The agent already knows you

AI agents,
forged for you.

Every other platform gives you a chatbot you have to teach over time. Forge interviews you across 27 fields before writing a single line. Your agent boots already knowing your role, your hard rules, your pet peeves. Then it runs on your hardware with your keys.

Build your agentSee how Phase 1.5 works
Your keys·Your hardware·No telemetry·Runs offline
Forge architect mid-conversation, capturing Phase 1.5 context

myagentos.ai/create — real product, real chat

myagentos.ai/create — Phase 1.5 Context Interview

Live recreation of Phase 1.5 capture

Not another chatbot wrapper.

Three things separate a Forge agent from everything else you've tried.

01

Phase 1.5 Context Interview

The architect captures 27 fields across 5 waves before generating code. Who you are, how you decide, what makes you throw your laptop. The agent ships pre-loaded with you.

27 fieldscaptured per agent
02

28 capabilities, all runnable

Memory, sub-agents, sandbox, gateway, ACP, file ops, code execution, web search, email, Slack, SMS, GitHub, Notion, Linear, calendar, webhooks, SQLite, logs — every capability ships real code. Drop your credentials in .env, the agent uses them.

28 / 28runnable today
03

Every byte in your zip

The runtime is vendored. No phone-home, no telemetry, no SaaS dependency. Your agent runs on your laptop, your container, or fully air-gapped. BYOK end-to-end.

zeroexternal calls
0
Context fields
0
Capabilities
0+
Real tools wired
0
External calls
What gets generated

Your context becomes your agent's bootloader.

Phase 1.5 answers flow directly into your workspace files. The agent reads these every time it boots — no fine-tuning, no embeddings, no opaque inference. Just markdown the agent and you can both edit.

  • USER.mdwho you are, in your words
  • SOUL.mdvoice, archetype, anti-patterns
  • STANDING_ORDERS.mdhard rules and escalation triggers
  • IDENTITY.mdname, emoji, catchphrases
  • AGENTS.mdoperational manual the agent reads
src/agent/workspace/USER.md
Captured during Phase 1.5. Your identity, role, expertise, communication style. Read on every boot.
# USER.md

## Identity

- Name: Dan
- Role: Run the CIQ Startup Program. Serial founder.
  Runs multiple workstreams in parallel — sales, AI research,
  internal tooling, advising founders. Context-switches 30+ times/day.
- Timezone: PST. Working hours 8am-7pm.
- Heartbeat: 7:30am daily, 6pm wind-down summary.

## Communication preferences

- Voice: terse
- Decision style: data + opinion. Quantify it, then recommend.
- Response length: 1-3 sentences default. Long only when warranted.

## Sharp domains (skip the basics)

- Startup operations, distribution, agentic AI architecture
- Sales process, founder psychology, LLM economics
- Rocky/RHEL infrastructure, code review (Python/TS/Rust)

## Where the agent fills in

- Calendar archaeology, email triage, meeting prep
- Follow-up tracking, saying-no drafts, end-of-day summaries

## Avoid (immediate distrust)

- "I'd be happy to help" / "Great question!"
- Corporate filler, fake enthusiasm
- Excessive disclaimers, soft-pedaling
- 5 clarifying questions when 1 would do
- Unsolicited emoji, "circle back", "touch base"
read on every agent boot · merged into the system prompt
How it works

From idea to running agent in one session.

No fine-tuning. No vector store setup. No infrastructure decisions before you've even described what you want.

01

Describe it

Tell the architect what you want the agent to do. One paragraph is enough.

02

Get interviewed

Phase 1.5 runs: 5 waves, 27 fields. Your role, hard rules, voice, escalation triggers, pet peeves.

03

Watch capabilities wire

The architect surfaces the design with all 28 capabilities pre-wired. You strike out what you don't want.

04

Ship and run

Click Build. Download the zip. Drop your keys in .env. python -m <agent>. Done.

