SpecterOps has released Blacklight, an open-source toolkit designed to identify and detect local artifacts from AI coding agents such as Codex, Claude Code, Cursor, and Antigravity CLI. As AI becomes more prevalent in writing code, troubleshooting issues, executing commands, inspecting repositories, and interacting with cloud resources, their local files may contain valuable but potentially exploitable data. While these tools enhance productivity, they also introduce a new security concern: the potential exposure of sensitive information to attackers who gain access to workstations.
Configuration files may also expose additional risks by revealing information about user models, trusted projects, command approval settings, sandbox rules, environment variables, and MCP server configurations.
Chat history from AI agents reveals what users are working on, internal file paths, repository names, debugging output, deployment instructions, internal URLs, hostnames, and commands. Blacklight Scout enables endpoint discovery across Windows, macOS, and Linux systems. Its stealthy loaders perform filesystem triage and report discovered agents, valuable paths, file sizes, and recency without accessing artifact contents.
The SpecterOps research underscores a pressing need for enterprise security teams: treating local AI agents as equivalent to browser profiles, cloud CLI configurations, and other valuable endpoints data stores is essential. Organizations should inventory AI agent usage, restrict access to local profile directories, monitor reads of credential files, review trusted-project settings, and establish retention policies for sessions.
