(S)AGE Governed Agent Memory

What does this agent do, and when should you use it?

(S)AGE (Sovereign Agent Governed Experience) is a local governed memory and messaging layer for AI agents. Every memory write is signed by an agent, passes a validation lifecycle before entering normal recall, is scoped by domain permissions and classification, and keeps an inspectable audit trail. Governance runs on CometBFT/ABCI consensus: personal mode runs one real validator plus a per-node auto-voter, while multi-validator deployments require a BFT quorum. Storage splits into two roles — BadgerDB is authoritative consensus state, and SQLite (personal) or PostgreSQL + pgvector (cluster) projects memory content and vectors at commit for authorized recall. It exposes MCP tools, a Python SDK (sage-agent-sdk) and a signed REST API, alongside CEREBRUM, a local dashboard centered on a 3D MRI memory brain that manages agent enrollment, Access Groups, federation and updates.

Agents submit memories and tasks through MCP tools (sage_inception, sage_remember, sage_recall, sage_task, sage_inbox, sage_message_claim, sage_message_replies, sage_node_health) or through SDK methods such as propose(), list_memories() and hybrid(). The node authenticates the signature and domain permissions, routes the transaction into the CometBFT memory consensus path, commits to BadgerDB, and projects content plus vectors into SQLite or PostgreSQL. Recall performs semantic retrieval inside the same authorization scope and returns confidence, status and attribution. Coordination covers an exact-agent task backlog, one-way assignment notices, a unified inbox, message claims with idempotent replies, and same-agent runtime handoff fenced by session and revision. The CEREBRUM console at http://127.0.0.1:8080/ui/ surfaces chain health, agents, federation, recall tuning, vault recovery and software updates.

1. Teams running multiple coding agents over long projects that need later sessions to recall decisions and failed approaches
2. Platform teams that must attribute every agent memory to a recorded author and gate admission through review
3. Multi-agent workflows that assign tasks to an exact agent, claim inbound requests and reply idempotently
4. Regulated engineering orgs that need to decay or deprecate memories while retaining audit history
5. Organizations that want bounded experience sharing across separate SAGE chains without granting local membership or write access

How do you install or deploy this agent?

Desktop: download the release for your platform from GitHub Releases (signed and notarized macOS DMG, Windows EXE, Linux tar.gz), open SAGE and complete setup in CEREBRUM.

Build from source (Go 1.26.10+):

git clone https://github.com/l33tdawg/sage.git && cd sage
go build -o sage-gui ./cmd/sage-gui/
./sage-gui setup
./sage-gui serve

Docker:

docker pull ghcr.io/l33tdawg/sage:latest
docker run -d --name sage \
  -p 8080:8080 \
  -v ~/.sage:/root/.sage \
  ghcr.io/l33tdawg/sage:latest

Python SDK:

python -m pip install sage-agent-sdk

How do you use this agent?

Check the node before starting another server:

curl http://127.0.0.1:8080/health

A healthy node returns {"status":"healthy"}. After connecting an MCP client, call sage_inception and follow its returned memory mode. Reading committed memories with the Python SDK requires Python 3.10+, a running node and an enrolled agent:

import os
from sage_sdk import AgentIdentity, SageClient

identity = AgentIdentity.from_file(os.environ["SAGE_IDENTITY_PATH"])
domain = os.environ["SAGE_DOMAIN"]

with SageClient(
    base_url="http://127.0.0.1:8080",
    identity=identity,
    trust_env=False,
) as client:
    page = client.list_memories(
        domain=domain, status="committed", limit=3,
    )
    for memory in page.memories:
        print(memory.memory_id, memory.content)

To record experience, call propose() in a domain the agent can write to and set classification explicitly; use hybrid() for semantic retrieval. Installations predating v5.0 should re-run the installer to add Claude Code hooks:

cd /path/to/your/project
sage-gui mcp install

What are this agent's strengths and limitations?

How does this agent compare with similar options?

Key facts side by side with the most closely related agents.

How does FollowAgents rate this agent?

Recent releases

Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.

1. v11.25.1 SAGE v11.25.1 2026-11-04
   Release notes excerpt

   SAGE v11.25.1 AMID can approve a distinct local agent. Operators can opt into --local-agent-key-dir (SAGE_LOCAL_AGENT_KEY_DIR) to supply identity-bound target consent keys to the existing access-policy route and Python SDK methods. Root authorization, same-machine checks, and consensus validation remain required. Unregistered targets are refused before signing or broadcast, and malformed Ed25519 private-key encodings fail closed. …

   View on GitHub ↗
2. v11.25.0 SAGE v11.25.0 2026-11-03
   Release notes excerpt

   SAGE v11.25.0 includes the merged implementation through #526 and release preparation #527. Fresh startup New personal chains start directly at the supported app29 version. Shared Unix deployments use one signed full-validator genesis, then init-fresh-node --genesis-sha256 <digest> on each prepared node before serve. Existing chains keep their keys, genesis and governed upgrade history. See the [fresh-genesis guide](https://github.com/l33tdawg/sage/blob/v11.25.0/docs/reference/fresh-genesis.md). App30 and compatibility …

   View on GitHub ↗
3. v11.24.5 SAGE v11.24.5 2026-10-31
   Release notes excerpt

   Sovereign Agent Governed Experience — persistent, governed memory for AI agents. What's Changed * Fix task identity across domains by @l33tdawg in https://github.com/l33tdawg/sage/pull/480 * release: prepare SAGE v11.24.5 by @l33tdawg in https://github.com/l33tdawg/sage/pull/483 Full Changelog: https://github.com/l33tdawg/sage/compare/v11.24.4...v11.24.5

   View on GitHub ↗

FAQ