Rafter is a security platform that integrates directly with AI coding assistants to scan for secrets, vulnerable dependencies, and unsafe commands as code is written. It offers real‑time local enforcement plus an optional remote engine for full SAST, SCA, and data‑flow analysis, delivering deterministic JSON results for seamless CI/CD integration.
Funding
Funding not disclosed
Founders
Product
Problem
Developers using AI coding assistants often introduce security flaws—such as leaked secrets, vulnerable dependencies, or unsafe commands—directly into code before any traditional review or CI pipeline can catch them. This “shift‑right” approach leaves organizations exposed to supply‑chain attacks, data leaks, and compliance violations.
Solution
Rafter provides an agent‑first security platform that embeds secret scanning, command interception, policy enforcement, and extension auditing directly into the AI coding assistant’s feedback loop. By installing a single CLI command, the tool integrates with popular agents (Claude Code, Codex CLI, Gemini CLI, Cursor, Windsurf, Continue.dev, Aider, OpenClaw) and runs locally to catch risks as code is written. For deeper analysis, an optional remote Code Security Engine performs full SAST, SCA, and data‑flow audits via an API key, delivering structured JSON reports that agents can act on automatically. The solution operates offline, supports deterministic results, and requires no complex setup, enabling fast‑moving teams to ship code with confidence.
Target Audience
Rafter is aimed at development teams that build software with AI coding assistants, including fast‑moving startups, enterprise engineering groups, and DevSecOps engineers seeking automated, shift‑left security.
Features
- Local agent integration that provides real‑time secret scanning, dangerous command interception, and policy enforcement during code authoring
- Extension auditing to validate third‑party plugins and MCP server configurations before they execute
- Remote Code Security Engine offering deep SAST, SCA, and data‑flow analysis on demand via a simple API key
- Deterministic JSON output and documented exit codes for reliable consumption by AI agents and CI pipelines
- One‑command installation (`rafter agent init --all`) that auto‑detects and configures across nine supported agents and IDEs
- Offline operation for local scans, ensuring security even without network connectivity
- CI/CD integrations (GitHub Actions, GitLab CI, CircleCI) for continuous compliance and audit logging