mixseek-debug_skill

This skill enables and configures MixSeek-Plus debug logging to help diagnose issues by activating verbose mode and controlling log output.

Python

0

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill drillan/mixseek-plus --skill mixseek-debug

SKILL.md7.3 KB

Overview

This skill enables MixSeek-Plus debug features and controls its log output. It exposes environment-variable and API ways to enable verbose logging, target specific components, and route logs to console or workspace files. Use it to diagnose agent behavior, inspect tool calls, and collect traces for troubleshooting.

How this skill works

The skill toggles built-in debug flags and patches MixSeek core to change logger levels at runtime. It can set MIXSEEK_VERBOSE for integrated debug, CLAUDECODE_MODEL_LOG_LEVEL for only the claude model, and MIXSEEK_LOGFIRE to enable Logfire instrumentation. You can apply settings via shell env vars, persistent session exports, or the Python API (enable_verbose_mode(), patch_core(), disable_verbose_mode()).

When to use it

Investigating failing agent runs or unexpected results.
Collecting detailed tool-call timing and parameters for performance analysis.
Tracing ClaudeCode model behavior separately from other components.
Capturing member-agent logs to workspace files for postmortem review.
Adding Logfire traces when deeper pydantic-ai instrumentation is needed

Best practices

Start with MIXSEEK_VERBOSE=1 for most debugging sessions; it enables comprehensive logs without extra config.
Use CLAUDECODE_MODEL_LOG_LEVEL=DEBUG when you need only the model-level logs to reduce noise.
Set MIXSEEK_WORKSPACE so file logs are written to a known location and rotated if needed.
Call mixseek_plus.patch_core() after setting env vars in Python to ensure loggers are patched.
Unset or disable verbose mode after debugging to avoid excessive log volume and disk usage.

Example use cases

Quick CLI debug: MIXSEEK_VERBOSE=1 uv run mskp team "Task" --config configs/agents/team.toml to see tool start/done and result previews.
Isolate model logs: CLAUDECODE_MODEL_LOG_LEVEL=DEBUG uv run mskp team ... to debug only claudecode_model messages.
Persistent session: export MIXSEEK_VERBOSE=1 in a shell to run multiple commands in verbose mode without prefixing each.
Python-managed debug: mixseek_plus.enable_verbose_mode(); mixseek_plus.patch_core() to enable verbose logs from a script and write agent traces to $MIXSEEK_WORKSPACE/logs.
Add Logfire traces: MIXSEEK_VERBOSE=1 MIXSEEK_LOGFIRE=1 and configure logfire to capture pydantic-ai traces for deeper diagnostics.

FAQ

Confirm the env var is exported in the shell or set in os.environ for Python, and ensure mixseek_plus.patch_core() is called when using the Python API. Also check logging handlers via logging.basicConfig or similar.

Where are file logs written?

Member agent logs are placed under $MIXSEEK_WORKSPACE/logs/member-agent-YYYY-MM-DD.log. Set MIXSEEK_WORKSPACE to a writable path before running to ensure files are created.