{"slug": "looking-for-an-arxiv-endorser-ai-computer-science", "title": "Looking for an ArXiv Endorser – AI / Computer Science", "summary": "EGA V9, an execution-governance framework for deterministic replay, provenance-aware verification, trust-state evaluation, and fail-closed containment in autonomous AI workflows, is now available via npm as 'ega-v9', with a free 90-day Evaluation License activated by running 'npx ega-v9 register'. The framework integrates with LangChain, OpenAI Agents SDK, CrewAI, AutoGen, and MCP Tool Server, adding a runtime governance layer without requiring framework migration or prompt rewrites, and claims runtime verification in milliseconds with near-zero cost. The official replication guide, published on GitHub at https://github.com/paibyun9/EGA-V9.git, includes a validation suite to verify seven core governance properties.", "body_md": "Official Replication Guide for the EGA V9 paper\n\nEGA V9 is an execution-governance framework for deterministic replay, provenance-aware verification, trust-state evaluation, and fail-closed containment in autonomous AI workflows.\n\nThis README serves as the official replication guide accompanying the EGA V9 paper.\n\n☐ AI agent tool calls cannot be verified.\n\n☐ Agent execution cannot be replayed.\n\n☐ State corruption is difficult to diagnose.\n\n☐ Workflow failures are hard to reproduce.\n\n☐ Prompt injection leaves little audit evidence.\n\n☐ Multi-agent execution becomes a black box.\n\n✓ Replay Verification\n\n✓ Runtime Governance\n\n✓ Trust-State Evaluation\n\n✓ Fail-Closed Containment\n\n✓ Execution Provenance\n\n✓ LangChain + EGA V9\n\n✓ OpenAI Agents SDK + EGA V9\n\n✓ CrewAI + EGA V9\n\n✓ AutoGen + EGA V9\n\n✓ MCP Tool Server + EGA V9\n\nNote;\n\nEGA V9 complements your existing agent framework—it does not replace it.Keep your orchestration logic, prompts, and tool definitions. EGA adds deterministic runtime verification underneath.\n\n| Existing Stack | Existing Stack + EGA |\n|---|---|\nLangChain / Framework |\nLangChain / Framework |\n| ↓ | ↓ |\nYour Agent |\nYour Agent |\n| ↓ | ↓ |\nLLM / External ToolsBlack Box |\nEGA Runtime Governance Layer |\n| ├─ Replay Verification | |\n| ├─ Runtime Governance | |\n| ├─ Trust-State Evaluation | |\n| ├─ Fail-Closed Containment | |\n| └─ Execution Provenance | |\n| ↓ | |\nLLM / External Tools |\n\nNo framework migration. No prompt rewrite. No workflow redesign.\n\nJust add EGA Runtime Governance.\n\n``` js\nconst { ega } = require(\"ega-v9\");\n\napp.use(ega.guard());\n```\n\n**Existing Stack + One Runtime Governance Layer = Deterministic Governance for AI Execution**\n\nGet EGA V9 running in less than one minute.\n\n```\nnpm install ega-v9\n```\n\nAfter the installation completes, run:\n\n```\nnpx ega-v9 register\n```\n\nThis activates your free 90-day Evaluation License.\n\n```\nYou will be prompted for:\n\nContact Name\nCompany Name\nWork Email\n\nAfter successful activation, you will see:\n\n✓ Evaluation License Activated\n✓ EGA V9 is now activated.\n```\n\nCreate a file named `quick-start.cjs`\n\n, paste the following code, and save it.\n\n``` js\nconst { verifyExecution } = require(\"ega-v9\");\n\nconst workflow = [\n  {\n    step: 1,\n    action: \"search_product\",\n    item: \"Laptop\"\n  },\n  {\n    step: 2,\n    action: \"checkout_request\"\n  }\n];\n\nconst result = verifyExecution(workflow);\n\nconsole.log({\n  status: result.status,\n  trustState: result.trust.currentTier,\n  executionAllowed: result.containment.executionAllowed,\n  containmentActivated: result.containment.activated\n});\nnode quick-start.cjs\n{\n  status: 'verified',\n  trustState: 'T1',\n  executionAllowed: true,\n  containmentActivated: false\n}\n```\n\n| Environment | Status |\n|---|---|\n| CommonJS | ✅ Verified |\n| ESM | ✅ Verified |\n| TypeScript | ✅ Verified |\n| Express | ✅ Verified |\n| npm install | ✅ Fresh Install Verified |\n| npm audit | ✅ 0 Vulnerabilities |\n\n*This step is required only once per machine.