FridayOS

Mission Control for serious AI agent work.

One premium control plane for Hermes, Claude Code, Codex, Perplexity, providers, MCP tools, workspaces, review queues, approvals, runtime history and live operational signals.

Control rooms4+Hermes, Codex, Claude Code, Perplexity and more

Runtime surfaces20+agents, providers, reviews, jobs, ledgers and artifacts

Refresh loop8snear-realtime cockpit polling with stale/error handling

Safety posturegatedworkspace policy, write policy, firewall and approval layers

Hermes

Codex

Claude Code

Perplexity

LIVEFridayOSpolicy · routing · ledger

Agent network

◎

Mission Control

Live snapshot and command cockpit

◈

Hermes

Research, orchestration and planning

⌘

Claude Code

Refactor, code review and implementation

</>

Codex

CLI coding worker and verification loops

⌁

Perplexity

Fresh source research and citation support

▧

MCP

Filesystem, Context7 and tool contracts

✓

Review Queue

Approval gates for sensitive actions

▣

Ledger

Replayable runtime records

Control-plane flow

flow

Observe

See every agent, provider, workspace and status in one cockpit.

flow

Route

Use workspace policy to decide which agent and provider should handle the job.

flow

Review

Hold sensitive provider connections, writes and external effects behind approval gates.

flow

Replay

Use ledger history and activity records to understand what happened and why.

How FridayOS operates

Observe

Snapshot the agents, providers, runtime stores, workspace map and readiness checks.

Decide

Apply identity, workspace, provider, firewall and write-policy rules before work starts.

Execute

Route the job to the right agent/provider pair while preserving approval and budget constraints.

Prove

Persist activity, usage, artifacts, review decisions and ledger entries for replayable execution.

Capability matrix

Built for agent teams that need control, not just chat.

FridayOS turns agents, tools, providers and approvals into a visible operating system. Every surface should explain what is allowed, what is blocked, where work is routed and what happened afterwards.

Identity + status

Agent command

Give every agent a role, permission tier, provider route and operational state instead of treating agents like loose chat windows.

Latency + policy

Provider routing

Route work across OpenAI, Anthropic, Gemini, Perplexity, local models and future providers with fallback and budget rules.

Allowlist + purpose

Workspace boundaries

Separate Thea, WorldMind, Rendetalje, Renos and FridayOS tasks so tools, memory and context stay scoped.

Approve + block

Review gates

Hold external effects, production impact and sensitive writes behind explicit review decisions before execution.

Trace + replay

Runtime ledger

Record jobs, artifacts, reviews, activity, usage and black-box audit events so execution becomes explainable.

Tools + contracts

MCP surface

Prepare a real operating layer for MCP tools, filesystem access, Context7, Sequential Thinking and future adapters.

Real workspaces

workspaceTekUp

Coordinate product, code and agent infrastructure work from one cockpit.

workspaceProject Thea

Keep health-assistant safety, verification and release gates visible before shipping.

workspaceWorldMind

Track simulation agents, saves, runtime state and creative production pipelines.

workspaceRendetalje

Use agent workflows for customer ops, billing, cleaning plans and follow-up automation.

workspaceRenos

Turn planning, QA, security and product execution into traceable agent work.

Ship the operating layer

Make every agent action observable, gated and replayable.

Start in Mission Control, inspect the live snapshot, then move into providers, jobs, reviews, ledgers and artifacts.

Open Mission Control Check Readiness

Landing narrative