Agent Self-Reflection Sentiment Scanner

Agent Self-Reflection Sentiment Scanner MCP Connector for Claude

A+

Analyzes agent execution logs to detect self-correction markers and measure stability.

3 tools Official Updated Oct 1, 2026 Official Vinkius Partner

This MCP server provides tools to analyze the 'sentiment' of agentic workflows by tracking deterministic self-correction markers. It identifies Error Recognition Markers (e.g., when an agent realizes a mistake) and Success Markers (e.g., task completion). By using scan_logs_for_markers, calculate_rate, and get_summary, users can calculate the Self-Correction Frequency Rate and assess the stability of agentic loops. It is designed to provide visibility into how often agents identify and fix errors during execution.

agentlogsself-correctionstabilitymetrics

3 tools expose this connector's capabilities to your AI agent.

calculate_rate

Computes the statistical frequency of self-corrections relative to the number of loops performed

get_summary

Aggregates all scanning and calculation data into a single high-level report

scan_logs

Analyzes raw log files to identify and count specific markers

See how to talk to your AI agent using Agent Self-Reflection Sentiment Scanner.

Can you scan these logs and tell me the correction rate?

The self-correction frequency rate is 0.15, indicating a High stability score for this execution loop.

What was the summary for execution ID 12345?

Execution 12345 had 2 errors detected and 5 successful completions, resulting in a stable execution profile.

How many error markers were found in the provided log content?

A total of 3 error recognition markers were identified in the log content.

They are specific linguistic phrases like 'I made a mistake' or 'The task is complete' that indicate an agent's internal state transition.

Related Connectors