Roadmap

Building the Complete Graph Database

Anvil is built phase by phase with full test coverage at every step. Track our progress across 1464 planned tasks.
1207 of 1464 tasks82%
22 Complete5 In Progress2 Planned
0

Project Scaffolding

Complete
8/8
Rust workspace with Cargo
React Router 7 browser app
CI pipeline (clippy, test, fmt)
Rust edition 2024, MSRV pinned
1

Core Graph Data Model

Complete
33/33
Primitive Types— NodeId, RelationshipId, LabelId, PropertyValue enum
Graph Entities— Node, Relationship, Path structs
Schema & Constraints— Unique, existence, composite key constraints
Graph API— Full CRUD trait with traversals and statistics
2

Storage Engine

Complete
30/30
Page-Based Store— Fixed-size pages, LRU cache, mmap option
Record Stores— Node, relationship, property, string, array stores
ID Allocation— Sequential with free-list recycling
Write-Ahead Log— Segmented WAL, fsync, crash recovery, checkpointing
File Layout— graph.db, properties.db, WAL directory, lock file
3

Indexing

Complete
19/19
B+ Tree— On-disk, range scans, bulk loading, concurrent access
Index Types— Unique, composite, full-text (BM25), spatial (R-tree)
Index Management— CREATE/DROP INDEX, automatic maintenance
4

Transaction Engine

Complete
17/17
Transaction Manager— Begin/commit/rollback, savepoints
Concurrency Control— MVCC, 3 isolation levels, deadlock detection
Recovery— ARIES-style redo/undo, WAL replay
Garbage Collection— Version chain pruning
5

Cypher Query Language

Complete
78/78
Lexer— 40+ token types, Unicode, nested comments
Parser— Full grammar: MATCH, CREATE, MERGE, DELETE, SET, WITH, UNWIND, FOREACH, CALL, UNION
Semantic Analysis— Variable scoping, type inference
Built-in Functions— 30+ functions: aggregation, scalar, string, math, list, path
Query Planner— Cost-based optimizer, index selection, join ordering
Query Executor— Volcano model, 25+ operators, multi-hop BFS, ALL SHORTEST
6

GraphQL API

Complete
36/36
Schema Generation— Auto-generated from graph, Relay connections
Queries & Mutations— Filtering, ordering, pagination, nested traversals
Subscriptions— Real-time via graphql-transport-ws
Cypher Passthrough— Raw Cypher via GraphQL
7

Server

Complete
39/39
HTTP Server— Axum/Tokio, CORS, REST + GraphQL + WebSocket
Authentication— JWT RS256, JWKS, Argon2id, per-request sessions
Configuration— TOML config, env vars, CLI flags, system path detection
Multi-Database— Schema isolation, database routing
7.2.1

Email Verification & Passwordless Login

Complete
17/17
SMTP / Email Provider— SMTP with STARTTLS, webhook alternative, HTML templates
Email Configuration— [auth.email] TOML section, 12 env vars, configurable TTLs
Account Verification— Verification token, GET /auth/verify, POST /auth/resend-verification
Login Verification Gate— Blocks login when require_email_verification=true and unverified
Passwordless OTP— POST /auth/otp/request + /auth/otp/verify, 6-digit codes, brute-force protection
Auto-Create User on OTP— New users auto-created with reader role on first OTP login
Rate Limiting— Max 5 OTP requests per email per hour, max 3 attempts per code
JWT Integration— Both flows produce identical JWTs, auth_method in refresh token
Hammer UI— Password/Email Code tabs, OTP flow, resend verification on error
Email Templates— Built-in verification, OTP, password reset templates with custom override
Document Collections— auth.email_verifications, auth.otp_codes with TTL
Unit Tests— 16 tests: OTP generation, verification, login gates, email templates
User Node Sync— Sync email_verified to :User graph node
Token Cleanup— Auto-delete expired verification tokens and OTP codes
Integration Tests— End-to-end OTP and verification flows in test-cypher.sh
8

CLI

Complete
20/20
Server Management— start/stop/status, daemon mode, log file
Interactive Shell— Cypher REPL, history, parameter binding
Data Operations— Import Cypher scripts, config init
9

