strata
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@stratasearch memory for the checksum of the last dataset"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Strata
Portable agent memory. Append-only records, verified provenance, MCP tools, and self-hosting.
Strata gives a team one durable memory that any agent can read and update without mixing projects, leaking secrets, or rewriting history. It is built for work that moves across people, sessions, machines, and runtimes.
Why it exists
Local agent memory breaks when work moves across machines, people, tools, or old sessions. Strata keeps the facts that should outlive one agent run:
experiment results and decisions;
dataset notes, source checksums, and extraction status;
procedures that worked and mistakes to avoid;
project handoffs for agents joining later.
Strata does not store secrets, raw transcripts, chain-of-thought, API keys, or provider settings.
Related MCP server: memhippo
What ships
Part | What it does |
| CLI, append-only store, REST gateway, MCP bridge, policy checks, export, and restore. |
| Shareable agent skill for reading, writing, and safe service control without hardcoded host values. |
| Rules and JSON contracts agents must follow before writes. |
| Local, Docker, and systemd helpers for running the service. |
Quick start
git clone https://github.com/rx6ru/strata.git
cd strata
python -m venv .venv
. .venv/bin/activate
python -m pip install -e .
strata --helpStart a local gateway:
strata serve --root /tmp/strata-demo --host 127.0.0.1 --port 8787Check it from another shell:
curl -fsS http://127.0.0.1:8787/healthz
curl -fsS -H 'X-Strata-Client: 1' http://127.0.0.1:8787/v1/statusIf you are running from a source checkout before installing the package:
PYTHONPATH=. python -m strata --help
PYTHONPATH=. python -m strata root-hash --root /tmp/strata-demoRead the full setup guide in docs/getting-started.md.
How agents use it
Copy
skills/strata/into the agent runtime skill directory.Give the agent only non-secret runtime values, such as the gateway URL, namespace, current task, SSH alias, and optional lifecycle wrapper path.
The agent fetches policy, prepares scoped context, reads canonical records, and writes only verified durable knowledge.
Corrections append superseding records. Reorganization appends relocation, split, merge, or invalidation events. Ordinary agents never edit history in place.
The skill does not hardcode hostnames, IP addresses, EC2 instance IDs, usernames, API keys, project namespaces, policy hashes, roots, or record IDs. See docs/agent-skill.md.
How writes stay clean
Strata checks every write before commit:
current policy hash is acknowledged before commit;
schemas validate the record;
idempotency keys prevent duplicate retries;
append-only SQLite triggers block in-place history edits;
content hashes and root manifests make exports checkable;
secret and raw transcript scanners reject unsafe payloads;
derived recall is labeled as navigation, not truth.
Self-hosting
For the Docker stack:
./scripts/configure.sh
./scripts/control.sh startFor a system service, use the installation guide and the constrained lifecycle wrapper in deploy/. Remote agents should call only that wrapper for status, start, stop, or restart.
The default deployment has no app-level users. Keep it on loopback, a private network, SSH, VPN, firewall, or your own proxy/auth layer. Read docs/security-model.md and SECURITY.md before team hosting.
REST and MCP
The REST gateway exposes health, policy, status, search, recall, record, append, supersede, relocate, split, merge, and invalidate endpoints under /v1.
The MCP bridge exposes matching tools such as memory_policy, memory_prepare_context, memory_search, memory_get, memory_capture_verified, and memory_status.
Evaluation
PYTHONPATH=. python -m strata eval --input tests/fixtures/strata_eval/cases.jsonThe fixture checks namespace isolation, provenance, promotion and rejection behavior, derived recall labels, and forbidden-value leakage.
License
MIT. See LICENSE. Third-party notes are in THIRD_PARTY_NOTICES.md.
This server cannot be installed
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Related MCP Servers
- AlicenseAqualityAmaintenanceProvides persistent, searchable memory for AI agents, enabling them to retain, recall, and reflect on information across conversations.191MIT
- AlicenseNot gradedqualityAmaintenanceProvides persistent, shared memory for AI agents by capturing conversations verbatim, distilling facts and summaries, and enabling retrieval through search, timeline, details, and explicit remember tools.MIT

Veracium MCP Serverofficial
AlicenseNot gradedqualityBmaintenanceProvides agents with durable, provenance-aware memory through tools for remembering, recalling, answering, and maintaining information, while structurally resisting injection and confabulation.8MIT- FlicenseNot gradedqualityCmaintenanceProvides AI agents with persistent memory across sessions, enabling recall of decisions, clients, and deadlines with verifiable citations.
Related MCP Connectors
Shared, governed long-term memory for AI agents across tools and sessions via MCP and REST.
Persistent memory for AI agents. Search, store, and recall across sessions.
Persistent memory for AI agents — verbatim conversations, searchable by meaning.
Latest Blog Posts
- Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)By Om-Shree-0709 on .mcpAgent IdentityOAuth 2.1
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/rx6ru/strata'
If you have feedback or need assistance with the MCP directory API, please join our Discord server