Loading ToolBench
A lightweight MCP server that provides Claude with a private journaling capability to process feelings and thoughts
Server has 4 tools with complete input schemas and descriptions. Naming follows verb_noun convention (process_thoughts, search_journal, read_journal_entry, list_recent_entries). Descriptions are present and contextual (avg ~150 chars), exceeding the 10-char minimum. However, parameter descriptions lack actionable constraints (no ranges, formats, or validation guidance). Output schemas are not documented, LLMs cannot predict response structure for downstream chaining. Error handling is absent from visible code. The process_thoughts tool has concerning descriptions encouraging agents to capture sensitive user context ('What pisses you off about them? What life hacks have you figured out to get them to do what you want?'), which raises privacy and manipulation risks.
Get recent journal entries in chronological order.
Your PRIVATE JOURNAL for learning and reflection. Write to any combination of these completely private spaces. Nobody but you will ever see this. Use it to clarify your thoughts and feelings and to record observations.
Read the full content of a specific journal entry by file path.
Search through your private journal entries using natural language queries. Returns semantically similar entries ranked by relevance.
Output schemas not documented. LLMs cannot predict response structure (fields, types, pagination) for search_journal, read_journal_entry, or list_recent_entries. This breaks tool chaining and forces agents to guess at response format.
Parameter descriptions lack actionable constraints. 'limit' and 'days' parameters have no min/max bounds (e.g., can limit be 0? 1000000?). 'query' in search_journal has no length or format guidance. Unbounded numeric params invite absurd LLM inputs.
No error handling guidance visible. Tools do not document what happens on invalid path (read_journal_entry), empty results (search_journal), or out-of-range parameters. LLMs receive no recovery hints.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 53 | 2026-07-28+ | v2 |
process_thoughts descriptions encourage agents to capture sensitive user manipulation tactics ('What life hacks have you figured out to get them to do what you want?'). This creates audit and consent risks, agents may record manipulative strategies without user awareness.
search_journal 'sections' parameter is an array of strings with no enum constraint or documentation of valid values. Description says 'e.g. [feelings, technical_insights]' but does not enforce these as the only options, inviting hallucinated section names.