agent-bom

Name: agent-bom
Rating: 1.4 (29 reviews)
Author: msaad00

AI 与智能体

by msaad00

面向 AI 供应链安全的扫描器，支持 CVE 扫描、blast radius 分析、策略执行与 SBOM 生成。

什么是 agent-bom？

面向 AI 供应链安全的扫描器，支持 CVE 扫描、blast radius 分析、策略执行与 SBOM 生成。

README

<picture> <source media="(prefers-color-scheme: dark)" srcset="https://raw.githubusercontent.com/msaad00/agent-bom/main/docs/images/logo-dark.svg"> <img src="https://raw.githubusercontent.com/msaad00/agent-bom/main/docs/images/logo-light.svg" alt="agent-bom" width="360" /> </picture> <a href="https://github.com/msaad00/agent-bom/actions/workflows/ci.yml"><img src="https://img.shields.io/github/actions/workflow/status/msaad00/agent-bom/ci.yml?branch=main&style=flat&label=Build" alt="Build"></a> <a href="https://pypi.org/project/agent-bom/"><img src="https://img.shields.io/pypi/v/agent-bom?style=flat&label=Latest%20version&cacheSeconds=60" alt="PyPI"></a> <a href="https://hub.docker.com/r/agentbom/agent-bom"><img src="https://img.shields.io/docker/pulls/agentbom/agent-bom?style=flat&label=Docker%20pulls" alt="Docker"></a> <a href="https://github.com/msaad00/agent-bom/blob/main/LICENSE"><img src="https://img.shields.io/badge/License-Apache%202.0-blue?style=flat" alt="License"></a> <a href="https://securityscorecards.dev/viewer/?uri=github.com/msaad00/agent-bom"><img src="https://img.shields.io/ossf-scorecard/github.com/msaad00/agent-bom?style=flat&label=OpenSSF%20scorecard" alt="OpenSSF Scorecard"></a> <a href="https://glama.ai/mcp/servers/msaad00/agent-bom"><img src="https://img.shields.io/badge/MCP-Glama-7c3aed?style=flat" alt="agent-bom on Glama"></a> <a href="https://smithery.ai/servers/agent-bom/agent-bom"><img src="https://img.shields.io/badge/MCP-Smithery-1f6feb?style=flat" alt="agent-bom on Smithery"></a>  Open security scanner and self-hosted control plane for AI, MCP, and cloud infrastructure. Scan from CLI, CI, Docker, or cloud connect on a control plane you run → centralize evidence → enforce runtime MCP/tool calls. One Finding + UnifiedGraph model across CLI, API, UI, and MCP. Brand lockup above is the product logo. The CLI prints the wordmark as text (<code>agent-bom</code>); the UI and <a href="https://demo.agent-bom.com">live demo</a> show the logo in app nav. <a href="https://demo.agent-bom.com">Live demo</a> · <a href="https://msaad00.github.io/agent-bom/">Docs</a> · <a href="#quick-start">Quick start</a> · <a href="#deploy--self-host">Deploy</a> · <a href="docs/FIRST_RUN.md">First run</a> · <a href="https://github.com/marketplace/actions/agent-bom">GitHub Action</a> · <a href="https://hub.docker.com/r/agentbom/agent-bom">Docker</a> · <a href="https://github.com/msaad00/agent-bom/releases">Changelog</a>

Quick start

Three commands to see value locally (no cloud account required):

bash

pip install agent-bom
agent-bom scan --demo --offline
# optional control plane (loopback UI + API)
pip install 'agent-bom[ui]' && agent-bom serve

Console report prints inventory, findings, and blast radius inline.
Optional file artifacts: -f html -o agent-bom-report.html (also json, sarif, SBOM).
Demo exit status 1 is expected — the sample includes blocking findings (same gate as CI).

Full walkthrough: docs/FIRST_RUN.md.

Who it is for

Team	Start here	Outcome
Developers	`agent-bom scan .`	Inventory, findings, blast radius before changes ship
AppSec	`agent-bom scan . -f sarif -o findings.sarif`	Reachability triage, graph paths, CI gates
Platform / SRE	`pip install 'agent-bom[ui]' && agent-bom serve` or Helm	Customer-controlled API/UI, fleet evidence, runtime policy
GRC / audit	`agent-bom scan . -f json -o scan.json && agent-bom report compliance-narrative scan.json`	Framework mappings, signed bundles, review context
AI / MCP owners	`pip install 'agent-bom[mcp-server]' && agent-bom mcp server` or `gateway serve`	Tool inventory and allow/warn/block decisions

Evidence helper, not a GRC system of record, IAM, SIEM, or certification program. Boundaries: PRODUCT_BOUNDARIES.md.

