Provides tools for loading, filtering, and querying GStreamer debug logs, enabling analysis of log files by time range, level, category, and object name.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@GStreamer Logs MCP List available log files and load the most recent one to see its summary."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
MCP server for GStreamer debug logs. One load per file (cached); the agent uses filters (path, category, time range in ms, level, object_name, etc.) and gets counts before requesting lines, so context stays small. Time: integer milliseconds from log start only (e.g. 10000 = 10s, 10300 = 10.3s). Standalone repo: no external project dependencies.
Python 3.10+
Related MCP server: Graylog MCP Server
From this directory:
pip install -r requirements.txtpython server.pyOr with uv:
uv run server.pyEnv var | Meaning |
| Directory containing log files (default: |
In Cursor MCP settings, add a server that runs this script, for example:
{
"mcpServers": {
"gst-logs": {
"command": "python",
"args": ["C:\\path\\to\\gst_logs_mcp\\server.py"]
}
}
}Use the path to your gst_logs_mcp clone and ensure the Python that has mcp installed is the one used by Cursor.
All time filters use integer milliseconds from log start only (no fractions). E.g. 10000 = 10s, 10300 = 10.3s. Use load_log first to get time_span so you can compute ms from the first timestamp.
Tool | Purpose |
| Call this first. Returns the full agent guide: workflow, filters, token-saving rules. Follow it when using the other tools. |
| List available log files (optionally from a given directory). |
| Load and index a log file once; returns total, time_span (first/last), levels, categories, object_count (no object list). Cached for later queries. |
| Counts only (no raw lines): total_matching, count_by_level, count_by_category, count_by_object. Optional filters: time_start, time_end (ms), level, category, object_name, etc. |
| Required: path, category, time_start, time_end (ms). Returns per-level count of distinct objects and full object list. Use to discover object names, then narrow with query_logs. |
| Get lines. Required: path, category, time_start, time_end (ms). Optional: level, object_name, search, etc. limit default 50, max 2000. If total_matching > 100 you get no rows — only total_matching, object_count, and a message; use log_summary first and narrow filters. |
Resource: gst-logs://agent-guide – same content as the agent guide (for clients that support MCP resources).
Workflow: 1) get_agent_guide (once). 2) list_log_files if path unknown. 3) load_log once per file → time_span, levels, categories, object_count. 4) log_summary or object_summary with filters (category + time in ms) → get counts. 5) If total_matching is small (≤100), query_logs with same category + time (+ optional object_name/level/search).
For agents: Call get_agent_guide_tool at the start, or read resource gst-logs://agent-guide. Full guide: AGENT_GUIDE_GST_LOGS_MCP.md.
From project root:
python scripts/test_mcp_tools.pyUses the same core as the server; prints SENT/GOT for list_log_files, load_log, log_summary (with time in ms), object_summary, query_logs. Time in tests is integer milliseconds (e.g. 0, 5000, 10500). The folder gst_log_files/ is gitignored; set GST_LOGS_MCP_LOG_DIR to a directory that contains GStreamer debug logs, or add a small sample log for CI.
This server cannot be deployed
Open-source agent that observes and fixes your application. Query logs, traces, metrics, incidents.
Securely search and manage workspace context files for AI agents and teams.
Structured knowledge base for AI agent solutions. Search, explore, and retrieve build logs.
Query application logs, traces, and metrics from your AI coding assistant via Foam's MCP server.
