Synapsor
Log inGet access
All docs
Documentation

Ship reviewable AI-agent writes to your existing database.

Connect selected Postgres/MySQL tables read-only, expose approved capabilities, record evidence, stage risky updates as row diffs, approve them, and apply approved changes through a trusted runner.

Start with agent database change control if your app already owns Postgres or MySQL. Use Synapsor-native tables later when workflow state belongs inside Synapsor and needs real branch, settlement, and merge semantics.

Run the write-proposal demo Start the quickstart OpenAPI spec
Start with the change-control path
Run the write-proposal demo

See one agent-requested row update become evidence, a row diff, approval, trusted-runner handoff, and replay.

Connect Postgres/MySQL

Select tables, fields, tenant scope, and object keys for governed read access.

Install the trusted runner

Apply approved Postgres/MySQL write proposals inside your environment with a scoped runner token.

Inspect a proposal and replay

Find the stored evidence, row diff, review decision, writeback result, and replay record later.

Advanced: store workflow data in Synapsor

Existing database change control is the default path. When the workflow data itself belongs in Synapsor, use Synapsor-native tables for real branch, settlement, and merge semantics.

Synapsor-native quickstartBranching modesNative workflowsSettlement policies
Agent Activity and evidence lookup

When a human reviewer starts from a ticket, invoice, order, customer, proposal, query fingerprint, or time window, Agent Activity is the indexed lookup surface that finds the stored run, evidence bundle, approval or settlement, proposal diff, replay record, and query audit.

Agent ActivityFind evidence laterEvidence bundlesReplay
Read this first
What is Synapsor?QuickstartExisting DB modesExisting DB UI setupExisting DB quickstartValue sourcesExisting DB live demoWrite proposal live demoAgent contextsCapabilitiesWorkflow examplesEvidenceFind evidence laterAgent ActivityWrite proposalsReplayKnown limitations

Comparing Synapsor with an existing Postgres/MySQL app? Start with the existing database demo, then read the connector docs and Labs pages.

Using an LLM to read these docs?

Give it Synapsor's LLM context first so it understands the custom SQL syntax, controlled-beta limits, and write proposal workflow before generating examples.

llm-prompt.txt
Read https://synapsor.ai/llms.txt first, then use https://synapsor.ai/llms-full.txt for detailed Synapsor syntax. When writing code, keep tenant scope in SESSION/HIDDEN bindings, store returned run/evidence/proposal lookup ids with your app audit row, and stage risky writes through EXECUTION PROPOSAL with auto_branch.
llms.txtllms-full.txtSyntax glossaryValue sources

Start here

Understand database MCP commit safety and create one reviewable agent database change.

Open

Database MCP

Expose semantic MCP tools, bind trusted context, create proposals, and commit through a guarded runner.

Open

Existing database change control

Govern reads, create row diffs, approve proposals, and apply conflict-checked writeback.

Open

Framework integrations

Use Synapsor beside your ORM and agent framework without exposing raw SQL to the model.

Open

Synapsor-native — advanced

Use real Synapsor branch, settlement, memory, search, and workflow state when data lives in Synapsor.

Open

Reference

REST, SQL, SDKs, CLI, system views, and generated API contracts.

Open

Operations and beta status

Controlled-beta limits, production-oriented architecture, cloud operations, labs, and private previews.

Open