Hammer UI

Complete
65/65
Query Editor— Cypher highlighting, table/JSON/graph/plan views
Graph Visualization— Force-directed, focus mode, lasso, minimap, export
Schema Browser— Labels, types, keys, indexes, schema dropdown
Document Manager— Collection CRUD, sync rules, schema-filtered
Admin Panel— User/role management, event log, alerts (admin-only)
Data Import— Cypher script upload with creation counts
Role-Based Access— Admin-only nav, schema protection, JWT roles
10

Graph Algorithms

Complete
16/16
Centrality— PageRank, betweenness, closeness, degree
Pathfinding— Dijkstra, A*, BFS/DFS, ALL SHORTEST
Community Detection— Louvain, label propagation, connected components
Other— MST, triangle counting, node similarity
11

Performance & Scalability

Complete
19/19
Query Performance— Plan caching, result caching, parallel operators
Storage Performance— LZ4/Zstd compression, tiered storage
Benchmarking— Criterion benchmarks, LDBC SNB workload
12

Operations & Reliability

Complete
13/13
Backup & Restore— Online snapshots, incremental, point-in-time
Observability— Prometheus metrics, structured logging, OpenTelemetry
Data Integrity— CRC32 checksums, consistency checks
13

Clustering & Replication

Complete
7/7
Raft Consensus— Leader election, log replication
WAL Streaming— Leader-follower replication, read replicas
Sharding— Hash/label-based, rolling upgrades
14

Ecosystem

Complete
22/22
Client Drivers— Rust, TypeScript, Python, Go
Migration Tooling— Neo4j Cypher dumps, CSV, GraphML
Documentation— mdBook site, API docs, tutorials
Testing— 3,270+ tests across 14 crates
15

Plugin System

Complete
98/98
Plugin Framework— Rust + Lua/Python/WASM/JS/TS/Starlark, sandboxed, hot-reload
APOC Library— 55 built-in functions: geo, text, collections, math, date
16

Native Document Store

Complete
57/57
Document Model— Schema-namespaced collections, composite keys, TTL
Query API— Filters, batch ops, paginated scan, REST endpoints
Graph-Document Sync— Bidirectional rules, conflict resolution, INCLUDE clause
Storage Modes— Unified, split, shared/separate WAL
17

Row-Level Security

Complete
48/48
Policy DDL— CREATE/DROP POLICY, ENABLE/DISABLE/FORCE RLS
Predicates— current_user(), current_roles(), session()
Sync Pairs— Label policies auto-protect paired collections
Schema Protection— Label reservation, cross-schema mutation blocking
18

Database Functions

Complete
33/33
Function DDL— CREATE/DROP/SHOW, typed parameters, defaults
Execution— CALL...YIELD, inline in expressions, body caching
Hammer UI— Create/edit/delete, test panel, call log
19

Triggers

Complete
39/39
Trigger DDL— CREATE/DROP/ENABLE/DISABLE, priority ordering
Event Types— BEFORE/AFTER on INSERT/UPDATE/DELETE
Trigger Body— OLD/NEW variables, RAISE, cascade detection
Hammer UI— Management, activity log, dependency analysis
20

Observability

Complete
46/46
Event Log— Ring buffer with configurable retention
Query Analytics— p95/p99 latency, slow query detection
Alerts— Configurable rules, error rate thresholds
Dependency Analysis— Trigger cascades, function calls, sync interactions
21

Data Import & Export

In Progress
7/57
Cypher Script Import— CLI, REST, browser UI, Neo4j-compatible
CSV Import— Header mapping, type inference, batch loading
JSON Import— Nested document to graph mapping
GraphML Import— Standard graph interchange format
Export— Cypher dump, CSV, JSON, GraphML
Example Datasets— Movies, social network, recommendations
22

Edge Functions

In Progress
6/64
Function Runtime— Deno/Node.js subprocess, HTTP triggers
Deployment & Management— Versioning, rollback, resource limits
Database Access— Graph/document queries from function context
Secrets— Encrypted environment variables
Local Development— Hot reload, debug logging, test harness
Hammer UI— Editor, logs, deployment status
23

Hooks

Planned
0/44
Hook Definition— CREATE/DROP/ENABLE/DISABLE HOOK DDL
System Events— ON SERVER START/SHUTDOWN, CONNECT/DISCONNECT, SCHEMA CHANGE
Scheduled Hooks— Cron expressions, distributed lock
Webhook Integration— HTTP POST, retry policy, signature verification
Hammer UI— Hook management, execution history
24

