Okahu — Claude Code Skills

Add observability to your Python apps using Claude Code slash commands. Scan your codebase, add tracing, run with telemetry, and view traces — all from the CLI.

Instrumentation is powered by Monocle2AI.

Install

/plugin marketplace add okahu-demos/claude-skills
/plugin install okahu@claude-skills

Commands

Instrumentation workflow

Command	Description
/ok-scan [path]	Scan codebase to recommend what to trace. Detects frameworks (LangChain, OpenAI, Flask, etc.) and entry points.
/ok-find [query]	Find methods by search or exact name, then trace execution paths back to entry points.
/ok-instrument [path]	Add tracing — zero-code (generates okahu.yaml) or code-based (injects setup code).
/ok-run [command]	Run your app with tracing enabled. Auto-detects instrumentation approach.
/ok-local-trace [query]	View local traces from .monocle/ folder. Supports natural language queries.
/ok-add-framework <name>	Add instrumentation support for a new AI/ML framework.
/ok-imagegen <prompt>	Generate visual assets using Codex CLI with gpt-image-2.

Session management

Command	Description
/ok-pause [path]	Save conversation context to SESSION.md for cross-session continuity.
/ok-resume [path]	Resume an instrumentation session from saved context.

Other

Command	Description
/ok-login [env]	Authenticate with Okahu Cloud via GitHub Device Flow.
/kahu <query>	Query the Okahu SRE Agent API (natural language).

Typical workflow

/ok-scan ./my-app          # Analyze codebase, pick entry points
/ok-instrument ./my-app    # Generate tracing config
/ok-run                    # Run with tracing enabled
/ok-local-trace            # View collected traces