Database access control, feature by feature — the middle ground between blanket access and a DBA ticket queue.
AccessFlow sits between your team and your databases: every query is parsed, risk-scored, routed to the right approver, and written to an audit log nobody can quietly edit. This page is the whole capability list — twelve of them, grouped by the three things AccessFlow governs plus the platform underneath. Each card is the short version; the three deep dives carry the mechanism. Everything runs inside your own infrastructure under Apache 2.0 — and with a local Ollama analyzer, air-gap friendly.
Database access governance
Five capabilities that follow one query end to end — parsed at the proxy, scored by AI, routed to a reviewer, optionally bundled with other steps, and eventually retired by a lifecycle rule. The deep dive walks the whole path →
/admin/review-plans — a plan decides who must sign off, how long they have, and what happens when nobody does.Full query proxy — SQL and NoSQL
Every query is parsed and policy-checked before it reaches the database. AccessFlow sorts each statement by type — read, write, or schema change — validates the tables it touches against a schema allow-list walked from the parse tree, and routes it through your review policy. NoSQL engines and the cloud warehouses run identical checks, with server-side and destructive commands blocked up front. Layer on row-level security, dynamic masking, and PII/PCI/PHI/GDPR classification tags that raise a query's risk score automatically. How the proxy inspects a query →
PostgreSQL · MySQL · MariaDB · Oracle · MSSQL · ClickHouse · MongoDB · Couchbase · Redis · Cassandra · ScyllaDB · Elasticsearch · OpenSearch · DynamoDB · Neo4j · Snowflake · BigQuery · Databricks · connector catalog · custom drivers · read-replica load balancing · result caching · dynamic masking · row-level security · data classificationAI query analysis
An AI reviewer reads every query before a human does. Each one comes back with a 0–100 risk score, missing-index detection, and anti-pattern flags, plus concrete fixes offered as a draft edit. Pick the provider per organization — Anthropic, OpenAI, Ollama, any OpenAI-compatible backend, or Hugging Face — and tune the analyzer with an editable system prompt. The analysis is evidence a reviewer reads, never a decision on its own. Voting, guardrails and anomaly detection →
Anthropic · OpenAI · Ollama · OpenAI-compatible · Hugging Face · Multi-model voting · Guardrails · Fallback pool · Per-model cost & latency · Optimization suggestions · Text-to-query · RAG · Langfuse · Anomaly detection (UBA) · Approval likelihood (advisory) · History dashboardConfigurable review workflows
Decide who must sign off before a query runs, and route each request there automatically. Approval chains are per-datasource and multi-stage, with auto-approved reads, approval timeouts, bulk decisions, and queues scoped to a user or a group. Policy-as-code routing can auto-approve, escalate, or raise the approval bar on query type, AI risk, requester, time of day, or origin. Users can request just-in-time access for a window; genuine emergencies use audited break-glass. Delegation, escalation and where reviewers decide →
State machine · Per-datasource reviewers · Groups · Policy-as-code routing · JIT time-bound access · Break-glass emergency access · Bulk decisions · Real-time co-authoring · Inline comments · Slack approve/reject · Mobile PWA · One-tap push approvals · Email · Slack · Discord · Telegram · Teams · PagerDuty · ServiceNow · Jira · WebhooksRequest chaining & grouping
Bundle several steps into one grouped request, reviewed and approved once, then executed in order. Members mix queries across datasources with calls to governed API connectors, and the builder shows a per-step risk preview beside an aggregate badge. Bundling never weakens a member's policy: the required approvers are the union of every member plan, and you can never approve your own group. Ordered execution, and why there is no rollback →
Grouped requests · queries + API calls · ordered execution · no distributed rollbackData lifecycle & right-to-erasure
Govern when data is retired, not just who reads it. Retention rules target a table, a column set, or a classification tag per datasource, with a retention window, arbitrary conditions, and an action — hard-delete, soft-delete, or pseudonymize — on an optional cron schedule with a dry-run preview. GDPR and CCPA right-to-erasure requests take the same configuration and route through peer review before anything is removed. How erasure is enforced at read time →
Retention · conditions · cron · pseudonymization · peer-reviewed erasureAPI access governance
The same pipeline, pointed at outbound REST, SOAP, GraphQL and gRPC calls instead of SQL — because the shared credentials and the missing audit trail are the same problem in a different protocol.
