[Integration]: TealTiger — Governance decisions as Langfuse spans

[nagasatish007]

Issue Body

What

We've built a tealtiger.integrations.langfuse module that exports AI agent governance decisions as Langfuse spans — giving teams governance visibility inline with their existing LLM traces.

How it works

from langfuse import Langfuse
from tealtiger.integrations.langfuse import LangfuseGovernanceExporter

langfuse = Langfuse()
exporter = LangfuseGovernanceExporter(langfuse)

# Each governance decision → Langfuse span
from tealtiger import observe
from openai import OpenAI

client = observe(OpenAI(), on_decision=exporter.trace)

Each governance decision becomes a Langfuse span with:

Field	Value
name	tealtiger.governance
level	ERROR (deny), WARNING (monitor), DEFAULT (allow)
metadata	action, reason_codes, risk_score, evaluation_time_ms, cost, PII findings
input	tool/action being governed
output	governance decision result

What is TealTiger?

TealTiger is an open-source (Apache 2.0) deterministic governance SDK for AI agents. It provides policy enforcement, PII detection, cost tracking, and audit evidence — with no LLM in the governance path and <5ms overhead.

• PyPI (4k+ downloads)
• npm
• Already integrated with: Haystack, AG2, Composio

Why Langfuse?

Teams already using Langfuse for LLM observability want to see governance decisions inline with their traces — without switching tools. The span model maps perfectly: one governance decision = one span with appropriate level coloring.

Source

• Integration module: src/tealtiger/integrations/langfuse.py
• Example: examples/langfuse_governance_traces.py
• Tests: tests/test_langfuse_integration.py

Ask

Would you be open to listing TealTiger on your community integrations page? Happy to submit a docs PR in whatever format you prefer. We can also add a cookbook/example if that's more appropriate.