AgentGuard
Local spend caps, kill-switch blocking and Ed25519-signed receipts for AI agent tool calls.
Open the source (opens github.com) https://github.com/merchantguard/agentguard-mcp
Before you use it
- Runs on your computer with your permissions. Read what it can do before connecting it to an agent.
- Needs credentials: AGENTGUARD_LICENSE_KEY. Use a key with the least access you can.
What it is
MCP servers from io.github.MerchantGuard.
The problem it solves and who it suits haven't been reviewed by InnerVault. The description above is the source's own.
How to get it
npm package (stdio)
@agentguard-run/mcp@0.3.1
From the server's entry in the official MCP Registry
Settings it asks for
AGENTGUARD_MCP_DAILY_CAP_CENTS· Local daily spend cap in cents. A nonnegative integer.AGENTGUARD_MCP_PER_CALL_CAP_CENTS· Local per-call spend cap in cents. A nonnegative integer.AGENTGUARD_MCP_TENANT· Tenant identifier stamped into local receipts.AGENTGUARD_HOME· Local signing-key, configuration and ledger directory. Defaults to .agentguard in the operating system home directory.AGENTGUARD_MCP_LEDGER· Local decision storage. ndjson persists receipts across restarts; memory retains this process only.AGENTGUARD_MCP_DISABLE_COST_OVERRIDE· Set to 1 to discard and prevent model pricing overrides for this server.AGENTGUARD_LOCK_COST_OVERRIDES· Spend SDK startup lock. Set to 1 to prevent model pricing overrides.AGENTGUARD_ACTOR_DIGEST· Set to 1 to hash actor identifiers before they enter local receipts.AGENTGUARD_LICENSE_KEYsecret · Optional AgentGuard license key for license validation and seat registration. Otherwise read from local configuration.AGENTGUARD_LICENSE_ENDPOINT· License and seat service base URL. No prompts, tool content or receipts are sent.AGENTGUARD_NO_BEACON· Set to 1 to disable optional activation telemetry regardless of saved consent.AGENTGUARD_TELEMETRY· Set to 1 to opt in to content-free activation telemetry. Otherwise saved consent applies, default off.
Related
-
agentguard
MCP serversAgentGuard — 20-tool AI safety MCP: policy preflight, risk scoring, audit logging, rate limits.
-
agentguard
MCP serversNetwork-egress firewall for agent tools: check URLs against a declarative policy.
@mukundakatta/agentguard-mcp@0.1.0 -
agentguard47
MCP serversRead-only MCP server for coding-agent traces, alerts, costs, usage, and budget health.
@agentguard47/mcp-server@0.2.2 -
AgentGuard by Audiscape
MCP serversStops runaway AI agent loops and token burn. Checks each tool call before it runs. No signup.
-
agentguard
MCP serversMCP policy proxy: spend caps, approvals for destructive tools, kill switch, dry-run, audit log.
@agentwares/agentguard@0.1.9 -
AgentGuard — security checks for AI agents
MCP serversSecret, CVE and dependency-vulnerability scanning for AI agents (free, OSV.dev).
ghcr.io/shuaicongxiaomai/agentguard:0.1.2