*\n\n- ⚡\n**Fast**— Runtime verification in milliseconds. - 💰\n**Near-Zero Cost**— Runtime verification without external LLM or API calls. - 🚀\n**Simple**— Integrate with just a few lines of code. - 🔒\n**Secure**— Protect AI workflows with deterministic governance and fail-closed execution.\n\n**Replay**— Reconstruct every workflow exactly.** Auditability**— Generate cryptographically verifiable runtime evidence.** Runtime Verification**— Detect execution inconsistencies.** Deterministic Governance**— Govern AI with predictable decisions.** Fail-Closed Execution**— Automatically contain unsafe workflows.\n\nDon't trust our claims. Verify them yourself.\n\nRun the governance validation suite locally and verify the core runtime governance properties of EGA V9 using the same SDK implementation included in this repository.\n\n```\ngit clone https://github.com/paibyun9/EGA-V9.git\ncd EGA-V9\nnpm ci\n```\n\nValidate the seven core governance and runtime-integrity properties.\n\n```\n# 1. Deterministic Replay Root Verification\nnpm run test:replay-root\n\n# 2. Workflow Divergence Detection\nnpm run test:workflow-divergence\n\n# 3. Trust-State Escalation\nnpm run test:trust-state\n\n# 4. Fail-Closed Containment\nnpm run test:fail-closed\n\n# 5. Tool Invocation Order Integrity\nnpm run test:tool-order\n\n# 6. Approval Bypass Defense\nnpm run test:approval-bypass\n\n# 7. Workflow-Level Tool Injection Detection\nnpm run test:tool-injection\n```\n\nNote\n\n- All validation runs locally.\n- Zero external API calls.\n- Deterministic execution.\n- Each test automatically generates reproducible JSON and Markdown evidence files.\n\nValidation artifacts are generated automatically under `publication/evidence/`\n\n.\n\nArtifacts include:\n\n- JSON report\n- Markdown report\n- Final PASS / FAIL status\n\n| Validation | Expected Result |\n|---|---|\n| Replay Root Verification | ✅ PASS |\n| Workflow Divergence Detection | ✅ PASS |\n| Trust-State Escalation | ✅ PASS |\n| Fail-Closed Containment | ✅ PASS |\n| Tool Order Integrity | ✅ PASS |\n| Approval Bypass Defense | ✅ PASS |\n| Workflow-Level Tool Injection Detection | ✅ PASS |\n\nThese tests validate the core deterministic-governance capabilities evaluated by EGA V9.\n\nAdditional adversarial testing was performed beyond the core validation suite.\n\n**Verified**\n\n- ✅ Fail-closed containment during an active mismatch\n\n**Not Established**\n\n⚠️ Persistent containment after the triggering condition is removed⚠️ Exactly-once side-effect execution under concurrent, retry, or duplicate execution⚠️ Complete evidence-contract integrity\n\n**Not Verified**\n\n⚠️ Multi-step compositional governance⚠️ Direct interception of already-attempted external side effects\n\nThese results define the current verified capability boundary of EGA V9.\n\nDon't trust our claims. Verify them yourself — including the capabilities we have not yet established.\n\n**We don't hide problems. We publish them, prioritize them, and work to resolve them.**\n\nThe next step is simple: evaluate EGA V9 inside a real enterprise AI workflow.\n\nBefore evaluating EGA V9, define the policies that your AI system must follow.