Capabilities¶
ForgeOS has grown through milestones H1-H49. This page groups those milestones into public capabilities instead of exposing them as an implementation timeline.
Capability map¶
| Capability | Enables | Main commands | Related docs |
|---|---|---|---|
| Compiler and graphs | Deterministic app understanding | forge generate, forge inspect all --json |
Architecture |
| Runtime entries | Commands, queries, liveQueries, actions, workflows | forge run, forge query, forge live |
Runtime Model |
| Data and policies | Tenant-scoped data, RBAC, RLS | forge policy, forge rls check |
Security and Data |
| Secrets and env | Safe secret access and env validation | forge secrets, forge env check |
Security and Data |
| Frontend contract | Hooks, routes, capability map | forge inspect frontend --json |
Frontend Integration Guide |
| LiveQuery | Durable reactive updates | forge live status --json |
Frontend |
| Package intelligence | Safe integrations and SDK API evidence | forge add, forge deps api |
Package Intelligence |
| Agent contract | Machine-readable project context | forge agent print-context --json |
Agent Contract |
| Guided dev loop | Central local diagnostics and next actions | forge dev, forge do |
Dev Loop |
| Authoring | Scaffold resources, UI, commands, workflows, AI chat | forge make, forge feature |
Authoring |
| Refactor safety | AST-aware rename/extract codemods | forge refactor |
Codemods |
| Testing and repair | Impact tests, repair plans, review, UI smoke | forge impact, forge repair, forge review, forge ui |
Testing and Repair |
| Native AI agents | Tools, agents, auto-tools, chat endpoints | forge ai, forge inspect agent-tools |
AI Agents |
| DeltaDB and timeline | Local operation memory, inferred sessions, semantic timeline, explain | forge delta, forge session, forge timeline, forge explain |
Forge DeltaDB |
| Agent memory bridge | Redacted external agent event ingestion and MCP context | forge agent memory, forge agent ingest, forge mcp serve |
Agent Memory |
| External runtimes | Executable manifest-backed Go, Java, and Spring Boot services | forge manifest, forge run, forge query |
Forge Protocol |
| Brownfield import | Static inventory for existing TypeScript/JavaScript apps | forge import analyze, forge inspect imported |
Brownfield Import |
| Operations | Self-host, release, Windows, Node path | forge doctor, forge self-host, forge verify |
Operations |
How the pieces fit¶
source
-> compiler graphs
-> runtime guards
-> generated contract
-> dev diagnostics
-> agent workflows
-> verification and repair
The core idea stays the same across every capability: make the app explicit enough that agents can operate it safely.
Milestone grouping¶
| Milestones | Public capability |
|---|---|
| H1-H3 | Compiler, examples, DataGraph |
| H4-H8 | Runtime, persistence, outbox, workflows |
| H9-H12 | Telemetry, auth/policy/tenant, secrets, AI workflows |
| H13-H17 | Queries, client SDK, liveQuery, React/Vue bindings, frontend template |
| H18-H24 | Self-host, agent contract, production auth, RLS, package planner, release bridge, production liveQuery |
| H25-H32 | Authoring, blueprints, codemods, impact tests, repair, adapters, review, UI bridge |
| H33-H38 | Intent router, full-stack contract, capability map, template hygiene, dev panel, fast generated checks |
| H39-H43 | Showcase app, Windows/Node hardening, verify observability, native AI tools and agents |
| H44-H47 | DeltaDB local recorder, inferred work sessions, change grouping, semantic timeline |
| H48 | Agent memory bridge for Codex, Claude Code, Cursor, and MCP |
| H49 | Brownfield import analysis for existing TypeScript/JavaScript apps |
Recommended reading paths¶
For new app builders:
For AI coding agents:
For production hardening:
For existing apps: