README.md

Open security scanner and self-hosted control plane for AI, MCP, and cloud infrastructure.

Scan repositories, images, and cloud accounts; centralize evidence across environments; and enforce AI and MCP runtime policy in infrastructure you control.

15 package ecosystems · 16 compliance surfaces · 77 MCP tools · no account required
Quick start · Docs · Live demo

What it is

agent-bom is an open scanner and self-hosted control plane for software, cloud, identity, AI-agent, and MCP evidence. One Finding + UnifiedGraph model powers CLI and CI artifacts, fleet and browser investigations, compliance evidence, and runtime policy.

Use the scanner without an account, or deploy the shared control plane inside your own cloud, VPC, Kubernetes cluster, database, identity, and audit boundary.

Graph provenance remains explicit: collected, inferred, static, and runtime relationships stay distinct, and unavailable evidence is never upgraded to observed.

Control-plane architecture

Who it is for

Role	Start here	Primary outcome
Developers	agent-bom scan .	Find and explain issues before code leaves the workstation
AppSec	agent-bom scan . -f sarif -o findings.sarif	Triage reachable findings and enforce CI gates
Security engineers	pip install 'agent-bom[ui]' && agent-bom serve	Investigate exposure paths, identities, and evidence provenance
Platform / SRE	agent-bom connect aws	Centralize estate inventory, jobs, and runtime controls
GRC / audit	agent-bom report compliance-narrative scan.json	Review control mappings and export evidence with explicit gaps
Leadership / CISO	pip install 'agent-bom[ui]' && agent-bom serve	Review posture, coverage, material risk, and change over time
AI / MCP owners	pip install 'agent-bom[mcp-server]' && agent-bom mcp server	Inventory tools and apply allow, warn, or block decisions

AppSec and GRC remain separate workflows: findings and reachability are not presented as audit certification. See product boundaries.

Product gallery

The gallery uses deterministic sample data, visibly labeled in the UI. It is product-state proof, not customer or advisory evidence.

Overview	Findings

Investigation	Remediation

Cloud and environment lineage	Agent mesh

Capture protocol

Quick start

Run against the repository in your current directory:

pip install agent-bom
agent-bom scan .

The console shows inventory, findings, and reachable impact. Save an artifact with agent-bom scan . -f sarif -o findings.sarif, or follow the first-run guide for exit codes, formats, and CI use.

Try without a repository

Use the curated, explicitly synthetic sample when you only want to inspect the output shape:

agent-bom scan --demo --offline

The sample intentionally contains blocking findings, so exit status 1 is expected.

Self-host

Start the loopback control plane:

pip install 'agent-bom[ui]'
agent-bom serve

For a shared deployment, use the documented Docker or Helm path and configure real identity, TLS, PostgreSQL, encryption, and audit keys before exposing it.

Target	Start here
Docker Compose	Pilot compose
Helm / Kubernetes	helm install agent-bom oci://ghcr.io/msaad00/charts/agent-bom --version 0.98.2
EKS	Terraform module
Snowflake SPCS / Native App	scripts/deploy/install.sh snowflake-native · install guide
Air-gapped	Image bundle guide

Examples target this release candidate; confirm release availability before copying an exact pin. Otherwise, use the latest version shown on PyPI.

Deployment overview · Enterprise configuration · Cloud connections

Advanced integrations and runtime entry points

Need	First action	Artifact or next step
GitHub CI	uses: msaad00/agent-bom@v0.98.2	SARIF, PR summary, and a policy exit code
Cloud evidence	agent-bom connect aws	Stored connection reference; run scans from the control plane
Runtime gateway	agent-bom gateway serve --from-control-plane http://127.0.0.1:8422 --bind 127.0.0.1:8090	Allow, warn, and block audit events
Agent interface	agent-bom mcp server	77 MCP tools, 6 resources, and 8 workflow prompts
Agent distribution	Smithery manifest · Glama · MCP registry · Docker MCP	Registry-specific installation metadata

MCP server mode exposes 77 MCP tools, 6 resources, and 8 workflow prompts, all read-first: discovery and analysis never mutate a scanned target.

The CLI, Docker, API, Helm chart, MCP server, gateway, and SDK are distribution surfaces of the same product. The Snowflake SPCS / Native App lane runs inside the customer's Snowflake account; it is a customer-owned deployment target, not an agent-bom-hosted service. Snowflake and Snowpark also remain connector and runtime integrations for the other deployment profiles.

Every way to install it

Surface	Get it
Python package	pip install agent-bom — PyPI
Container	docker pull agentbom/agent-bom — Docker Hub
Kubernetes	helm install agent-bom oci://ghcr.io/msaad00/charts/agent-bom
GitHub Action	msaad00/agent-bom
MCP server	pip install 'agent-bom[mcp-server]' && agent-bom mcp server
MCP registries	Smithery manifest · Glama · MCP registry · Docker MCP
SDKs	Python · TypeScript · Go

Trust

• Read-only discovery by default; runtime write decisions are separate and explicit.
• Credentials are write-only where stored, encrypted at rest, and never returned by API responses.
• API and control-plane routes are tenant scoped and auth protected outside explicit local mode.
• Missing evidence is shown as unavailable or partial, never converted into a factual zero.
• Public examples and screenshots use deterministic synthetic identifiers only.

Threat model · Release verification · Security policy · MCP security model

Contributing

Start with CONTRIBUTING.md, AGENTS.md, and the open issues.

Apache-2.0 licensed.