\n\n**Example: AI Shopping Policy**\n\n| Workflow | Company Policy |\n|---|---|\n| Purchase | Payment must be completed before shipment. |\n| Refund | Refunds are allowed within 30 days and require manager approval. |\n| Return | Product inspection is required before approval. |\n\n**Example Policy Configuration**\n\n```\n{\n  \"purchase\": {\n    \"paymentRequired\": true\n  },\n  \"refund\": {\n    \"managerApproval\": true\n  }\n}\n```\n\nIllustrative example only. Actual policy integration depends on your application architecture.\n\nInstall EGA V9:\n\n```\nnpm install ega-v9\n```\n\nPlace EGA V9 at the governed execution boundary:\n\n```\nCustomer\n  → AI Shopping Agent\n  → Company Policy\n  → [ EGA V9 ]\n  → Inventory API | Payment API | Refund API | Shipping API\n```\n\nEGA V9 evaluates configured governance conditions before governed tool execution and records deterministic governance evidence.\n\nIt does not replace the AI agent or the company policy. It governs the execution path between the agent decision and the protected tool or API.\n\n**Standard Order Pipeline**\n\n```\nCustomer Order\n  → Agent\n  → [ EGA V9 ]\n  → Payment API\n  → PASS\n```\n\n**Policy-Enforced Refund Pipeline**\n\n```\nRefund Request\n  → Agent\n  → Manager Approval\n  → [ EGA V9 ]\n  → Refund API\n  → PASS\n```\n\nA valid governed workflow proceeds to execution.\n\nWhen EGA V9 detects a policy or integrity violation on the governed execution path, it fails closed before the protected tool call and records governance evidence.\n\nThis protection has a defined boundary.\n\nFail-closed containment remains effective while the triggering policy or integrity violation is actively detected.\n\nEGA V9 does **not** currently establish persistent containment across renewed execution attempts after the original triggering condition has been removed.\n\nExactly-once side-effect execution under concurrent, retry, or duplicate execution is also **not established** in EGA V9.\n\nSee **Capability Boundaries** above for the complete current validation status.\n\nNow test EGA V9 against your own workflows, policies, and failure scenarios.\n\nThen make the decision:\n\n```\nYES / NO\n```\n\nShould EGA V9 be deployed at the execution boundary of your production AI workflow?\n\nThe answer should come from your own evaluation and evidence.\n\nAfter running the evaluation, examine what changes when execution governance is present.\n\n**Operational Reliability**\n\n- Enforces configured governance conditions before protected tool execution.\n- Detects workflow mutations before protected execution.\n- Records deterministic evidence when governance decisions are made.\n\n**Security**\n\n- Detects approval-bypass attempts.\n- Detects unauthorized tool execution.\n- Detects policy and workflow mutations.\n- Applies fail-closed containment when a governed violation is detected.\n\n**Governance**\n\n- Replayable governed workflows.\n- Traceable execution.\n- Auditable governance decisions.\n- Reproducible execution evidence.\n\n**Engineering**\n\n- Deterministic workflow debugging.\n- Reproducible evidence for incident investigation.\n- Clear PASS / BLOCK execution outcomes.\n\n**Business**\n\n- Adds a governance checkpoint before high-impact AI actions.\n- Provides evidence for internal security and governance review.\n- Makes execution failures and policy violations easier to inspect.\n\nActual operational impact will depend on your workflow, policies, architecture, and deployment environment.\n\nAfter completing the evaluation, ask one question:\n\nAssume the company policy requires manager approval before a refund.\n\n**Without an Execution-Governance Check**\n\n```\nRefund Request\n  → AI Agent\n  → Refund API\n```\n\nIf the required approval is missing or bypassed and no equivalent execution-governance check exists at this boundary, the refund request can reach the Refund API.