Loading ToolBench
MCP-native, local-first memory search and distillation for AI conversations
RecallNest provides 41 tools with consistent naming (verb_noun pattern) and descriptions for all tools. However, parameter schemas lack depth: most parameters have type and description, but many lack constraints (enums, ranges, patterns). Output schemas are not documented. Error handling guidance is minimal. Tool composition is reasonable but some tools combine multiple concerns (e.g., resume_context does search + checkpoint synthesis). Descriptions average ~80 chars, below the 194-char baseline for A+ tools. No tool annotations (readOnlyHint, destructiveHint) despite clear risk classifications.
Audit memory conflicts for patterns and insights.
Automatically capture and store relevant conversation segments based on heuristics.
Store multiple memory entries in a single batch operation.
Create a brief summary of memory entries.
Create a checkpoint of the current conversation state for later resumption.
Clean and update dirty briefs.
Consolidate similar memories into unified entries.
Output schemas not documented. Tools return results but LLMs cannot predict field structure, forcing trial-and-error parsing and wasting tokens.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Tools are classified by risk (READ_ONLY, WRITE, DESTRUCTIVE) but this metadata is not exposed to clients, preventing safe agent planning.
Parameter constraints missing. Many parameters (e.g., 'tier' in promote_memory, 'format' in export_graph) accept enums but lack explicit enum definitions in schema. LLMs guess valid values.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 60 | 2026-07-28+ | v2 |
Run a data quality checkup on the memory store.
Distill a memory into a concise summary.
Distill a session into a concise summary.
Generate creative associations and insights from memories.
Escalate unresolved conflicts for user review.
Generate an explanation of a memory entry.
Export memory relationships as a graph.
Export memory entries to a file.
Permanently delete a memory entry.
Import conversation history into memory.
Retrieve the most recent checkpoint for a scope or session.
List all pinned and exported memory assets.
List memory conflicts detected by the system.
List briefs that need cleaning or updating.
List all pinned memory assets.
List available MCP tools with descriptions and tier information.
Drill down into memory details and relationships.
Lint memory entries for quality issues.
Get statistics about the memory store.
Pin a memory as a persistent asset for easy reference.
Promote a memory to a higher tier or category.
Resolve a memory conflict by selecting the correct version.
Compose a resume context from the latest checkpoint and scoped memory search.
Retrieve a stored skill by name or query.
Scan for skills that should be promoted.
Search indexed memories by semantic similarity and keyword matching.
Set a prospective memory reminder for future recall.
Store a case memory with problem, solution, and outcome.
Store a new memory entry with optional metadata and conflict detection.
Store a reusable skill or technique.
Store a reusable workflow pattern for future reference.
Build evidence pack for workflow analysis.
Inspect workflow health and performance metrics.
Store an append-only workflow observation for self-evolution. Use this to record whether a continuity primitive or reusable workflow succeeded, failed, was corrected by the user, or was missed entirely.
Error handling guidance absent. Tools do not document what errors can occur, whether they are retryable, or what the LLM should do next. Agents cannot self-correct on failures.
Descriptions are brief (avg ~70 chars vs 194-char baseline). Many lack context on WHEN to use the tool vs similar ones. E.g., 'distill_memory' vs 'brief_memory' distinction unclear.