How the tool works

Three product lanes on one Finding + UnifiedGraph model:

Scan — CLI / CI / Docker / cloud connect → inventory, findings, SARIF/SBOM/HTML, graph
Control plane — pip install 'agent-bom[ui]' && agent-bom serve → tenant UI/API, attack paths, compliance, audit (self-host with Docker or Helm + Postgres). Loopback is the default; non-loopback hosts need real auth or an explicit --allow-insecure-no-auth (env vars alone are not enough).
Runtime — agent-bom gateway serve --from-control-plane http://127.0.0.1:8422 --bind 127.0.0.1:8090 → allow/warn/block on live MCP/tool calls

Discovery and static/cloud scanning are read-only. The control plane stores evidence; gateway/proxy modes make explicit policy decisions at a separate boundary.

One finding fans out to the MCP servers that load it, reachable tools, credential references, and agents that can reach it — not a CVE list in isolation.

See the product

Story order: Scan → Connect → Posture → Investigate → Enforce (findings + remediation together under Scan). Screenshots are the same UI as the live demo (logo in app nav). Capture notes: docs/CAPTURE.md.

Scan — findings and remediation

Findings queue	Remediation
<img src="https://raw.githubusercontent.com/msaad00/agent-bom/main/docs/images/dependency-map-live.png" alt="Findings queue with severity, reachable agents, fixes, and review actions" width="430" />	<img src="https://raw.githubusercontent.com/msaad00/agent-bom/main/docs/images/remediation-live.png" alt="Prioritized remediation with risk reduction, ownership, and verification" width="430" />

Connect — sources and new scan

Connections	New scan
<img src="https://raw.githubusercontent.com/msaad00/agent-bom/main/docs/images/cloud-accounts-live.png" alt="Connections hub across cloud, code, AI, and data sources" width="430" />	<img src="https://raw.githubusercontent.com/msaad00/agent-bom/main/docs/images/new-scan-live.png" alt="New Scan workspace with collector plan and read-only boundary" width="430" />

Posture — risk overview

Investigate — lineage, mesh, context

Three graph lenses tell different stories: package-level lineage, multi-agent mesh overlap, and lateral context — not three copies of the same CVE chain.

Lineage (package hop)	Agent mesh (shared MCP)
<img src="https://raw.githubusercontent.com/msaad00/agent-bom/main/docs/images/lineage-graph-live.png" alt="Lineage graph focused on developer-copilot, GitHub MCP, next package version, and DEMO-VULN finding with labeled edges" width="430" />	<img src="https://raw.githubusercontent.com/msaad00/agent-bom/main/docs/images/mesh-live.png" alt="Agent mesh showing developer-copilot and sre-runbook-agent converging on shared filesystem MCP with relationship labels" width="430" />
Investigation drilldown: agent → MCP → `next@` → critical finding.	Two agents, one shared server — tools and credentials in the same frame.

<img src="https://raw.githubusercontent.com/msaad00/agent-bom/main/docs/images/context-map-live.png" alt="Context map with path focus off: repo-write tool, GitHub credential, MCP servers, and agent lateral links" width="820" /> Context map: neighborhood topology (tools, creds, servers) — not the same hero CVE strip as lineage. <details> <summary>Attack path — prioritized path with export/handoff chrome</summary> <img src="https://raw.githubusercontent.com/msaad00/agent-bom/main/docs/images/security-graph-live.png" alt="Prioritized attack path connecting identity, agent, MCP server, package, and critical finding" width="820" />

Path view is a single-row hop strip (scroll horizontally on long chains).

</details>

Enforce — runtime gateway

<img src="https://raw.githubusercontent.com/msaad00/agent-bom/main/docs/images/gateway-policies-live.png" alt="Runtime gateway KPI rollup and tool-call feed" width="820" /> <details> <summary>CLI walkthrough — 0.98.0 console demo</summary> <img src="https://raw.githubusercontent.com/msaad00/agent-bom/main/docs/images/demo-latest.gif" alt="agent-bom terminal demo showing inventory, findings, remediation, and package gate" width="820" />

In the terminal the brand is the text wordmark (agent-bom), not the SVG logo. Seeded requests typosquat produces an expected non-zero security-gate exit — a demonstrated finding, not a failed recording.