Governed outbound API calls
Govern outbound API calls with the machinery that governs queries. Register a REST, SOAP, GraphQL, or gRPC connector with its base URL, default headers, and an auth method — API key, bearer, basic, OAuth2, custom header, or mTLS — every secret encrypted and never returned. Upload a schema and AccessFlow parses a normalized operation catalog with read/write classification, then shares governed connectivity through per-user permissions. Calls then flow through the same AI scoring, routing, approval, and audit path. API access governance in depth →
REST · SOAP · GraphQL · gRPC · OpenAPI / WSDL / SDL / proto / PostmanDeployment approval governance
A release is an access decision too. The same reviewers, routing policies and audit trail gate a production deploy, behind a gate your CI job blocks on until the release is actually approved.
Gated CI/CD releases
Gate CI/CD deployments behind the same review machinery as a query. Define a pipeline with ordered environments — each with its own required approvals, review plan, and break-glass opt-in — and grant trigger rights per user or group. A CI job opens a request with an API key, the AI analyzer scores the release, routing policies set the bar, and reviewers decide in the UI, where nobody can approve their own deployment. The job then blocks on a fail-closed gate. Deployment approval governance in depth →
GitHub Actions · GitLab CI · Azure Pipelines · Jenkins · CircleCI · Bitbucket · freeze windows · break-glass · rollback trackingPlatform, auth & operations
The parts that hold the rest up: who may sign in, what gets recorded, what you see when you land, and how the whole thing is deployed and declared as code. These have no deep-dive page — the operator docs carry them.
Workforce-ready auth
Sign in with the accounts your company already runs — SAML 2.0 SSO and OAuth 2.0 / OIDC, with templates for Google, GitHub, Microsoft and GitLab and a generic provider for any other IdP. Roles and group memberships sync on every login, with optional TOTP two-factor. SCIM 2.0 provisioning hands the whole lifecycle to the IdP: a leaver's sessions and standing access grants are revoked the moment they are offboarded. How authentication is wired →
JWT · SAML · OAuth · SCIM provisioning · IdP group sync · TOTP · Multi-tenant orgs · Per-org quotasTamper-evident audit & compliance reports
Every decision lands in an append-only, cryptographically chained log that cannot be quietly edited after the fact — searchable, exportable, and holding no row data. Pre-built compliance reports cover classified-data access and the DDL/DELETE trail with approver names, export as digitally signed PDF or CSV that verify offline, and are surfaced to a dedicated read-only Auditor role. The chain, the exports and the SIEM sinks →
HMAC-SHA256 chain · signed exports · SIEM streaming · WORM archivalPersonalized dashboard
A self-scoped home that answers "what's waiting on me?" — pending approvals, your recent queries with status and risk trends, an AI optimization backlog you can open straight in the editor, and your own anomaly alerts. Drag the widgets into the layout you want and it sticks; export the week as a signed PDF or CSV, or take it as a weekly email digest. Dashboard docs →
Customizable widgets · weekly digest · signed exportCloud-native deploy
Run it on a laptop with one command, or on Kubernetes for production. One docker compose up covers local and small environments; a Helm 3 chart covers Kubernetes, with horizontal autoscaling, pod disruption budgets, and Bitnami Postgres/Redis subcharts. Redis-backed distributed locks keep scheduled jobs safe across replicas. Install & first-run docs →
Docker · Helm · KubernetesInfrastructure as Code
Manage AccessFlow the way you manage the rest of your stack. An official Terraform / OpenTofu provider declares datasources, review plans, routing, row-level-security and masking policies, AI configs, and notification channels as code, applied idempotently. Reusable GitHub Actions and a GitLab CI template wrap provisioning a datasource and submitting a governed query from a pipeline, each authenticating with a service-account API key. Terraform provider docs →
Terraform · OpenTofu · GitHub Actions · GitLab CI · GitOps · Service-account API keys