Governance, audit trails, and control for every AI coding assistant. Works with Claude Code, Codex, Gemini CLI, Aider, gptme, Goose, OpenCode, Amazon Q, Cline, Ollama, and more. Free. Open source. Five minutes to set up.
Declarative TOML rules evaluated on every request. Allow, deny, or require-approval. Circuit breakers trip automatically on violations.
Ed25519-signed Merkle DAG of every agent action. SHA-256 hash-chained. Tamper-evident proof that the agent did exactly what it claimed.
11 screens including live agent output, governance rule list, audit chain, diff viewer, traffic monitor, and branch manager.
Agents pause on flagged operations. Approve or deny from the TUI, VS Code, or JetBrains. Every decision recorded in the proof chain.
Exposes governance state, knowledge base, tasks, and agent controls via Model Context Protocol. Agents can query their own governance status.
Launch up to 12 parallel agents with shared governance. Merkle DAG tracks concurrent branches. Fan-out dispatch routes tasks to the right model.
A governance layer that interposes between your AI coding tools and their APIs. Every action is policy-checked, cryptographically attested, and hash-chained into a tamper-evident record. Not logs. Proof of what happened, in what order, under what policy.
Explore ARF →Transparent encrypted storage using Telegram's infrastructure. Files encrypted on your machine before any network contact. A dead man's switch destroys keys if you stop checking in. What Telegram stores, no one can read.
Explore tgcryptfs →ARF interposes between your AI tools and their APIs, checking every action against declarative policy and sealing each decision into a cryptographic chain. Your team directs; agents execute; ARF attests to every step.
tgcryptfs encrypts before data leaves your machine. The cloud stores opaque blocks it cannot read. Only your keys, on your device, produce files. The blocks stay opaque at rest, in transit, and under compulsion.
When governance, provenance, and access control are handled at the infrastructure layer, your team is freed from the mechanics of compliance and returned to what only people should do.
Process augmentation isn't a one-time gain. Every mechanical task absorbed by governed agents accumulates capacity for the work that defines what your organization becomes.
Professional services for small and medium companies integrating AI into how they build and operate. Strategy, engineering, and the hard parts in between.
Compress design and delivery cycles without compressing quality. Agent-assisted iteration with human taste at every decision gate.
Map where your people spend time on coggery. Replace the mechanical with governed agents. Return your team to the decisions only people can make.
AI-accelerated development of the products and tools your team needs. From greenfield to production, with governance built in.