MCP Server · Model Context Protocol

Give your AI agent
terminal memory.

11 tools and 5 resources that let Claude Code, Cursor, and other AI agents query your shell history directly. 100% local. Zero config.

$ suv init claude-code
Copied!

or suv init cursor for Cursor

Your AI agent is flying blind.

Every time your AI coding agent runs a shell command, it forgets the result. It can't see what failed five minutes ago. It can't check if a command is risky before running it. It has no memory of your terminal history.

MCP (Model Context Protocol) changes that. It's the open standard that lets AI agents access structured data sources. Suvadu is an MCP server that turns your shell history into a queryable knowledge base your agent can access in real time.

No cloud. No API keys. No config files to edit. Just run one command and your agent gets terminal memory.

11 tools your agent can call.

Every tool reads from your local SQLite database. No network, no latency, no permissions to configure.

Discover

search_commands

Search history by text, directory, executor, and date range

recent_commands

What just happened in this directory

command_status

Has this command been run before? What happened?

Analyze

get_stats

Command counts, success rates, activity patterns

what_changed

File-modifying operations: writes, deletes, moves

what_failed

Failed commands and the prompt that caused them

Predict

suggest_next

Frecency-based prediction of your next command

assess_risk

Pre-execution safety check with risk classification

Context

session_history

Full chronological history of a session

list_sessions

Browse sessions with metadata

get_prompts

AI agent prompts and the commands they triggered

Context before the first prompt.

5 resources are automatically injected into your agent's context at session start. No tool calls needed.

Most MCP servers wait for the agent to call a tool. Suvadu goes further — it pre-loads context so your agent already knows what's happening in your terminal before you even ask.

MCP Resources
suvadu://history/recent Last 20 commands with exit codes
suvadu://failures/recent Recent failures grouped by prompt
suvadu://stats/today Today's command count & success rate
suvadu://risk/summary Risk summary of recent agent commands
suvadu://agents/activity Per-agent activity breakdown

Set up in one command.

No config files. No API keys. No network ports.

1

Install Suvadu

curl -fsSL https://downloads.appachi.tech/suvadu/install.sh | sh
2

Connect your agent

suv init claude-code

or suv init cursor

3

Done. Agent has memory.

Start coding. Your AI agent can now search history, check failures, and assess risk.

Works with every major AI coding agent.

One-command MCP setup for Claude Code and Cursor. Command tracking for everything else.

MCP auto-configured

Claude Code logo
Claude Code
suv init claude-code
Cursor logo
Cursor
suv init cursor

Commands tracked automatically

OpenCode logo
OpenCode
auto-detect
VS Code logo
VS Code
auto-detect
Windsurf logo
Windsurf
auto-detect
Codex logo
Codex
auto-detect
Antigravity logo
Antigravity
auto-detect

MCP can be manually configured for any agent that supports the Model Context Protocol.

What it looks like in practice.

Real queries your AI agent can answer with Suvadu's MCP tools.

What failed in this project in the last hour?

called what_failed

3 commands failed: npm test (exit 1, 23 min ago), cargo build (exit 101, 45 min ago), docker compose up (exit 137, 58 min ago). The npm test failure was triggered by your prompt "fix the auth middleware".

Is it safe to run git push --force origin main?

called assess_risk

Risk level: Critical. This command has been run 0 times in this project. Force-pushing to main is destructive and irreversible. I recommend git push --force-with-lease instead.

100% local. By design.

Your shell history never leaves your machine.

No network ports

JSON-RPC over stdin/stdout only. Never opens a TCP or UDP port.

No cloud

All data stays in your local SQLite database. No accounts. No telemetry.

Same data as CLI

Reads from the same database that powers suv search. No separate data store.

Secret redaction

API keys, tokens, and passwords are auto-redacted before they touch disk.

Ready to give your agent memory?

Install Suvadu, run one init command, and your AI agent gets terminal context forever. For the complete reference, see the MCP Server documentation.

Get started in 30 seconds

Three commands. That's it.

zsh
# Install (macOS & Linux — Zsh / Bash)
$ curl -fsSL https://downloads.appachi.tech/suvadu/install.sh | sh
# Add shell hook
$ echo 'eval "$(suv init zsh)"' >> ~/.zshrc
# Activate & verify
$ source ~/.zshrc && suv status

Also available via brew tap AppachiTech/suvadu && brew install suvadu, cargo install suvadu, and manual install.