# NodeLoom — Complete Reference

> NodeLoom is the AI agent operations platform. Discover, monitor, and govern AI agents in production.

NodeLoom enables organizations to find every AI agent running in their infrastructure, instrument agents with lightweight SDKs (Python, TypeScript, Java, Go), monitor behavior in real time with anomaly and drift detection, enforce guardrails (keyword, regex, LLM-as-judge, semantic similarity, PII redaction), run adversarial red team security scans, and maintain compliance with automated audit trails and reporting for SOC 2, HIPAA, GDPR, ISO 42001, NIST AI RMF, and PCI-DSS.

NodeLoom is available as a cloud-hosted SaaS platform or as a self-hosted deployment (Docker, Kubernetes, Helm) for air-gapped and on-premises environments.

---

## Features

### Agent Discovery
NodeLoom Agent Discovery automatically identifies AI agents running across an organization's infrastructure without requiring code changes. It scans cloud providers (AWS, GCP, Azure), GitHub repositories, container orchestrators, and MCP gateways to build a complete inventory. An optional eBPF kernel-level agent intercepts LLM API calls at the TLS layer on Linux hosts, providing zero-instrumentation observability for any process making calls to OpenAI, Anthropic, Google, Cohere, Mistral, and other LLM providers.

### Observability SDKs
Four official SDKs (Python, TypeScript, Java, Go) let developers instrument any AI agent built with any framework — LangChain, CrewAI, LlamaIndex, Autogen, or custom code. SDKs use a fire-and-forget architecture: events are queued locally, batched (100 events or 5 seconds), and sent asynchronously with exponential backoff retry. SDKs add less than 1ms overhead per trace. Each SDK includes a full REST API client for programmatic access to workflows, executions, guardrails, evaluations, feedback, and more.

### Real-Time Monitoring
- **Anomaly Detection**: Automatic scoring (0-100) of every execution against learned baselines. Detects duration spikes, token overuse, output size anomalies, and data exfiltration patterns.
- **Drift Detection**: Identifies gradual performance changes over time. Tracks duration drift, token drift, output size drift, and error rate drift with configurable per-team thresholds.
- **Sentiment Tracking**: Analyzes AI response tone over time. Alerts on negative spikes, score drops, and unusual volume.
- **Token Budget Management**: Daily, weekly, and monthly token limits with per-model cost attribution. Warning (80%) and critical (95%) threshold alerts.

### Guardrails
- **Keyword & Regex Filtering**: Block specific terms or patterns in AI inputs/outputs.
- **LLM-as-Judge**: A separate LLM evaluates AI outputs against custom criteria on 5 dimensions (groundedness, relevance, factual accuracy, tone adherence, safety), each scored 1-5.
- **Semantic Similarity**: Cosine similarity against reference embeddings to detect prompt manipulation attempts.
- **PII Redaction**: Automatic masking of emails, SSNs, credit cards, phone numbers, API keys, and JWT tokens.
- **Prompt Injection Detection**: Pattern and keyword scoring to detect instruction override attempts.
- **Custom Rules**: REGEX, KEYWORD_LIST, JAVASCRIPT, and LLM_PROMPT rule types with WARNING, BLOCK, or LOG severity levels.

### Compliance & Audit
- **Cryptographic Audit Trail**: SHA-256 hash chain for tamper detection with forensic verification.
- **Compliance Reports**: One-click generation for SOC 2, ISO 42001, NIST AI RMF, GDPR, HIPAA, and SOX.
- **Data Retention Policies**: Configurable per data type with automatic purge.
- **SIEM Integration**: Export to Splunk HEC, Datadog Logs, Elasticsearch, or custom webhooks.
- **RBAC**: 5 roles (Admin, Builder, Operator, Viewer, Compliance Officer) with granular permissions.

### Adversarial Testing (Red Team)
Automated security scans test AI agent defenses. An attacker LLM generates targeted attacks across 6 categories (prompt injection, jailbreak, data exfiltration, tool abuse, PII leakage, harmful output), sends them to the target agent, and a judge LLM evaluates whether each attack succeeded. Results include a resilience score (1.00-5.00) and per-finding severity ratings.

### Incident Response Playbooks
Map detection events (guardrail violations, anomalies, drift alerts, evaluation failures) to automated response workflows. Configurable severity filtering, cooldown windows, and enable/disable toggles. Response workflows use the full workflow builder capabilities (Slack, Jira, Email, HTTP, conditionals).

### Multi-Environment Deployments
Promotion pipeline: Development → Staging → Production. Each environment runs its own definition snapshot, activation state, webhook token, schedule, and credentials. Enterprise only.