Bolt Protocol

Planned
0/76
PackStream Serialization— Binary encode/decode for all Cypher types
Chunked Transport— TCP framing with Tokio async
Bolt Handshake— Version negotiation (v4.4/v5.0)
Authentication— HELLO/LOGON, credential validation
Query Execution— RUN/PULL/DISCARD streaming
Transactions— BEGIN/COMMIT/ROLLBACK, state machine
Driver Compatibility— Official Neo4j drivers (Python, JS, Java, Go)
TLS Support— Optional bolt+s://
25

File Storage

In Progress
14/16
25.1 Storage Schema— Protected :storage.Bucket / :storage.Object / :storage.Upload labels, unique + content-hash indexes
25.2 Pluggable Backends— Local content-addressed FS done; S3 / GCS / Azure scaffolded, return NotImplemented
25.3 RESTful API— 30+ /storage/v1 endpoints: bucket CRUD, object CRUD, copy/move/list, public, signed, render, usage, TUS
25.4 Cypher Integration— Bucket DDL + file_url / signed_url / file_size / file_hash helpers shipped; MATCH FILE / CREATE FILE deferred
25.5 Access Control (RLS)— RLS auto-enabled on :storage.* labels, public-bucket short-circuit, service_role bypass, default policy set
25.6 Signed URLs— HMAC-SHA256 tokens, separate signing key, configurable TTL, bucket-versioned revocation
25.7 Image Transformations— Pure-Rust pipeline (image crate), width/height/resize/format/quality, content-addressed variant cache
25.8 Resumable Uploads (TUS 1.0.0)— creation / expiration / checksum / termination extensions, staging-dir chunk merge, atomic finalize
25.9 Trigger Integration— BEFORE/AFTER INSERT/UPDATE/DELETE on :storage.Object — reject by MIME, kick off thumbnails, block referenced deletes
25.10 Event Logging— StorageObjectCreated/Updated/Deleted/Accessed, BucketCreated/Deleted, QuotaExceeded, SignedUrlIssued
25.11 Quotas & Limits— Per-bucket / per-user / global byte caps, pre-flight Content-Length check, atomic doc-store accounting
25.12 Storage Configuration— [file_storage] anvil.toml block, 16 ANVIL_STORAGE_* env vars, duration + size suffix parsing
25.13 Client SDK (anvilent)— TypeScript / Rust / Python / Go drivers — storage namespace, .from() builder, TUS, signed URLs, transforms, streaming download
25.14 CLI— anvil storage create-bucket / list-buckets / upload / download / ls / rm / sign / usage / gc — 41 tests
25.15 Browser UI (Hammer)— Bucket browser, drag-drop upload (auto-TUS), inline preview, signed URL generator, public-link panel, usage dashboard
25.16 Documentation— 12 in-route docs sections under File Storage group: overview, quickstart, buckets, objects, cypher, rls, signed, transforms, resumable, backends, triggers, tutorials
26

Service Accounts & Long-Lived Tokens

In Progress
37/45
Service Account Model— Non-login principals with scoped roles, optional service_role flag, full AuthStore lifecycle
API Key Model— anvil_sk_ keys, SHA-256 hashed, one-time secret display, opaque server-side lookup for instant revocation
Management API— Admin-only /auth/service-accounts CRUD + key mint/list/revoke, persisted to auth.* collections
Middleware Integration— Bearer key auth, throttled last-used stamping, scope/role narrowing, service_role RLS bypass
Security & Auditing— Prefix-only audit events, privilege-escalation guard, scope-subset enforcement
CLI— anvil service-account create/ls/show/rm + key create/ls/revoke
Hammer UI— Accounts panel, create dialog, key drawer, audit tab
Documentation— Overview, quickstart, security guide, config reference
27

Index Integration

In Progress
383/402
Write-Path Maintenance— CREATE/DROP INDEX DDL kept consistent via a single MemGraph IndexObserver hook covering executor, WAL replay, and sync-engine paths
Query Planner Integration— Index-backed fast paths for single-property equality and range predicates
Full Index Type Coverage— Composite, fulltext, relationship-property, spatial-radius, and covering/index-only scans
Production Scale-Out— Disk-backed index stores, online ALTER INDEX REBUILD, non-blocking populate, replica convergence, snapshot v7 restart
IndexAdvisor— Slow-query attribution + suggestion engine with memory-budget recommendations
Documentation— Indexes guide and reference
Hammer UI— IndexAdvisor admin panel polish
Checks & Hardening— Phase 27 testing-gap closeout and open items