</details>

Deploy & self-host

You run the control plane in your own boundary (no managed public SaaS in this repo yet):

bash

curl -fsSL https://raw.githubusercontent.com/msaad00/agent-bom/main/deploy/docker-compose.pilot.yml -o docker-compose.pilot.yml
docker compose -f docker-compose.pilot.yml up -d
# Dashboard -> http://localhost:3000

Pilot compose binds 127.0.0.1 with loopback CORS. Before sharing a link, use docker-compose.platform.yml or docs/HOSTED_POC.md.

Target	Start here
Docker Compose	pilot compose
Helm / Kubernetes	chart — `helm install agent-bom oci://ghcr.io/msaad00/charts/agent-bom --version 0.98.0`
EKS	Terraform module
CloudFormation	templates
Snowflake SPCS	install guide

Guides: Deploy anywhere · deployment overview · HOSTED_POC.

<details> <summary>Surfaces and entry points</summary>

Need	First action	Artifact
Scan repo / image / agent config	`agent-bom scan .` or GitHub Action	JSON, SARIF, SBOM, HTML
Cloud / data estate	`agent-bom connect aws` then `agent-bom cloud scan`	assets, CIS findings, graph edges
Team posture UI	`pip install 'agent-bom[ui]' && agent-bom serve`	findings, graph, audit, compliance
MCP tools for agents	`pip install 'agent-bom[mcp-server]' && agent-bom mcp server`	strict MCP tool responses
Skills playbooks	docs/skills/ · OpenClaw / Cortex wrappers under `integrations/`	first command → findings / SBOM
Runtime tool governance	`agent-bom gateway serve --from-control-plane http://127.0.0.1:8422 --bind 127.0.0.1:8090`	allow/warn/block audit
Audit package	`agent-bom scan . -f sarif -o findings.sarif`	SARIF, CycloneDX, SPDX, bundles

Also: agent-bom graph, agent-bom remediate -p ., CI pin uses: msaad00/agent-bom@v0.98.0. Maps: CLI · start here · product map.

</details> <details> <summary>Auth, connectors, and architecture</summary>

Control plane: connect once; later jobs use the stored connection reference. Humans: OAuth / OIDC / SAML (+ Snowflake OAuth PKCE) and SCIM. Agents/CI: scoped API keys. Secrets are write-only (encrypted at rest, never read back). Non-loopback deploys terminate TLS at the edge; API/UI stay on loopback or a private network.

Cloud connectors (opt-in, default-off, no secret values read):

Cloud	Enable	Scan
AWS	`AGENT_BOM_AWS_INVENTORY=1`	`agent-bom cloud aws`
Azure	`AGENT_BOM_AZURE_INVENTORY=1`	`agent-bom cloud azure`
GCP	`AGENT_BOM_GCP_INVENTORY=1`	`agent-bom cloud gcp`
Snowflake	SSO or key-pair	`pip install 'agent-bom[snowflake]'` then `agent-bom scan --snowflake`

CLOUD_CONNECT.md · DATA_SOURCES.md · ENTERPRISE.md.

Accuracy: match tiers distro_confirmed > osv_range > osv_ecosystem > unfixed_distro > nvd_cpe_candidate. End users do not need an NVD API key; NVD_API_KEY is an optional operator freshness knob. MCP server mode exposes 77 MCP tools, 6 resources, and 8 workflow prompts over strict arguments. Agent distribution includes a committed Smithery manifest; external catalog liveness is verified separately. Deep dive: ARCHITECTURE.md · vulnerability matching.

</details>

Trust

Read-only discovery by default; no mandatory telemetry
Credential values redacted; env names kept for explainable exposure paths
Exports: JSON, SARIF, CycloneDX, SPDX, Parquet, CSV, Markdown, HTML, PDF, compliance bundles
Tenant scope, auth boundaries, and audit evidence on API/runtime paths

Threat model · Pentest readiness · Python client · Go client · Release verification · MCP security model

Contributing

Start with CONTRIBUTING.md, AGENTS.md, and the open issues. Community chat: Discord.

License: Apache-2.0.

常见问题

agent-bom 是什么？

面向 AI 供应链安全的扫描器，支持 CVE 扫描、blast radius 分析、策略执行与 SBOM 生成。

agent-bom

什么是 agent-bom？

README

Quick start

Who it is for

How the tool works

See the product

Scan — findings and remediation

Connect — sources and new scan

Posture — risk overview

Investigate — lineage, mesh, context

Enforce — runtime gateway

Deploy & self-host

Trust

Contributing

常见问题

agent-bom 是什么？

相关 Skills

Claude接口

RAG架构师

多智能体架构

相关 MCP Server

知识图谱记忆

顺序思维

Codebase Memory

评论