How security teams gain auditability, cost attribution, and gateway policy control over GitHub Copilot usage—stated with complete architectural honesty.
Copilot Chat in Agent mode in VS Code now supports native agent hooks. ReclaimLLM registers pre-execution hooks there, the same way it does for Claude Code, Cursor, and Gemini CLI, to redact secrets before tool calls run and to capture the full session. Inline completions don't pass through hooks, so ReclaimLLM governs them through Enterprise LLM Gateway proxy routing and OpenTelemetry (OTEL) trace auditing.
Current Support & Controls
Register native pre-execution hooks for Copilot Chat in Agent mode via rclm-hooks-install --copilot. Redact secrets before tool calls run, capture the full session, and configure MCP recall.
Route enterprise model endpoints through the centralized ReclaimLLM Gateway. Manage provider credentials, issue team-scoped keys, and enforce model policy restrictions—especially for inline completions outside the hook path.
Ingest standard OTLP/HTTP traces from corporate proxy layers to capture models, prompt sizes, token counts, and latency into the ReclaimLLM session stream.
The Mixed Toolchain Problem
In modern software teams, engineers use GitHub Copilot for code completion, but switch to Cursor for multi-file refactoring, Claude Code for CLI debugging, and ChatGPT for architectural questions. Securing Copilot alone leaves massive blind spots.
ReclaimLLM unifies telemetry from Copilot, Cursor, Claude Code, Gemini CLI, and browser assistants into a normalized session record. CISOs see total AI spend, active users, and risk flags across every tool in one dashboard.
Explore enterprise observability →Agent-mode tools, including Copilot Chat in Agent mode, Claude Code, and Cursor, execute commands that can read production .env files. ReclaimLLM provides deep pre-execution hook DLP where the leak risk is highest.
| Capability | GitHub Copilot | Cursor | Claude Code |
|---|---|---|---|
| Pre-execution hook secret redaction | Available now (VS Code Agent mode) | Available now | Available now |
| Session capture + MCP recall | Available now | Available now | Available now |
| Enterprise LLM Gateway routing | Available now | Available now | Available now |
| OpenTelemetry (OTEL) trace audit | Available now | Available now | Available now |
| Token spend attribution by team/project | Available now | Available now | Available now |
| AES-256 session blob encryption | Available now | Available now | Available now |
Yes, for Copilot Chat in Agent mode in VS Code. ReclaimLLM installs native VS Code agent hooks (rclm-hooks-install --copilot) that redact secrets before tool calls run and capture the full session, and it configures the ReclaimLLM MCP server so Copilot can search past sessions. Inline code completions don't go through hooks; ReclaimLLM covers those through the Enterprise LLM Gateway and OpenTelemetry trace exports.
For Copilot Chat in Agent mode in VS Code, ReclaimLLM registers native pre-execution hooks for secret redaction and full session capture, plus MCP recall. For inline completions, organizations route enterprise endpoints through the ReclaimLLM LLM Gateway, ingest OTLP telemetry traces into unified observability dashboards, and enforce zero-training and data retention policies across all developer AI usage.
Most engineering organizations have mixed environments where developers use GitHub Copilot alongside agentic tools like Cursor, Claude Code, and Gemini CLI. ReclaimLLM provides native pre-execution hook DLP for Copilot (VS Code Agent mode), Cursor, and Claude Code while aggregating Copilot OTEL telemetry into a single, unified CISO audit log.
Run rclm-hooks-install --copilot, or rclm-hooks-install --local --copilot to commit the hook config to a repository (.github/hooks/rclm.json). See the launch post at /news/github-copilot-and-vs-code-agent-hooks-with-mcp-capture.
Observe Copilot usage, enforce gateway policies, and deploy hook DLP for Copilot (VS Code Agent mode), Cursor, and Claude Code.