// reference
Glossary
Definitions of terms used in the Superagent dashboard and documentation.
Agreement
Contributor License Agreement (CLA) legal text stored as a template with versioned hashes. Managed under Agreements.
Contributor trust
Security signals about a GitHub contributor (identity, origin, spray patterns), published as a GitHub check and tracked on the Contributors page. See Contributor Trust.
Finding
A security issue in the org-wide Findings queue. Kinds: repository_red_team, web_app_red_team, and github_advisory. See Findings.
GitHub App
An installation that grants Superagent access to selected repositories. Superagent Security powers scans, trust, and reports; Open CLA powers agreement checks. See Install GitHub Apps.
Context Guardrails
Trust scoring for the external content agents consume. Supported origins are web pages, email, files, agent skills, MCP repositories, and packages. See Context Guardrails.
Runtime Guardrails
Endpoint-local monitoring and enforcement for coding agents. Superagent clients normalize agent activity, evaluate security rules locally, and report findings and enforcement decisions. See Runtime Guardrails.
MCP server
Superagent's REST API exposed as a remote Model Context Protocol server, so coding agents can read findings and start reports. See MCP server.
MCP repository scan
A static Context Guardrails scan of a public GitHub repository that contains an MCP server. The scan evaluates identity, source, tool descriptions, instructions, and schemas without connecting to or executing the MCP server. It requires the Superagent Security GitHub App and fails closed when repository contents cannot be scanned. See MCP repositories.
Package scan
A Context Guardrails scan of a registry package (ecosystem:name@version). It uses the same supply chain pipeline as pull request dependency checks and maps package intelligence into identity, behavior, and content scores. See Packages.
PR scan
The checks Superagent runs on every pull request: Security scan, Superagent Supply Chain Scan, and Contributor trust. See PR Scans.
Report (red team)
An adversarial engagement against one of four targets: repository, Web app, agent, or model. All reports live under Red team (/app/reports). See Red Team.
Repository
A GitHub repository connected through a GitHub App installation, with per-repo scan and CLA settings on its detail page. See Repositories.
Template (agreement)
A versioned CLA document in Agreements, assigned to repositories on the repository detail page.