28 capabilities. Wired by default.

Six categories. Every agent ships with all 28 — you opt out of what you don't want, not in to what you do.

Core
5
  • ·Persistent memory
  • ·Scheduled commitments
  • ·Heartbeat loop
  • ·Skills
  • ·Standing orders
Execution
5
  • ·Sub-agent spawning
  • ·Python execution
  • ·Shell execution
  • ·Flows
  • ·Sandbox
Communication
6
  • ·Email (IMAP)
  • ·Slack
  • ·SMS
  • ·Webhooks
  • ·Gateway (HTTP+WS)
  • ·TTS
Data
4
  • ·File ops
  • ·Web search
  • ·Web extract
  • ·SQLite
Integration
6
  • ·GitHub
  • ·Google Calendar
  • ·Notion
  • ·Linear
  • ·ACP (multi-CLI)
  • ·Plugins
Observability
2
  • ·Conversation logging
  • ·Metric tracking

Full reference in the documentation.

Built for the work that doesn't fit a SaaS box.

The agent already knows your domain. The capabilities are already there. You build what nobody else can sell you.

Personal

Chief of staff

  • →Catches what you drop across email, Slack, calendar
  • →Drafts replies — never sends without your nod
  • →Pre-loads meeting briefs 15 min before each call
  • →End-of-day "what slipped" + morning heartbeat
Personal

Inbox triage

  • →Reads every inbound email, scores urgency
  • →Tracks who is waiting on you and for how long
  • →Auto-drafts polite-no replies for invites
  • →Surfaces forgotten threads on Sunday evening
Personal

Calendar archaeologist

  • →"What did I commit to 3 weeks ago that I forgot?"
  • →Cross-references DMs against calendar events
  • →Flags conflicts and double-bookings
  • →Knows context for every meeting before it starts
Operational

Trading & monitoring

  • →Watches price feeds, technical indicators, news
  • →Runs your strategy with your risk rules in standing orders
  • →Surfaces setups that match your historical wins
  • →Never executes without explicit confirmation
Operational

Lead intelligence

  • →Monitors Slack, CRM, email for inbound leads
  • →Sub-agents run founder + company dossiers
  • →Scores against your ICP, your win patterns
  • →Drafts personalized outreach with talking points
Operational

Compliance watch

  • →Tails logs, monitors APIs, reads security feeds
  • →Cross-references against your policy in STANDING_ORDERS
  • →Spawns sub-agents to investigate anomalies
  • →Daily audit log + weekly compliance summary

Three ways to run. All sovereign.

Choose how it boots. The agent code, workspace, and vendored runtime are identical — only the deployment shell changes.

Local Python

Fastest to start

Run directly on your machine. Best for development, daily use, dogfooding.

Python 3.10+Any OSBYOK
$ python3 -m venv .venv && source .venv/bin/activate
pip install -e .
python -m <agent>

Rocky Linux container

Production-ready

Isolated container. Best for long-running production agents on your infrastructure.

Podman/DockerRocky 9Hardened
$ ./container/build.sh
./container/run.sh
# Listens on 127.0.0.1:7891 via gateway

Fully sovereign

Air-gapped, no LLM API

Bundles a local LLM (llama.cpp / Ollama) into the image. Zero external dependencies.

Local LLMZero netCompliance
$ ./container/start-sovereign.sh
# All inference local
# No internet required

Bring your own everything.

Your API keys. Your LLM — Anthropic, OpenAI, Gemini, Grok, or any OpenAI-compatible endpoint. Your data never touches our servers. Zero telemetry. Zero data collection. Full source code in every download.

Zero data collectionNo vendor lock-inLinux · macOS · WindowsRun anywhereFull source includedVendored runtime

Stop teaching every AI tool who you are.

Forge captures it once, in your words. Then ships you an agent that already knows.

Build your agent

Free to design and download. You bring your own LLM key at runtime.