### Enterprise Security
- AES-256-GCM encryption for all credentials at rest
- Mandatory Docker sandboxing for code execution
- SAML 2.0 / OIDC single sign-on
- SCIM 2.0 provisioning (compatible with Okta, Azure AD, OneLogin)
- Per-tier API rate limiting
- HTTP security headers (HSTS, CSP, X-Frame-Options, X-Content-Type-Options)

---

## Pricing

### Team — $500/month
- Agents: 10
- Executions: 10,000/month
- Team members: 5
- Credentials: 20
- Free trial: 14-day
- Key features: SDK access (all languages), Behavioral monitoring, Basic guardrails (keyword, regex, reject), Anomaly detection, Duration & error rate drift, Audit logs (30 days, view only), Webhooks, Basic analytics, Email support

### Business — $1500/month
- Agents: 50
- Executions: 100,000/month
- Team members: 25
- Credentials: Unlimited
- Free trial: 14-day
- Key features: SSO (SAML 2.0, OIDC), SCIM 2.0 provisioning, Custom RBAC roles, Compliance dashboard & reports, Cryptographic audit trail, Exportable audit logs (1 year), Advanced guardrails (LLM-as-judge), Semantic guardrails & embeddings, Sentiment analysis, Full drift detection suite, Incident response playbooks, Agent discovery & inventory, Priority support

### Enterprise — Starting $25,000/year
- Agents: 500+
- Executions: Unlimited/month
- Team members: Unlimited
- Credentials: Unlimited
- Key features: LLM-as-judge evaluation, SIEM integration (Splunk, Datadog, Sentinel), Self-hosted & air-gapped deployment, Red team security scanning, Multi-environment (Dev/Staging/Prod), Unlimited audit log retention, Custom compliance frameworks, Dedicated account manager, Onboarding & training, Custom contracts & invoicing, Data residency controls, Early access to new features, Dedicated support channel

---

## Integrations (97+)

- **AI & Machine Learning**: OpenAI, Anthropic, Google Gemini, Ollama, Hugging Face, Replicate
- **Cloud & Infrastructure**: AWS S3, Google Cloud, Azure, Docker, Kubernetes, Cloudflare
- **CRM & Sales**: Salesforce, HubSpot, Pipedrive, Zoho CRM, Freshsales, Close
- **DevOps & Development**: GitHub, GitLab, Bitbucket, Jira, Linear, Jenkins
- **Databases**: PostgreSQL, MySQL, MongoDB, Redis, Elasticsearch, Supabase
- **Communication**: Slack, Discord, Microsoft Teams, Twilio, SendGrid, Mailchimp
- **Productivity**: Google Sheets, Notion, Airtable, Todoist, Trello, Asana

---

## Industry Solutions

### Financial Services
Every bank, asset manager, and fintech has the same problem: critical compliance processes running on spreadsheets, email chains, and manual copy-paste between systems. A single missed AML alert can mean millions in fines. A broken reconciliation workflow means your ops team works through the weekend. NodeLoom replaces these fragile manual processes with automated workflows that log every step, flag every exception, and produce audit-ready reports on demand.

**Use Cases:**
  - KYC/AML Screening Pipeline: New client application triggers the workflow. NodeLoom pulls identity data, screens against OFAC/SDN sanctions lists and PEP databases via API, runs through your internal risk scoring model, and routes the result. Clean applications get auto-approved with documentation. Flagged applications go to an analyst queue with all supporting evidence pre-attached. Every decision is logged.
  - SOX Control Testing Automation: On a schedule (daily, weekly, or quarterly), NodeLoom pulls control evidence from source systems (general ledger entries, access logs, approval records), validates it against your testing criteria, and generates a formatted report. Exceptions are flagged and routed to the control owner. No more copy-paste from SAP into Excel.
  - End-of-Day Trade Reconciliation: After market close, NodeLoom pulls trade records from your OMS, matches them against broker confirmations and custodian statements, and identifies breaks by type (price, quantity, settlement date, counterparty). Matched trades are auto-confirmed. Breaks are categorized, assigned, and escalated based on age and materiality.
  - Transaction Monitoring Alert Triage: When your TMS flags a transaction, NodeLoom enriches the alert with customer history, related accounts, and transaction patterns. AI agents score the alert against your typology models. Low-risk alerts get documented and closed with a rationale. Medium and high-risk alerts go to investigators with a pre-built case file, reducing investigation time from hours to minutes.
  - Regulatory Filing Assembly: For recurring filings (CCAR, FR Y-14, Call Reports), NodeLoom collects data from source systems, applies validation rules and cross-checks, formats output to the regulator's specifications, and generates a review package. The workflow tracks who reviewed what, when approvals were given, and maintains a complete audit trail for examiner requests.
  - Credit Decisioning Pipeline: Loan application comes in. NodeLoom pulls bureau data, runs it through your credit scorecard, checks concentration limits, and applies your pricing matrix. Applications within policy are auto-decisioned with terms. Exceptions (overrides, policy exceptions) route to the appropriate approval authority with all supporting data attached.

### Healthcare
In most healthcare organizations, the "integration" between systems is a person. Someone copies a lab result from the EHR into a fax. Someone re-keys a prior auth into a payer portal. Someone manually checks if a referral was received. These handoffs are where errors happen, where data gets lost, and where your staff burns out. NodeLoom connects your systems with automated workflows so your people can focus on patients instead of data entry.

