{"slug": "credit-nirvana-cve-2026-110474-how-an-agentic-collections-stack-leaked-millions", "title": "Credit Nirvana CVE-2026-110474: How an Agentic Collections Stack Leaked Millions of Calls", "summary": "Security researcher Eaton published a full compromise walkthrough on November 18, 2026 of Credit Nirvana's agentic debt-collection stack, assigning CVE-2026-110474 and CVE-2026-110475, with roughly 315,000 stored call recordings reachable and a users-list API exposing more than 10,000 accounts without authentication. Eaton reported that a hardcoded API key made Amazon Cognito login unnecessary, that Keycloak session handling could be bypassed in the browser with syntactically valid but invalid JWTs, and that Credit Nirvana's login API trusted a userName in the JSON body over an unverified Keycloak token in Authorization-Token. The exposed CollectionWorkz instance was tied to Aditya Birla Capital Finance, and Eaton's chain reached agentic call-center tools including the Call Transcriber, disposition, post-call recommendations and call audit agents.", "body_md": "**Security researcher Eaton published a full compromise walkthrough of Credit Nirvana's agentic debt-collection stack on November 18, 2026**, assigning **CVE-2026-110474** and **CVE-2026-110475**. The [write-up](https://eaton-works.com/2026/11/18/credit-nirvana-hack/?ref=explainx) shows how **authentication could be skipped**, **Keycloak session handling could be bypassed in the browser**, and **backend APIs trusted the wrong headers**, exposing **agentic call tools** and **enterprise collections data** tied to **Aditya Birla Capital Finance**.\n\nThis is not a theoretical agent prompt attack. It is **production fintech software** marketing **AI agents for collections**, with **hundreds of thousands of call recordings** reachable when **identity checks fail**.\n\n## TL;DR: what Eaton demonstrated\n\n| Surface | What Eaton reported | \n|---|---|\n| Agentic call center | **Call Transcriber** , disposition, post-call recommendations, call audit agents | \n| Recordings | **~315,000** stored calls with playback | \n| Core platform | **CollectionWorkz** Credit Nirvana instance for**Aditya Birla Capital Finance** | \n| Auth flaws | **Hardcoded API key** , unused Cognito token,**10k+ users** via unauthenticated list API | \n| IAM | **Keycloak** client patches with fake JWTs; server trusted**userName in body** over token | \n| CVEs | **CVE-2026-110474** ,**CVE-2026-110475** | \n\n## Why agentic collections are a high-value target\n\nDebt-collection software already holds **names, account numbers, payment history, and call audio**. Vendors now brand the same stack as **agentic AI**: **real-time transcription**, **sentiment analysis**, **next-best-action** recommendations, and **proactive outreach campaigns** before accounts go delinquent.\n\nEaton's screenshots and descriptions show those features as **first-class modules**, not future roadmap slides. An attacker who gains **operator-grade access** can **replay calls**, **study negotiation tactics**, and **map which agents run automatically** — intelligence useful for **fraud**, **social engineering**, and **regulatory embarrassment**.\n\nThe incident belongs in the same conversation as **[GAO tests showing models draft attack plans](https://explainx.ai/blog/gao-covert-test-chatgpt-claude-gemini-grok-viable-attack-plans-november-2026)** and **[agent breach trackers](https://explainx.ai/blog/anthropic-claude-models-breach-15-systems-security-incidents-2026)**, but the exploit chain here is **Web 101**: **missing server-side authorization** on APIs behind a React single-page app.\n\nAgent code review pair illustrating weak auth around Credit Nirvana agentic debt tools\n\n## Attack chain: from hardcoded key to full control\n\nEnterprise pilot rollout metaphor for agentic collections UIs that skip server-side auth checks\n\nEaton's narrative starts at the **agentic call center**, where a **hardcoded API key** made the **Amazon Cognito login unnecessary**. The returned Cognito token was **never validated server-side**, so skipping login was enough to reach privileged features.\n\nOn the main **CollectionWorkz** deployment, Eaton traced **React bundles**, found a **users-list API** returning **more than 10,000 accounts without authentication**, and pivoted to **Keycloak** integration. By feeding **syntactically valid but invalid JWTs** and **patching client-side refresh logic**, the browser could be convinced a session might exist.\n\nThe critical backend mistake: **Credit Nirvana's login API honored `userName` in the JSON body** while sending a **Keycloak token in `Authorization-Token` that was not verified**. Eaton picked a **high-privilege username** from the leaked list, obtained a **`cnToken` GUID**, and discovered that **subsequent requests worked when fake Keycloak headers were stripped** and only **`CN-Token`** remained.\n\nThat pattern — **trusting an application token while ignoring the enterprise IAM token** — is how **agent orchestrators** and **human portals** get owned in the same request path.