Skip to content
SEEPIENT AGENTv0.7.2

The complete runtime for
an autonomous agent species.

Seepient pairs a full-screen terminal UI with a CLI, a TypeScript SDK, and a standalone server — one agent loop, one permission engine, one audit trail across all of them.

Live runtime activity
  • permission granted — execute_shell_command
  • exact commit verified — 4f2c9a1
  • model routed — language · complex tier
  • sandbox entered — seatbelt profile
  • skill activated — release-notes
  • session resumed — 128 messages

Clarity and control
for every action the model takes.

An agent that can touch your system needs more than a prompt. Seepient brokers capability, verifies every write, and records what happened.

[01]

Every action, brokered

Tools do not run until the policy engine says so. Capabilities are granted, escalated, and audited per action — in the terminal or over the wire.

[02]

Isolated to the kernel

Shell and file effects run inside Seatbelt or Bubblewrap profiles, and writes land through a Rust commit helper that verifies pre-images atomically.

[03]

Routed by purpose

Models are picked per purpose and tier — standard, complex, efficient — from a live provider catalog, never a hardcoded list.

Permission broker

  • fs.writesession
  • net.fetchask
  • secret.refwithheld

Nothing runs until policy says so.

Purpose routing

  • language · complexsonnet
  • vision · standardgemini
  • commit · efficienthaiku

Models assigned per purpose and tier.

Audit trail

  • 12:04:11fs.writeallow
  • 12:04:12net.fetchdeny
  • 12:04:15shellallow

Every decision, append-only.

Architecture

One platform.
Multiple strict layers.

Prompts arrive through any surface and cross four one-way layers. Every decision is brokered, every effect is isolated, and everything that matters is kept.

SEPIENT CORE · RUNTIME MAP

Surfaces

Talk to it

Terminal UI (TUI)CLI & REPLServer APIWorker SDK

The Brain

Decides what happens

Agent LoopModel RouterPermission EngineApproval InboxSkillsSecret Shield & Egress Gate

The Hands

Acts on the world

Sandboxed Execution (Seatbelt / Bubblewrap)Rust Exact-Commit BrokerBuilt-in ToolsMCP Gateway

Memory

What it keeps

Session History0600 Audit LogSettingsCredential Vault
FOUNDATIONSshared types · errors · contracts · settings — imported by every layer, importing no one

One-way imports

UI → Transport → Domain → Capabilities → Vendors. Nothing imports upward; no service SDK ever leaves the vendor quarantine.

One policy engine

All four surfaces delegate to the same agent loop and permission broker, so behavior never forks between terminal and server.

Shared foundations

Types, errors, contracts and settings schema live in foundations — imported by every layer, importing no one.

Surfaces

Four ways in. One runtime.

Pick the surface that fits the job. Each one resolves the same config, asks the same policy engine, and writes the same audit trail.

Exact commits

Writes land exactly —
or not at all.

Model file edits are applied by a native commit helper that re-reads the file and compares it against the pre-image the model saw. If anything moved underneath — a symlink, a race, a stale buffer — the write is refused.

How exact commit works →
Why Seepient

Built for production agents.
Not demo loops.

The gap between a chat completion and an agent you can leave unattended is everything around the model call. Seepient ships that everything.

CapabilitySeepientTypical DIY loop
Permission policy & consent modes Built-inBolt-on, if any
OS-level process sandbox Seatbelt · Bubblewrap
Exact atomic file commits Native helperBest-effort writes
Durable, replayable audit trail Append-onlyConsole.log
Purpose × tier model routing Live catalogHardcoded names
Embedder-owned storage (SDK) Store contracts
Ship agents you can trust

Take control of your agents.
Operate at a new level

Released under the Business Source License 1.1.