Manasvi is a policy-governed runtime for building agents with approval-gated actions, auditable execution, tool mediation, sandboxing, and memory provenance.
Manasvi separates conversation, policy, approval, memory, tools, and execution so agents can be useful without becoming blindly autonomous.
Agents that propose, policies that decide, humans that approve. Every capability is mediated — not bolted on after the fact.
Every sensitive action is evaluated against explicit policy before execution. The model proposes; policy decides.
Agents propose actions. Humans or policies approve them. Executors only run approved, cryptographically-signed intents.
Every decision, tool call, approval, denial, and execution result is written to an append-only, integrity-checked trail.
Tool execution is isolated with controlled filesystem, network, secret, and process access. Plugins can't escalate trust.
Memory is separated by provenance and trust level to reduce poisoning risks and prevent unsafe context reuse across sessions.
Telegram, Gmail, Calendar, filesystem, and web tools can all be governed consistently through the same policy engine.
In most agent frameworks, the model outputs a tool call and the system runs it. Manasvi puts a governance layer in between. The model proposes. Policy decides. A signed intent is created. Execution is sandboxed. Everything is recorded.
This means you get real control — not just logging after the fact, but actual gates that can stop, redirect, or require approval for any action.
Free, open source, and runs entirely on your machine. Policy-first execution from the ground up.