Archipelo vs Endor Labs

About Endor Labs

Endor Labs sells AURI, a security intelligence layer for AI coding agents. AURI plugs into the coding agent harness to block malicious packages before install, verify code is secure, fix vulnerabilities, and flag secrets before they are leaked. AURI inventories every coding agent, model, MCP server, and skill. AURI learns architecture from context files such as AGENTS.md and CLAUDE.md, design docs, and custom prompts. Endor Labs replaces disconnected scanners, triage queues, and remediation handoffs with one continuously managed workflow.^[3][4]

Endor Labs describes a problem in which coding agents require privileged access but can expose sensitive data, credentials, or run unsafe commands. Endor Labs states that coding agents are already writing code, installing dependencies, and shipping changes across an organization, and that most security teams cannot yet answer for any of it. Endor Labs also states that every model, skill, MCP server, and tool call extends the supply chain.^[4]

Endor Labs anchors every finding to its code context graph, a continuously updated model of application behavior across code, dependencies, secrets, and containers. AURI can orchestrate its own agents, with their own sub-agents, that navigate the code context graph to detect, triage, validate, and remediate vulnerabilities. AURI by Endor Labs combines LLM-based reasoning with program analysis trained on ground-truth vulnerability data. AURI maps how data flows through source code, through direct and transitive dependencies, and into container base images. AURI security agents are catching emerging malware campaigns across npm, PyPI, Maven, and other package ecosystems in real time. AURI is the engine behind the AI SAST of Endor Labs.^[3][5]

Developer, architect, and security engineer agents review every PR for architectural risk. Pre-built workflows triage findings and remediate dependencies using codebase context from AURI. The pre-built workflows run on the customer's infrastructure, read-only by default, with every change approval-gated. Every agent action is checked against policy before it runs, with the outcomes allow, block, or ask a human, and every action is recorded. AI Coding Agent Governance shows which models, MCP servers, and tools are active in a development environment and whether their behavior matches policies.^[4][5]

Only vulnerabilities that application code can reach are added to the backlog. Endor Labs analyzes call graphs across first-party code, direct and transitive dependencies, and container layers to determine whether an application can actually invoke a vulnerable function. Container Reachability by Endor Labs extends function-level reachability through the runtime and OS layers, classifying every package as Reachable, Potentially Reachable, or Unreachable, application by application. Reachability from code to image needs nothing to be installed in the cluster. Upgrade impact analysis shows which breaking changes a patch will trigger before anyone touches a version number. Endor Patches fix vulnerabilities in hard-to-upgrade libraries without forcing a major version bump. SBOM generation, artifact signing, and CI/CD posture management are part of coverage from dependency selection through delivery.^[1][4][5][6]

Package Firewall was released in May to help organizations stop software supply chain attacks by preventing installs of malicious packages on developer laptops or in CI/CD pipelines. Package Firewall scans newly published packages across npm, PyPI, and other ecosystems and flags emerging malware campaigns in under 10 minutes. Package Firewall blocks both developers and coding agents from pulling malicious packages into a local machine or CI runner. Package Firewall enforces a cooldown policy so that package versions need to be a sufficient age to be installed. Package Firewall cross-checks package versions against a malware database in case a malicious version is available on the registry beyond the cooldown period. Malicious packages are blocked at install, with one control point for every dependency entering the organization from an agent or a developer.^[1][4][5][7]

Endor Labs launched AURI alongside a partnership with OpenHands. AURI integrates through a plugin that bundles dozens of security skills into a full-stack security review spanning code, secrets, dependencies, and container images. AURI is available to everyone, and its Skills plugin, MCP, and CLI are free for developers. Developers can use these tools to detect vulnerabilities and exposed secrets in code and open source dependencies, block malware attacks, and fix security bugs.^[1][2][3][8]