Duvaxa

Security & Compliance · AI agent inventory

AgentLedger

See every AI agent in your code.

A complete inventory of the AI agents that can touch your repositories: CI agents, MCP servers, agent settings and organisation-level AI apps.

Bank and NBFC CISOsIT services firmsPlatform teams
agent-guard
$ agent-guard --help
→ AI agent register
→ Risk findings
→ SARIF report
What the product produces.
FindsAgents, MCP servers, AI apps
ScopeRepositories and org
AccessRead-only
OutputAuditor-ready inventory

The problem

Developers connect AI agents to repositories faster than security can track. Auditors now ask for the list.

What it does

01

Finds AI agents in CI, MCP configs, settings and instruction files

02

Flags settings that let an outsider steer an agent

03

Auditor-ready register in HTML, every finding mapped to the OWASP Top 10 for LLM Applications (2025)

04

Rates toxic combinations: outsiders able to steer an agent that holds secrets or write access

05

Flags remote MCP servers that send your code and context to a third party

06

SARIF output for code-scanning dashboards

07

ZIP upload in the workbench

How it works

  1. Scan repositories or upload a ZIP
  2. Review risky settings
  3. Export the register for audit

Works with

GitHubMCPClaude CodeCopilot

Runs in your CI

GitHub Actions workflow step
- uses: ./.github/actions/agent-guard
with:
path: .
fail-on: high
Licensed customers get the action or command-line tool installed in their own organisation; it runs on your runners, so your code and plans never leave them. The command-line tools run in GitLab CI the same way.

Your data

  • Read-only token.
  • Configuration is read, not code content.
  • Results stay with you.

Who it's for

Bank and NBFC CISOsIT services firmsPlatform teams

Questions

Why now?

Auditors have started asking for the list, and agents are being connected faster than security can track.

Does it block agents?

It inventories and flags risky permissions; MaskWall enforces.

Related

Often used with AgentLedger

See AgentLedger on your own data

A 30-minute call. We run the product on your export, repository or document and walk you through the results.

Book a demo