contextpulse mcp
Real-time context budget tracking and tool-loop detection for MCP-compliant coding agents.
$9
Works with the AI Agents you already use
contextpulse mcp
Example session with this skill installed
Check my current context budget and tell me if I am in danger of looping.
- Read your context and instructions
- Compiled the contextpulse mcp
The current session context is being tracked. I have verified your usage against the defined CONTEXT_LIMIT.
ContextPulse Status Report
| Metric | Value | Status |
|---|---|---|
| Current Tokens | 142,500 | 🟢 Nominal |
| Context Limit | 200,000 | - |
| Utilization | 71.25% | 🟡 Warning |
| Last Tool Call | read_file | - |
| Repeat Count | 1 | 🟢 No Loop |
| Storage | PostgreSQL | 🟢 Connected |
Threshold Analysis
Current usage has crossed the
70% Warning Threshold. You have approximately 57,500 tokens remaining before the hard limit is reached. No tool loops detected; the last three calls were ls, grep, and read_file.
Next steps
- Review the current file context and remove any large, unnecessary buffers.
- Use
git diffinstead of reading entire files to save token space. - Archive the current session to the database if you need to reset the counter.
Connects securely to your tools. The creator never sees your data.
About this skill
Coding agents often burn through context windows or get stuck in repetitive tool loops, leading to wasted tokens and degraded performance. This skill provides real-time monitoring and alerting to keep your agent runs efficient and within budget.
What it does
- Token monitoring tracks live context usage against a defined budget during active sessions.
- Loop detection triggers alerts if the same tool is called three times sequentially.
- Visual alerts provide warnings at 70% and critical notifications at 90% context capacity.
- Persistent logging stores full run history in a local PostgreSQL database for post-mortem analysis.
- Local execution runs entirely on your hardware with no external telemetry or cloud dependencies.
How it works
- Initialize the local PostgreSQL and Redis containers to handle state and history.
- Configure the MCP server in your agent settings (e.g.,
mcp.json) with your database credentials and limits. - Monitor real-time feedback as the agent performs tasks, receiving alerts if usage spikes or loops occur.
- Analyze historical run data to identify which tasks or prompts consume the most context.
Frameworks & tools
Built for MCP-compatible environments. Requires Docker for local PostgreSQL 16 and Redis 7. Optimized for Claude Code, Cursor, and GitHub Copilot.
Why this beats prompting it yourself
Standard prompts cannot track actual token counts or tool execution frequency across a session. This skill uses a dedicated server to maintain state, providing objective metrics that an LLM cannot accurately report on itself.
Use cases
- Prevent runaway costs during long autonomous coding sessions.
- Debug recursive tool-calling patterns in complex agent workflows.
- Audit agent efficiency by reviewing historical token usage across different projects.
- Manage memory constraints on smaller context window models.
Known limitations
Requires local Docker installation to run database dependencies. Only detects loops based on immediate sequential tool calls.
Requires Docker for PostgreSQL and Redis. Loop detection is currently limited to 3x sequential tool calls. No support for remote database hosting without manual env configuration.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 24 days ago
Requires