💡 Feature Description and Proposed Solution
TealTiger is a deterministic governance SDK for AI agents — it evaluates tool allowlists, PII detection, secret scanning, and cost budgets before every consequential action, producing a structured decision event each time.
Proposed integration: Emit TealTiger governance decisions as AgentOps ActionEvents so they appear in the session timeline alongside LLM traces and tool calls.
What this would look like:
import agentops
from tealtiger import TealOpenAI
agentops.init(api_key="your-key")
# TealTiger decisions auto-emit to AgentOps
client = TealOpenAI(
api_key=os.environ["OPENAI_API_KEY"],
guardrails={"pii_detection": True, "secret_detection": True},
budget={"max_cost_per_session": 5.00},
)Each governance decision would emit an event with:
action_type: "governance_decision"
result: allow / deny / monitor
params: {tool_name, reason_codes, risk_score, evaluation_time_ms}
This gives teams visibility into why an agent was blocked, right next to the LLM trace — without switching tools.
Why AgentOps specifically: You already track cost and LLM calls. Governance decisions are the missing dimension — they explain why something didn't happen, which is invisible in traces today.
TealTiger: https://github.com/agentguard-ai/tealtiger
PyPI: pip install tealtiger
Already integrated with AG2, CrewAI, LangChain, Vercel AI SDK
🤔 Related Problem
🤔 Related Problem:
When agents are governed (tool calls blocked, budget exceeded, PII detected), there's no visibility into those denials in the observability dashboard. The session timeline shows a gap — the tool call just... doesn't happen — with no explanation of why. Teams end up correlating separate governance logs manually.
🤝 Contribution
💡 Feature Description and Proposed Solution
TealTiger is a deterministic governance SDK for AI agents — it evaluates tool allowlists, PII detection, secret scanning, and cost budgets before every consequential action, producing a structured decision event each time.
Proposed integration: Emit TealTiger governance decisions as AgentOps
ActionEvents so they appear in the session timeline alongside LLM traces and tool calls.What this would look like:
Each governance decision would emit an event with:
action_type: "governance_decision"
result: allow / deny / monitor
params: {tool_name, reason_codes, risk_score, evaluation_time_ms}
This gives teams visibility into why an agent was blocked, right next to the LLM trace — without switching tools.
Why AgentOps specifically: You already track cost and LLM calls. Governance decisions are the missing dimension — they explain why something didn't happen, which is invisible in traces today.
TealTiger: https://github.com/agentguard-ai/tealtiger
PyPI: pip install tealtiger
Already integrated with AG2, CrewAI, LangChain, Vercel AI SDK
🤔 Related Problem
🤔 Related Problem:
When agents are governed (tool calls blocked, budget exceeded, PII detected), there's no visibility into those denials in the observability dashboard. The session timeline shows a gap — the tool call just... doesn't happen — with no explanation of why. Teams end up correlating separate governance logs manually.
🤝 Contribution