log-mcp
log-mcp is a Python-based MCP server that analyzes large log files for AI agents. It provides tools to scan log overviews, search by regex/level/time, extract segments, deduplicate and summarize errors, compute log statistics, compare patterns across files, and classify lines using an optional Rust TF-IDF classifier plus an optional BERT-mini re-scoring stage. It can run with or without the Rust classifier (Python fallback).
Score Breakdown
⚙ Agent Friendliness
🔒 Security
Likely runs locally as a local MCP server (no auth described). README includes an external installation command and references handling of an ANTHROPIC_API_KEY for training/labeling, but does not document secret storage practices or logging redaction. Network/TLS/auth for the MCP transport are not described; dependency hygiene cannot be verified from provided content.
⚡ Reliability
Best When
You have large (tens of MB to many GB) logs where you want an AI agent to narrow down interesting lines/patterns before sending summaries back into a context window.
Avoid When
You need a standards-based web API with documented HTTP error codes, strict auth/scope semantics, or guaranteed consistent ordering/results; or you must run in environments without permission to execute local commands/models (it appears to run locally via MCP server).
Use Cases
- • Analyzing large log files efficiently for root-cause style insights (errors, anomalies, security events).
- • Generating grouped/deduplicated error summaries with stack traces/fingerprints.
- • Comparing two or more runs to find patterns unique to each file or frequency outliers.
- • Surfacing anomalies that may not use explicit ERROR/FATAL levels via classification.
- • Building agent workflows that need structured log querying (overview/search/extract/analyze).
Not For
- • Interactive, manual browsing of raw log contents (the tools compress output for AI).
- • Use as a general-purpose log search UI without agent integration.
- • Situations requiring strongly validated data contracts, strict pagination semantics, or guaranteed deterministic outputs (not evidenced).
Interface
Authentication
README describes installing/running the MCP server locally and registering it with Claude; no authentication or scope model is described for the MCP transport.
Pricing
No pricing model for the package itself is described; README references potential Claude Batch API costs for labeling.
Agent Metadata
Known Gotchas
- ⚠ No explicit tool I/O schema, argument validation rules, or MCP error taxonomy is provided in the README content; agents may need to handle unexpected tool failures themselves.
- ⚠ Optional ML stages (Rust TF-IDF, optional BERT-mini with GPU/Metal) may cause variability in runtime/latency and failure modes depending on environment.
- ⚠ Classifier can be used where explicit ERROR/FATAL is missing, so agents should not assume level-based results are complete.
- ⚠ Large-file processing may be slow depending on hardware and classifier availability; agent timeouts/retries are not documented.
Alternatives
Full Evaluation Report
Comprehensive deep-dive: security analysis, reliability audit, agent experience review, cost modeling, competitive positioning, and improvement roadmap for log-mcp.
AI-powered analysis · PDF + markdown · Delivered within 30 minutes
Package Brief
Quick verdict, integration guide, cost projections, gotchas with workarounds, and alternatives comparison.
Delivered within 10 minutes
Score Monitoring
Get alerted when this package's AF, security, or reliability scores change significantly. Stay ahead of regressions.
Continuous monitoring
Scores are editorial opinions as of 2026-03-30.