Skip to main content
Glama
littlebigbrains

@littlebigbrain/mcp

by littlebigbrains
OverviewSchemaSourceRelated ServersDiscussionsAdmin
TypeScript
Hybrid

lbb_commit

Idempotent

Write graph facts, retract wrongly added ones, or label search results with relevance grades; supports dry-run validation and idempotent deduplication.

Instructions

Write graph facts, retract them, or label ranked results. mode=facts writes triplets/embeddings/properties; mode=retract removes a wrongly-added fact (by edge or by entity) without a full reset; mode=search_feedback stores query/result relevance labels (Feedback grades: 3=ideal/good, 1=partial, 0=bad; include query, search_id when available, target, rank, score). Explicit idempotency_key wins; when omitted, MCP derives a stable content hash so content-identical retries dedupe. Facts mode defaults edge_idempotency to append; pass skip_unchanged for re-runnable backfills.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
modeNo
graphNoGraph to target; defaults to the connection's graph
dry_runNoValidate a facts commit and return its structured SHACL report without writing. Only supported for mode=facts.
tripletsNo
observed_atNoBackfill timestamp (RFC3339). Records this commit AS OF that instant: stamps transaction time and defaults each triplet's valid_time.start. Replay history in order with observed_at per commit so as-of reads by date work. Omit for live writes.
retract_edgesNomode=retract: specific edges to remove, matched by (source, relation, target).
idempotency_keyNo
search_feedbackNo
edge_idempotencyNoDefaults to append in MCP. Use skip_unchanged for backfills; it skips exact current-edge duplicates and drops evidence-only repeats.
retract_entitiesNomode=retract: entities whose every current edge is removed (a current-state tombstone; the record and its history are kept for as_of reads).
entity_embeddingsNo
entity_propertiesNoTyped scalar attributes per entity. Each item is { type, name, properties }. `properties` is a flat map of field -> value, e.g. { "type": "PERSON", "name": "Ada Lovelace", "properties": { "h_index": 52, "title": "VP", "last_contact": "2026-06-26" } }. Values are coerced to each field's declared type, so a string like "2026-06-26" lands in a date_time field and "52" in an i64 field. (The verbose form [{ field, value: { i64: 52 } }] is also accepted.) Register a field first with lbb_configure evolve_ontology add_property; the commit response echoes written_properties so you can confirm what landed.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed1 schema field changedv0.5.2
    • removedInput schema / properties / branch
      Removed value: -{
-  "description": "Branch to target; defaults to the connection's branch",
-  "type": "string"
-}
  2. Changed1 schema field changedv0.5.1
    • addedInput schema / properties / dry_run
      Added value: +{
+  "description": "Validate a facts commit and return its structured SHACL report without writing. Only supported for mode=facts.",
+  "type": "boolean"
+}
  3. First observedv0.2.1

TDQS

A4.3/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Beyond the annotations, the description adds valuable behavior: explicit idempotency_key wins, omitted keys derive a stable content hash for dedupe, facts mode defaults to append, and retract supports both edge and entity removal. It does not spell out every side effect in the top-level text, but the schema covers tombstone/history preservation for retract_entities.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is dense but economical: five sentences front-load the mode enum, then cover idempotency, feedback grading, and backfill defaults without repetition or filler. Every clause adds operational value.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a complex 12-parameter tool with no output schema, the description covers mode semantics, idempotency, dry_run, backfill timestamps, and feedback labels well. However, mode itself is not required in the schema and the description never states what happens when mode is omitted, nor what the commit returns. Those gaps matter for an agent deciding whether and how to invoke the tool.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The description meaningfully explains the mode enum, idempotency semantics, edge_idempotency behavior, and even the feedback grading scale. Combined with schema descriptions for dry_run, observed_at, retract_edges, and entity_properties, most parameters are actionable. Some params like triplets and entity_embeddings rely on mode prose rather than dedicated descriptions, but the context is sufficient.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description opens with a specific verb and resource: 'Write graph facts, retract them, or label ranked results.' It then enumerates three distinct modes with concrete behaviors, making the tool immediately distinguishable from read/schema siblings like lbb_query and lbb_configure.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description clearly says when each mode is appropriate: retract is for removing wrongly-added facts without a full reset, search_feedback is for storing relevance labels, and skip_unchanged is for re-runnable backfills. It also indirectly points to lbb_configure for registering fields. It stops short of a fully explicit when-not-to-use or comparison with every sibling, so it is not a 5.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Deploy Server

Other Tools