agent-gate — fail-closed quality gate + hash-chained receipt ledger for agent work (stdio via PyPI)
@Jott2121
About agent-gate — fail-closed quality gate + hash-chained receipt ledger for agent work (stdio via PyPI)
MCP server that adds a fail-closed quality gate and hash-chained receipt ledger to any AI agent workflow.
Basic information
Config
Add this server to your MCP-compatible client using the configuration below.
{
"mcpServers": {
"agent-gate": {
"command": "python",
"args": [
"-m",
"agent_gate.server"
]
}
}
}Tools
No tools detected
We auto-extract tools from the README. The maintainer can list them under a ## Tools heading to populate this section.
Overview
What is agent-gate?
agent-gate is an MCP server that lets an AI agent gate its own work before claiming completion. It enforces deterministic checks, an independent refute-first review, and records tamper‑evident, hash‑chained receipts. It is designed for developers operating AI agent systems who need formal quality assurance rather than self‑grading.
How to use agent-gate?
Install with pip install mcp-agent-gate. Add it to an MCP client configuration (e.g., Claude Desktop or Claude Code) using the command python -m agent_gate.server. The agent then calls verify_gate() before reporting done, and receipts are stored at ~/.agent-gate/receipts.jsonl (overridable with AGENT_GATE_LEDGER). It can also be used directly via Python imports.
Key features of agent-gate
- Fail‑closed checklist: a check passes only if explicitly true.
- Tamper‑evident, hash‑chained receipt ledger.
- Human‑gated by default for irreversible actions.
- Pure stdlib core (gate and ledger modules).
- MCP adapter for easy integration with AI clients.
Use cases of agent-gate
- Agent must satisfy a ‘ship’ gate before delivering output.
- Record and later verify an append‑only decision log.
- Hard‑enforce independent review and no‑secrets policies.
- Detect tampering by past receipt edits via
verify_chain().
FAQ from agent-gate
What does fail‑closed mean?
A check counts as satisfied only if it is explicitly true. Missing proof is treated as not passing, mirroring a promotion gate.
How does the receipt ledger work?
Every receipt is appended as a (decision, metric, value, verdict) entry linked by a SHA‑256 hash chain. Calling verify_chain() returns false if any past receipt is edited or deleted.
What are the runtime dependencies?
The core agent_gate/gate.py and agent_gate/ledger.py use only the Python standard library. The MCP server adds one runtime dependency: the mcp package.
Can I use agent-gate without an MCP client?
Yes. Import the gate and ledger classes directly from Python. See the demo in the README for a complete example.
What checks are in the default ‘ship’ gate?
The default gate requires: deterministic checks pass, independent refute review, no secrets exposed, human approval if the action is irreversible, and an honest receipt logged.
More AI & Agents MCP servers
Open Multi-Agent Canvas
CopilotKitThe open-source multi-agent chat interface that lets you manage multiple agents in one dynamic conversation and add MCP servers for deep research
Just Prompt - A lightweight MCP server for LLM providers
dislerjust-prompt is an MCP server that provides a unified interface to top LLM providers (OpenAI, Anthropic, Google Gemini, Groq, DeepSeek, and Ollama)
Sequential Thinking Multi-Agent System (MAS)
FradSerAn advanced sequential thinking process using a Multi-Agent System (MAS) built with the Agno framework and served via MCP.
🔎 GPT Researcher
assafelovicAn autonomous agent that conducts deep research on any data using any LLM providers
Model Context Protocol for Unreal Engine
chongdashuEnable AI assistant clients like Cursor, Windsurf and Claude Desktop to control Unreal Engine through natural language using the Model Context Protocol (MCP).
Comments