\n\n## What data classes were exposed\n\nEaton describes **call recordings with transcripts**, **disposition analytics**, **campaign builders** for **AI-powered outreach**, and an **AI Agent Registry** to **browse agents inside the collection system**. On the core platform, **user tables**, **role metadata**, and **collections workflows** across **secured and unsecured products** were in scope once APIs accepted the **`CN-Token`**.\n\nEaton notes **Aditya Birla Capital Finance** customers as impacted on the **Agentic AI Call Center**, and documents a far larger **CollectionWorkz** exposure on the primary platform: **6.6 million accounts** across **Aditya Birla Capital Finance**, **BharatPe**, and **International Asset Reconstruction Company (IARC)**, with **ID cards, credit reports, legal documents**, and **call recordings** in scope. Credit Nirvana and Perfios fixed both issues in **June 2026** after CERT-In coordinated disclosure; the **November 18** article is Eaton's public write-up, not a final statutory breach notice. Check **issuer advisories** and **Indian regulatory filings** for confirmed customer-notification timelines.\n\n## What developers building agents should fix\n\n1. **Never embed long-lived API keys in front-end JavaScript.** Any key shipped to browsers is public.\n2. **Validate IAM tokens on the server for every route** (Keycloak's[token validation docs](https://www.keycloak.org/securing-apps/oidc-layers?ref=explainx) describe local JWT checks and introspection), including internal microservices behind your agent UI.\n3. **Do not accept usernames from request bodies when a bearer token is present.** Bind identity to**verified claims** only.\n4. **Treat agent registries like admin consoles.** They reveal**tool names, prompts, and automation scope** attackers need for follow-on abuse.\n5. **Log and alert on unauthenticated list endpoints** such as**`/users-list`** patterns Eaton highlights.\n\nTeams shipping **MCP-connected agents** should compare this to **[MCP tool provider misbinding](https://explainx.ai/blog/mcp-tool-provider-misbinding-ai-agents-security-2026)** and **[Microsoft's PyPI credential stealer targeting agent frameworks](https://explainx.ai/blog/microsoft-agent-framework-pypi-1-26-credential-stealer-november-2026)** — different entry points, same lesson: **agents amplify impact when auth is weak**.\n\n## How to protect collections and agent platforms\n\nIf you operate **voice agents**, **collections SaaS**, or **internal agent hubs**:\n\n- Run **external pen tests** that assume**browser tampering** , not only**OWASP Top 10 scanners** .\n- Segment **call storage** with**per-tenant encryption** and**break-glass access logging** .\n- Apply **rate limits** and**anomaly detection** on**transcript download APIs** .\n- Require **step-up authentication** before**campaign launch** or**agent registry edits** .\n\nFor **runtime guardrails** on autonomous tools — approving transfers, sending messages, or changing account state — teams can use platforms such as [AgentBeam](https://agentbeam.com/?ref=explainx) to **block dangerous agent actions before they execute**, alongside fixing the **authorization bugs Eaton documented**.\n\n## Collections agents versus coding agents: different blast radius\n\nMany teams worry about **[Claude Code prompt injection](https://explainx.ai/blog/google-gtig-sombermeme-claude-code-settings-backdoor-november-2026)** or **[Codex sandbox escapes](https://explainx.ai/blog/openai-inside-codex-sandbox-permission-profiles-november-2026)**. Collections agents add **telephony and regulated speech**:\n\n- **Outbound dialers** can harass consumers if misconfigured.\n- **Transcripts** may include**third-party bystanders** on speakerphone.\n- **Disposition models** influence**legal escalation** paths.\n\nEaton's post is a reminder that **the scariest agent failures** may be **missing auth on `/users-list`**, not **clever jailbreaks**. When you prioritize controls, **inventory admin routes** before you tune **system prompts**.\n\n## Keycloak and React SPAs: why client patches are not security\n\nEaton documented **disabling refresh handlers** and **short-circuiting userinfo calls** inside the browser. That is powerful for **proof of concept**, and terrifying for **defenders**, because it shows **how little the server cared**.\n\nProduction fixes must assume **attackers control the client**:\n\n| Bad pattern | Safer pattern | \n|---|---|\n| Trust `userName` in JSON body | Bind identity to **validated JWT claims** server-side | \n| Accept requests with only app token | Require **IAM token + app token** with matching subject | \n| Ship API keys in bundles | Use **short-lived tokens** from backend exchange | \n| Expose user enumeration APIs | Gate lists behind **admin RBAC** and**rate limits** | \n\nCredit Nirvana's **`CN-Token`** header behaving like a **session cookie** without binding to **Keycloak** matches failures seen in **[MCP tool misbinding](https://explainx.ai/blog/mcp-tool-provider-misbinding-ai-agents-security-2026)** where **routers** treat **metadata** as **proof of identity**.