\n\n**With EGA V9**\n\n```\nRefund Request\n  → AI Agent\n  → [ EGA V9 Runtime Governance ]\n       │\n       ├─ Valid workflow\n       │     → Refund API\n       │     → PASS + evidence\n       │\n       └─ Policy / integrity violation detected\n             → BLOCK\n             → Containment + evidence\n```\n\nThe difference is the execution boundary.\n\nEGA V9 checks the configured governance conditions before the protected tool call.\n\nIf the workflow is valid, execution proceeds.\n\nIf a policy or integrity violation is detected, the governed execution path fails closed before reaching the protected API and the decision is recorded as deterministic governance evidence.\n\n**Example Containment Evidence (Simplified)**\n\n```\n{\n  \"decision\": \"BLOCK\",\n  \"reason\": \"policy_violation\",\n  \"containment\": true,\n  \"executionAllowed\": false\n}\n```\n\nEGA V9 does not replace your AI agent.\n\nIt does not replace your foundation model.\n\nIt does not replace your company policies.\n\n**It adds a deterministic governance boundary before protected AI execution.**\n\n```\nAI Decision\n     ↓\nCompany Policy\n     ↓\n[ EGA V9 ]\n     ↓\nPASS  → Execute + Evidence\nBLOCK → Contain + Evidence\n```\n\nThe core question is not whether you trust the AI agent.\n\n**The core question is whether you can verify and govern what the agent is about to execute.**\n\nTest EGA V9 in your own environment.\n\n**Verify the evidence. Understand the capability boundaries. Then decide whether to deploy it.**\n\nWhether you are evaluating EGA V9, exploring enterprise adoption, or experimenting with new AI workflows, feedback and collaboration are welcome.\n\n-\nGitHub Issues:\n\n[https://github.com/paibyun9/EGA-V9/issues](https://github.com/paibyun9/EGA-V9/issues)Use GitHub Issues for questions, bug reports, feature requests, documentation feedback, and independent reproducibility reports.\n\n- Live Vercel Demo:\n[https://ega-v9.vercel.app/](https://ega-v9.vercel.app/) - LCM Official Website:\n[https://lcm3.com/](https://lcm3.com/)\n\nReleased under the MIT License.\n\n**Immediate Priority**\n\n- Persistent containment across renewed execution attempts (address the gap identified in Capability Boundaries above)\n\n**Subsequent Priorities**\n\n- Exactly-once side-effect governance for concurrent and retry scenarios\n- Stronger evidence-contract integrity\n- Compositional governance across multi-step execution paths\n- Deeper interception at external execution boundaries\n\n**Ongoing**\n\n- Improve SDK integrations\n- Expand language support\n- Add enterprise deployment examples\n- Continue benchmark reproducibility\n\nDo not trust a published result merely because it appears in a paper. Reproduce the benchmark, regenerate the publication artifact, and verify the release gates.", "url": "https://wpnews.pro/news/looking-for-an-arxiv-endorser-ai-computer-science", "canonical_source": "https://github.com/paibyun9/EGA-V9", "published_at": "2026-08-19 05:15:32+00:00", "updated_at": "2026-08-19 05:41:11.798518+00:00", "lang": "en", "topics": ["ai-agents", "ai-safety", "ai-infrastructure", "developer-tools"], "entities": ["EGA V9", "LangChain", "OpenAI Agents SDK", "CrewAI", "AutoGen", "MCP Tool Server", "npm", "GitHub"], "alternates": {"html": "https://wpnews.pro/news/looking-for-an-arxiv-endorser-ai-computer-science", "markdown": "https://wpnews.pro/news/looking-for-an-arxiv-endorser-ai-computer-science.md", "text": "https://wpnews.pro/news/looking-for-an-arxiv-endorser-ai-computer-science.txt", "jsonld": "https://wpnews.pro/news/looking-for-an-arxiv-endorser-ai-computer-science.jsonld"}}