Collect
3Aggregates Claude Code / Codex / Cursor local history.
AI Agent Governance Platform
Claude Code
Codex
Cursor
Aggregate prompt history.
Raise safety and productivity together.
Renue Agent Monitor aggregates the local history from Claude Code / Codex / Cursor and gives you usage statistics, risky-operation alerts, and concrete prompt improvement points — all in one place. Beyond the output itself, you can review how your team asked the AI and how they validated the answer.
From our internal operations, opened up externally
Works for organizations and individuals
Mac / Windows installers provided
Catch risky operations early
Continuously monitor access to secrets and risky prompt patterns.
See usage per user
Make it visible who uses which AI and how.
Evaluate conversation quality
Turn good prompting into organizational knowledge.
Detect
24/7Continuously watches for risky operations and secret access.
Review
Per userCompare usage per user to see who is where on the adoption curve.
Improve
4 axesReview-oriented scoring lifts the whole team's level.
Problem
Reviewing only the output misses the risky part of AI use
The more plausible output AI produces, the heavier output review becomes. But looking only at finished code never tells you how your team asked, whether secrets sneaked into prompts, or whether verification conditions were clear.
- Risky prompts and pasted secrets end up hidden inside each person.
- Good and bad usage alike fail to accumulate as team knowledge.
- The know-how of heavy AI users never becomes the org-wide standard.
Solution
Collect the dialogue and turn it into an operable loop
Renue Agent Monitor externalizes the monitoring and evaluation system we run in-house. By aggregating usage logs and connecting statistics, alerts, and scoring, it raises AI usage maturity while protecting security.
Goal
Reward heavier AI users, and lift the whole team's level by looking at how the AI is actually used.
Core Functions
The core features that turn monitoring into improvement
Not just observation: a full loop from visibility to improvement actions.
01
Aggregate local history on the server
Collect scattered Claude Code / Codex / Cursor history from each device and make it reviewable in one place.
02
Show statistics and per-user usage
Volume, tool-level trends, per-user adoption — all laid out so you can see gaps and potential.
03
Risky-operation alerts and improvement scoring
Detects secret-access and other risky signals and returns concrete improvement scoring for your prompts.
Always On Collection
After setup, the hook runs itself on every employee's device
Once the agent is installed on each employee's device, hooks fire automatically as people use Claude Code / Codex / Cursor. No manual upload — usage logs, prompt history, and risky signals are sent to the server continuously.
Per-device continuous collection
Hooks on Mac / Windows watch activity and send relevant data as it happens.
Centralized across the whole team
AI usage logs scattered across employees are consolidated into an org-wide, comparable view.
Admins just read the Insight
Admins review the collected data on the dashboard and focus on alerts and improvement work.
What Gets Sent
What is automatically collected
Prompt / response historyRecords what was asked and how it was verified.
Usage statisticsSessions, messages, and active time.
Risk signalsDetects secret exposure and risky commands.
Improvement materialScores dialogue quality and surfaces points to review.
Screen Preview
Key screens beyond the dashboard — PC layout
From the usage ranking down to per-session detail and the security alert list — the set of screens needed for real operations.
Ranking
Per-employee usage ranking
Compare sessions, messages, and active time per employee to spot heavy adopters and who needs help.
Session Detail
Session detail, one click from the list
Drill into the model used, tool calls, edited files, and actual prompt history — the dialogue behind each output.
Security Alerts
Security alert list
Severity, type, affected employee, and status — track secret exposure and risky commands by priority.
AI Insight
Generative AI reads the usage and proposes improvements
From collected sessions, scores, and work patterns, a generative model summarizes today's usage and points to concrete next things to fix.
MCP Extension
Analyze your own AI usage with an AI agent
With the MCP (Model Context Protocol) extension, admins can pull monitoring data directly from agents like Claude Code and analyze it in place.
AI agents read your data and analyze on the fly
From Claude Code or other MCP-aware agents, access usage statistics, session detail, and per-employee data via the MCP server. With a natural-language request like "find patterns from heavy users and propose things worth spreading", the agent cross-analyzes your data.
Analyze by natural language
Instead of driving a dashboard, ask the AI to analyze directly.
Cross-cutting access to all data
Pull usage stats, session detail, per-employee efficiency, and security info in one go.
Drill in on your own angle
Get company-specific insight beyond the fixed reports.
Spread top users' techniques to the whole team
The agent compares session quality scores and patterns across your team — "who does what well", "which prompt patterns deserve to be spread" — and returns concrete next actions, not just numbers.
Note
The MCP extension is a paid add-on available after contacting us. It is not included in the free plan. If you're interested, please contact us.
How It Works
From onboarding to improvement
Create an account
Choose organization or individual use. Individuals can also start with Google / GitHub / X sign-in.
Connect devices
Distribute the Mac / Windows installer so targeted members can have their usage history continuously collected.
Read the Insight
Use usage stats, risky-op findings, and instruction-quality weak points to revise rules and prompt patterns.
Architecture
How collection works
Hooks fire on each employee's device, data flows to a central server, and admins see it in the dashboard.
Step 1
Each employee's PC
- Uses Claude Code / Codex / Cursor
- Set-up agent runs in the background
- Hooks start automatically per employee
Step 2
Local hook / collector
- Detects usage events
- Extracts history, statistics, and risky operations
- Shapes records for upload
Step 3
Renue Agent Monitor
- Aggregates every employee's logs
- Updates rankings and usage statistics
- Generates alerts and improvement metrics
Step 4
Admin dashboard
- Review per-employee ranking
- Drill into session detail
- Act on security alerts
Security
How we handle security
Encryption in transit, authenticated access, AI-based anonymization on social share — the set of mechanisms we use to handle data safely.
See the security detailsExample review axes
- Whether purpose, constraints, and done-criteria are shared with the AI
- Whether secrets or risky operations are mixed into the prompt
- Whether the verification conditions for generated code are stated up front
- Whether the dialogue process can be reproducibly rolled out to the team
Start Now
Start by making your AI usage visible
This LP is written around making AI usage easier to review, not heavier to review.