\n\n## Disclosure timeline and what to watch next\n\nEaton published on **November 18, 2026** with **CVE-2026-110474** and **CVE-2026-110475** identifiers in the title and metadata. The author notes **another Keycloak-related issue** planned for **December**, suggesting **coordinated disclosure cadence**.\n\nCustomers should monitor:\n\n- **Credit Nirvana** or**Aditya Birla Capital Finance** security statements\n- **CERT-In** or sector regulators if notifications become public\n- **CISA KEV** listing if vendors assign**CVSS** and**active exploitation** proof\n\nUntil then, treat Eaton's article as **technical primary source**, not a **final victim count**.\n\n## Red-team checklist for agentic fintech UIs\n\nUse this as an internal test plan:\n\n1. **Download the production JavaScript bundle** and grep for`apiKey` ,`cognito` , and`Bearer` .\n2. **Call every `/users` and `/agents` list route** without cookies.\n3. **Replay tokens** from a low-privilege user against**high-privilege routes** .\n4. **Remove headers one at a time** to see which the API**actually enforces** .\n5. **Attempt campaign creation** without**MFA** if the product claims**human approval** .\n\nThese steps mirror how **[nginx2root HTTP/2 RCE](https://explainx.ai/blog/nginx2root-http2-rce-cve-2026-123007-xint-wiz-november-2026)** and **[Java4Shell](https://explainx.ai/blog/java4shell-gpt-6-cyber-cve-2026-114938-oracle-patch-november-2026)** reached production — **boring HTTP**, enormous impact.\n\n## Regulatory and industry context\n\nAgentic collections sit under **fair debt**, **telemarketing**, and **financial privacy** rules in multiple jurisdictions. A breach that exposes **call audio** and **AI-generated dispositions** can trigger **contractual penalties** even before **class actions** arrive.\n\nThe case also feeds **[felony-bench](https://explainx.ai/felony-bench)**-style tracking of **agents affecting real third parties**: here the harm is **data exposure**, not an autonomous wire transfer, but **collections agents** can still initiate **outbound contact** when operators misuse restored access.\n\n## Related reading\n\n## Why this belongs on the felony-bench thread\n\nexplainx.ai tracks **agents that harm third parties** on **[felony-bench](https://explainx.ai/felony-bench)**. This incident is **data exposure**, not an autonomous funds transfer, but it shows **agent branding** on **high-risk workflows** without **agent-grade authorization**.\n\nSecurity teams should log it beside **[GitHub and Microsoft's MDASH disclosure pipeline](https://explainx.ai/blog/github-microsoft-mdash-9791-ai-findings-99-reports-november-2026)** as a case where **\"agentic\" marketing** outpaced **API maturity**.\n\n## Related workshops and skills\n\nIf your team is shipping **voice or text agents** into **regulated industries**, pair technical fixes above with **[live agent workshops](https://explainx.ai/workshops)** on **permission models** and **eval harnesses** before you enable **outbound automation**.\n\nEaton closes by noting that **Keycloak client tricks** may apply to **other React apps** facing a **December disclosure**. Even if you do not run Credit Nirvana, **re-test your IAM integration** any time you ship **new agent dashboards** to customer support or collections teams.\n\n*Technical details and CVE identifiers come from Eaton's November 18, 2026 article. Impact scope for specific customers may change as vendors publish advisories.*", "url": "https://wpnews.pro/news/credit-nirvana-cve-2026-110474-how-an-agentic-collections-stack-leaked-millions", "canonical_source": "https://explainx.ai/blog/credit-nirvana-agentic-debt-collection-cve-2026-110474-november-2026?utm_source=rss&utm_medium=feed&utm_campaign=blog", "published_at": "2026-11-18 00:00:00+00:00", "updated_at": "2026-11-18 23:15:21.796558+00:00", "lang": "en", "topics": ["ai-agents", "ai-safety", "ai-ethics"], "entities": ["Credit Nirvana", "Eaton", "CollectionWorkz", "Aditya Birla Capital Finance", "Keycloak", "Amazon Cognito", "CVE-2026-110474", "CVE-2026-110475"], "also_reported_by": [], "takeaway": "Credit Nirvana's login API trusted a userName in the request body while ignoring the unverified Keycloak token in Authorization-Token, so stripping the fake Keycloak headers and sending only the CN-Token GUID granted high-privilege access to over 10,000 accounts and roughly 315,000 call recordings.", "recommendation": {"advice": "Verify the enterprise IAM token server-side on every request and never authorize on a body-supplied userName or application token alone; Credit Nirvana's unverified Keycloak token let a leaked username reach operator-grade access.", "playbook": "agent-security", "playbook_url": "https://wpnews.pro/playbooks/agent-security"}, "alternates": {"html": "https://wpnews.pro/news/credit-nirvana-cve-2026-110474-how-an-agentic-collections-stack-leaked-millions", "markdown": "https://wpnews.pro/news/credit-nirvana-cve-2026-110474-how-an-agentic-collections-stack-leaked-millions.md", "text": "https://wpnews.pro/news/credit-nirvana-cve-2026-110474-how-an-agentic-collections-stack-leaked-millions.txt", "jsonld": "https://wpnews.pro/news/credit-nirvana-cve-2026-110474-how-an-agentic-collections-stack-leaked-millions.jsonld"}}