# [RFC] TealTiger v1.3 — Portability, Ecosystem & Enterprise Adoption (Lock-in, Integrations, Cost, State)

[nagasatish007]

[RFC] TealTiger v1.3 — Portability, Ecosystem & Enterprise Adoption (Lock-in, Integrations, Cost, State)

Status: OPEN FOR FEEDBACK
Covers: Requirements 13, 14, 17, 18, 20, 21
Target audience: Enterprise architects, FinOps leads, platform teams, anyone evaluating governance tooling

---

Summary

Enterprise adoption requires confidence that you're not creating vendor lock-in. v1.3 addresses this head-on: portable policy export, open schemas, standalone modules, migration documentation, and ecosystem-level positioning. Plus: expanded provider support, agentic platform integrations, cost governance, and state/temporal controls.

---

Proposed Capabilities

Portability & Lock-In Reduction (Req 20)

• Policy export to OPA-compatible Rego format
• GovernanceProvider interface: swap the evaluation engine behind a stable API
• Published evaluation context schema (open, versioned JSON Schema)
• Every module usable as a standalone package without TealEngine
• Migration guide with effort estimates for OPA, Cedar, and custom engines
• All audit/evidence data exportable in open formats (JSON, SARIF, JUnit XML)

Ecosystem Defensibility (Req 21)

• Integration adapters for: LangChain, LangGraph, CrewAI, MCP, OpenAI SDK, Anthropic SDK, custom runtimes
• Max 5 lines of code to add governance to an existing agent
• Open "Governance Event Schema" for telemetry (not proprietary to TealTiger)
• OTel semantic convention extension for governance spans
• Policy schema, TEEC format, and TealFlow schema published as independent open specs
• Governance dashboard with RBAC, fleet management, approval workflows
• Hybrid deployment: TealTiger handles Layer 2 (runtime), integrates with Layer 1 (gateways/meshes)

Provider Expansion (Req 13)

• 5 new providers: Groq, DeepSeek, Together AI, Hugging Face TGI, xAI (Grok)
• Total: 12 providers (from 7)
• Same guarded client interface, cost tracking, and governance modules across all

Agentic Platform Integrations (Req 14)

• AWS Bedrock Agents guardrail adapter
• AWS AgentCore governance plugin
• Azure AI Agent Service middleware
• Cross-platform: same policy logic, same Decision objects, full module set

Enhanced Cost Governance (Req 17)

• Cost limits as governance artifacts (not developer-configured)
• Per-agent and per-workflow cost attribution
• Cost anomaly detection (spike alerts before execution)
• Separate reasoning-token budgets (thinking/chain-of-thought tokens)
• Cost evidence in TEEC envelopes

State & Temporal Governance (Req 18)

• Context window size limits per agent/provider
• Session TTL enforcement with automatic termination
• Cooldown periods between high-impact actions
• Time-of-day restrictions (e.g., block production deploys outside business hours)

---

Questions for the Community

1. Lock-in concerns — What would make you confident enough to adopt a governance SDK? Is policy export to Rego sufficient, or do you need something else (Cedar support? Custom DSL?)?

2. Provider priority — Of the 5 new providers (Groq, DeepSeek, Together AI, HF TGI, xAI), which would you use first? Are we missing any?

3. Platform integrations — Are you deploying agents on Bedrock Agents, AgentCore, or Azure AI Agent Service? Would a governance adapter change your adoption decision?

4. Cost governance — Is per-agent cost attribution useful for your FinOps workflow? Do you track reasoning tokens separately today?

5. Temporal controls — Would time-of-day restrictions be useful in your environment? What about cooldown periods between high-impact actions?

6. Integration effort — "5 lines of code to add governance" — is that a realistic bar for your team? What's the maximum integration effort you'd accept?

7. Open schemas — Would you adopt TealTiger's governance event schema even if you weren't using TealTiger itself? (i.e., is there value in a shared governance telemetry format across the industry?)

---

How to Give Feedback

• Comment below with thoughts, concerns, or use cases
• React with 👍 on capabilities you'd use
• React with 🤔 on capabilities that seem unclear
• React with 👎 on capabilities you think are wrong

---

Full Spec Reference

See Requirements 13, 14, 17, 18, 20, and 21 in the complete specification