retrieve file contents for analysis or display. Supports text files up to 10MB.'), prerequisites, and common errors.
• Document output schemas for all tools. E.g., list_files should return {files: [{name, type, size, path}], total_count, next_cursor}. read_file should return {content, encoding, size}.
• Add tool annotations: destructiveHint=true for write_file and edit_file; readOnlyHint=true for read_file, find_in_file, list_files, read_memory. This enables agents to reason about safety.
• Add parameter constraints and examples. E.g., path parameter: 'Relative path from workspace root (e.g., src/main.py). Must not contain .. or absolute paths.' replace_all: 'If true, replace all occurrences; if false, replace only the first.'
• Include recovery guidance in descriptions. E.g., read_file: 'If file not found, use list_files() to verify the path.' edit_file: 'If old_string not found, use find_in_file() to locate the exact text.'

Inferred effective spec: <=2025-11-25.

• Relies on Stateful initialize / Mcp-Session-Id (removed; protocol is stateless) - make each request self-contained