actionlog
# actionlog
[](https://www.npmjs.com/package/actionlog)
[](https://opensource.org/licenses/MIT)
**Tamper-Evident Autonomous Agent Action Audit & Governance Logging** MCP Server. Designed for AI agent compliance, recording irreversible audit trails and verifying tool execution integrity.
## Features
- **Cryptographic Audit Chain**: Every agent action, tool invocation, and API execution is hashed into a tamper-evident sequential log.
- **Agent Governance & Compliance**: Built for enterprise AI requirements (SOC2, EU AI Act traceability, agent liability auditing).
- **Model Context Protocol (MCP)**: Native integration for Claude Desktop, Cursor, and multi-agent orchestrators.
## Quick Start
### Direct Execution
```bash
npx actionlog
```
### Claude Desktop Integration
Add to your `claude_desktop_config.json`:
```json
{
"mcpServers": {
"actionlog": {
"command": "npx",
"args": ["-y", "actionlog"]
}
}
}
```
## Tools Included
1. `record_agent_action`: Commits an agent action into the cryptographic hash chain.
2. `verify_log_integrity`: Validates historical log entries against checksum tampering.
## License
MIT © [tudadada](https://github.com/tudadada)
TDQS
Scored across 2 tools
The two tools have completely distinct purposes: one appends an action to the log, the other verifies the log's integrity. There is no overlap or ambiguity in their intended usage.
Both tool names follow the same verb_noun pattern (record_agent_action, verify_log_integrity), are in snake_case, and clearly communicate the action and target. The naming is fully consistent.
With only two tools, the server feels thin even for a focused audit-logging purpose. While the count is not excessive, it falls at the low end where a single extra tool for retrieval would make the set feel more complete.
The server covers the core record and verify capabilities, but there is no way to read, list, or query the logged actions. This creates a notable gap: an agent can write and validate entries but cannot retrieve them for normal use.