llm-analytics-apps
CLAUDE.md6 starsChanged 8 months agoMITView on GitHub
Internal tooling for the PostHog LLM Analytics team. Provider examples have been moved to the SDK repos (posthog-python and posthog-js) as examples/example-ai-* directories. What remains here: - Demo data generator (scripts/generate_demo_data.py) — generates realistic multi-turn conversations using PostHog AI SDKs - Trace generator (trace-generator/) — creates mock LLM trace data without real API calls - Test scripts (scripts/test_*.py) — integration tests for specific SDK features - Example runner (run-examples.sh) — discovers and runs examples from sibling SDK repos - Screenshot…
- Reads credentials
The file
Files
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
Repository Overview
Internal tooling for the PostHog LLM Analytics team. Provider examples have been moved to the SDK repos (posthog-python and posthog-js) as examples/example-ai-* directories.
What remains here:
- Demo data generator (
scripts/generate_demo_data.py) — generates realistic multi-turn conversations using PostHog AI SDKs - Trace generator (
trace-generator/) — creates mock LLM trace data without real API calls - Test scripts (
scripts/test_*.py) — integration tests for specific SDK features - Example runner (
run-examples.sh) — discovers and runs examples from sibling SDK repos - Screenshot demo (
screenshot-demo/) — UI screenshot tool
Key Commands
# Setup
make setup # install deps via uv
# Run SDK examples from sibling repos
./run-examples.sh --list # list all available examples
./run-examples.sh --parallel # run all in parallel via phrocs
./run-examples.sh anthropic # run by name filter
# Generate demo data
make demo-data # 5 conversations, random providers
make demo-data-quick # 3 short, single provider
# Trace generator (no LLM calls)
make run-trace-generator
# Run individual scripts
uv run scripts/test_litellm.pyProject Structure
├── pyproject.toml # Python dependencies (managed by uv)
├── package.json # Node dependencies (managed by pnpm)
├── Makefile # all common tasks
├── run-examples.sh # SDK example runner
├── scripts/ # demo data and test scripts
│ ├── generate_demo_data.py # multi-provider demo data
│ ├── test_*.py # Python integration tests
│ └── test_*.ts # Node integration tests (Vercel AI, OTel)
├── trace-generator/ # mock trace builder
└── screenshot-demo/ # UI screenshot toolConfiguration
Environment variables in .env:
POSTHOG_API_KEY,POSTHOG_HOST— PostHog connectionANTHROPIC_API_KEY,GEMINI_API_KEY,OPENAI_API_KEY— provider keysPOSTHOG_PYTHON_PATH,POSTHOG_JS_PATH— local SDK paths for development
Trace Generator Event Types
The trace generator creates three event types:
$ai_trace— top-level trace events$ai_span— intermediate span events (nestable)$ai_generation— LLM generation events with model and token info
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

