remembero Back home

Same database. Same model. Different powers.

Both lanes share one browser-local SQLite database and the same model. The only difference is what the tool can return — and these four cases pick questions where that difference is structural: SQL hands back rows, silence, or a silently corrupted state; Remembero hands back a checkable proof, a named missing premise, or a refused contradiction.

Model modeSQLite data ready · WebLLM model not loaded
Recall depth3 recalled facts · 1 rule · 1 absence check
Data sourceStarting SQLite WebAssembly…
SQLite bootnot run · shared in-memory database
Tool executionSQL not run · Remembero not run
Answer contractAwaiting model output
Scenario
On demand · about 4.0 GB VRAM · weights cached by the browser
Ask

Give both agents the same question and shared SQLite database.

Model calls

Hermes emits a native function call before it can answer.

SQLite executes

One tool returns raw SQL rows; Remembero returns bindings and proof.

Answer

The tool result goes back to the same local model for synthesis.

Data-only agentQuery → prepared SQL · raw rows
MODEL PROMPT:
Call the Query function exactly once with query "What is really blocking the Atlas review?".
Do not rename the function, invent another function, or write SQL or Datalog yourself.
After the tool result is returned, answer the question using only that evidence and cite its concrete values.
Question: What is really blocking the Atlas review?

TOOL SCHEMA:
{
  "name": "Query",
  "description": "Run the prepared read-only SQL query for the active case and return raw SQLite rows.",
  "parameters": {
    "type": "object",
    "properties": {
      "query": {
        "type": "string",
        "enum": [
          "What is really blocking the Atlas review?"
        ]
      }
    },
    "required": [
      "query"
    ],
    "additionalProperties": false
  }
}
Model tool call · not run
Not run yet
Validated call
Not validated
Executed SQL
Not executed
Tool result
Not executed
Final answer prompt
Not run yet
Remembero agentQuery → Remembero · bindings + proof
MODEL PROMPT:
Call the Query function exactly once with query "What is really blocking the Atlas review?".
Do not rename the function, invent another function, or write SQL or Datalog yourself.
After the tool result is returned, answer the question using only that evidence and cite its concrete values.
Question: What is really blocking the Atlas review?

TOOL SCHEMA:
{
  "name": "Query",
  "description": "Evaluate the prepared Remembero relation for the active case and return bindings plus proof from SQLite.",
  "parameters": {
    "type": "object",
    "properties": {
      "query": {
        "type": "string",
        "enum": [
          "What is really blocking the Atlas review?"
        ]
      }
    },
    "required": [
      "query"
    ],
    "additionalProperties": false
  }
}
Model tool call · not run
Not run yet
Validated call
Not validated
Executed Datalog
Not executed
Tool result
Not executed
Final answer prompt
Not run yet
Shared SQLiteStarting SQLite WebAssembly…
  • One in-memory SQLite database for both agents
  • Startup probe: not verified
  • Data Query adapter returns prepared, read-only SQL rows
  • Remembero Query adapter uses datalog_query + datalog_explain
  • No database snapshot is pasted into either model prompt
SELECT
  (SELECT COUNT(*)
     FROM sqlite_schema
    WHERE type = 'table'
      AND name IN (
        'status',
        'blocker',
        'waits_on',
        'prefers_meeting',
        'pending_meeting',
        'review_slot'
      )) AS tableCount,
  (SELECT COUNT(*) FROM status)
    + (SELECT COUNT(*) FROM blocker)
    + (SELECT COUNT(*) FROM waits_on)
    + (SELECT COUNT(*) FROM prefers_meeting)
    + (SELECT COUNT(*) FROM pending_meeting)
    + (SELECT COUNT(*) FROM review_slot) AS rowCount
Answer + proof contractroot_blocker(atlas, Root)
Data-only final raw
Not run yet
Remembero final raw
Not run yet
Awaiting model output
  1. Proof output appears after the recall run.

Data only

Question → model tool call → prepared SQL → raw rows

Plugin off
User

What is really blocking the Atlas review?

Answer

Run the comparison to see the baseline answer.

With Remembero

Question → model tool call → Datalog → bindings + proof

Memory packet
User

What is really blocking the Atlas review?

Answer

Run the comparison to execute recall, proof, and answer synthesis.

Model calls use a labeled simulator (SQLite data ready · WebLLM model not loaded); SQLite, SQL, Remembero evaluation, and proof execution are real and browser-local.

Add this loop to your agent harness →