TealTiger provides an open-source runtime governance layer for AI agents, with a focus on enforcing deterministic policy, controlling operational costs, and generating audit-ready evidence before an agent interacts with a model, tool, or workflow. The platform is designed for organizations deploying autonomous AI systems in production environments where security, compliance, cost accountability, and operational integrity are critical. It targets sectors such as financial services, healthcare, and platform infrastructure, addressing needs like model risk management, regulatory audit trails, PHI governance, and multi-agent orchestration.
The system offers a unified runtime approach to governance, integrating security, cost, identity, evidence, and workflow controls. Features include a policy engine that operates before execution, options for freezing rules, plan-only mode, hot-swapping policy bundles, anti-tamper detection, and deterministic decision-making that does not rely on an LLM within the governance path. TealTiger covers multiple governance domains—such as security, cost, reliability, memory, registry, evidence, identity, workflow, temporal, and drift—each mapped to real risk scenarios. It supports cryptographic receipts, reason codes, OpenTelemetry spans, SIEM exports, SARIF, JUnit, and JSON evidence formats for audit purposes. The platform also implements non-human identity controls like scope-bound access, zero standing privilege, workload attestation, and enforcement of revoked agents.
Data protection features include PII detection, prompt-injection checks, content policy enforcement, Unicode normalization, and encoded output detection. Financial controls allow for setting budgets by request, session, agent, and day, with anomaly alerts and reasoning-token ceilings. TealTiger integrates with various LLM providers—including OpenAI, Anthropic, Gemini, Bedrock, Azure OpenAI, Cohere, and Mistral—through a provider adapter layer. It installs via standard developer channels such as npm for TypeScript, pip for Python, and Docker, and does not require a hosted governance service for policy enforcement.
As an open-source project, TealTiger recognizes contributions from its community and is part of the NVIDIA Inception ecosystem for AI startups. The platform is positioned as a comprehensive governance solution for autonomous AI, enabling organizations to maintain control, accountability, and compliance across their AI agent workflows.
TealTiger is a LLM eval & observability product. It focuses on securing and managing AI agent applications with guardrails, policy enforcement, and cost tracking. It is built as an open-source project for llm application developers. TealTiger is open source under the Apache-2.0 license. It runs on the web, the command line, and API, and it can be self-hosted.
TealTiger first shipped in 2026. Development happens publicly on GitHub with 28 commits in the last 90 days. PulseGate's similarity index finds few close equivalents — TealTiger occupies a relatively distinct niche. Key capabilities include guardrails, cost tracking, and policy management.
Latest indexed changes and source events
tealtiger.ai discovered by the PulseGate indexer
Other apps tracked under the same category.