**Use Cases:**
  - Prior Authorization Automation: Provider submits a prior auth request from the EHR. NodeLoom checks patient eligibility against the payer, attaches required clinical documentation from the patient's chart, formats the request per the payer's requirements, submits it electronically, and monitors for the response. Approvals update the EHR automatically. Denials route to the appeals team with the original documentation attached.
  - Referral Coordination: When a provider creates a referral, NodeLoom sends the referral to the specialist, attaches relevant clinical records, verifies the patient's insurance covers the specialist, schedules a follow-up check, and closes the loop when the specialist's notes come back. No more calling the specialist office to ask "did you get the referral?"
  - Claims Scrubbing Before Submission: Before a claim goes to the clearinghouse, NodeLoom validates it against a rules engine: correct CPT/ICD-10 combinations, valid NPI numbers, patient eligibility status, timely filing limits, and payer-specific requirements. Clean claims go straight through. Flagged claims route to the billing team with the specific errors highlighted, so they fix once instead of reworking after denial.
  - Lab Result Routing: Lab results arrive via HL7 feed or API. NodeLoom parses the result, matches it to the ordering provider and patient, checks for critical values that need immediate physician notification, files the result in the EHR, and sends the patient a notification through your engagement platform. Critical values trigger an immediate page to the ordering physician with escalation if not acknowledged.
  - Patient Scheduling and Reminders: New appointment booked. NodeLoom sends a confirmation, checks if pre-visit paperwork is complete, sends reminders at configurable intervals (7 days, 2 days, morning of), and collects patient intake forms digitally. No-shows trigger a rebooking outreach. All touchpoints are logged for the patient record.
  - PHI De-identification for Research: When clinical data needs to be shared with a research team or external partner, NodeLoom applies automated de-identification rules (Safe Harbor or Expert Determination method), strips the 18 HIPAA identifiers, generates a de-identification log, and routes the scrubbed dataset for review before release. The original data never leaves the secure environment.

### Insurance
The average property casualty claim takes 30+ days to close. Not because the work is that complex, but because information moves through email, gets re-keyed into three different systems, and sits in someone's queue waiting for a follow-up that should have been automatic. Your adjusters spend more time chasing documents and updating systems than actually adjusting claims. NodeLoom automates the handoffs, the data gathering, and the status updates so your team can focus on the decisions that actually require expertise.

**Use Cases:**
  - FNOL Intake and Triage: Policyholder reports a claim (web form, call center, agent portal). NodeLoom captures the information, validates coverage, assigns a claim number, routes to the right adjuster based on line of business and severity, sends acknowledgment to the policyholder, and creates tasks for initial documentation requests. A simple auto/homeowners claim that used to take 2 hours of manual setup now takes 2 minutes.
  - Underwriting Data Assembly: New submission comes in from a broker. NodeLoom pulls the applicant's loss history from ISO/A-PLUS, financial data from D&B, property data from CoreLogic, and prior claims from your own system. All of this is assembled into a structured underwriting workbook and delivered to the underwriter, who can now spend their time on risk analysis instead of data collection.
  - Policy Issuance and Servicing: Underwriter binds coverage. NodeLoom generates the policy document from your templates, sends it to the insured, sets up billing (direct or agency bill), creates the policy record in your admin system, and schedules renewal processing. Endorsements and cancellations follow the same automated path with appropriate approval gates.
  - Regulatory Filing Preparation: When a filing deadline approaches, NodeLoom pulls the required data from your policy admin, claims, and financial systems, validates it against the filing's specifications, flags discrepancies for review, and formats the output for submission. NAIC annual statement data, state rate filings, and market conduct data calls all follow the same pattern.
  - Policyholder Onboarding: New policyholder. NodeLoom sends a welcome package, sets up online account access, collects additional information needed for coverage (vehicle schedules, property photos, employee census), confirms receipt, and flags incomplete submissions. The agent gets a dashboard showing which clients have completed onboarding and which need follow-up.
  - Claims Reserve and Severity Scoring: When a new claim is opened or updated, NodeLoom runs it through your reserving methodology: injury type, jurisdiction, attorney involvement, prior claim history, and comparable claim outcomes. The suggested reserve is attached to the claim record for adjuster review. If the reserve exceeds authority limits, it auto-escalates to a supervisor.

### Legal
Legal teams are expensive. Yet a significant portion of legal work is not actually legal analysis. It is tracking deadlines in spreadsheets, copying contract terms into databases, chasing counterparties for signatures, and manually checking whether new regulations affect existing obligations. These tasks do not require a law degree. They require a workflow. NodeLoom automates the operational side of legal work so your team can focus on the judgment calls that actually need them.

**Use Cases:**
  - Contract Intake and Clause Extraction: A new contract arrives (email attachment, upload portal, DocuSign). NodeLoom extracts key terms using AI: effective date, termination provisions, indemnification caps, liability limitations, IP ownership, governing law, and renewal terms. These are compared against your standard playbook. Deviations are flagged with the specific clause highlighted. Standard agreements can be approved automatically. Non-standard terms route to the appropriate attorney with a redline summary.
  - Deadline Monitoring and Escalation: NodeLoom maintains a master calendar of all critical dates from your contracts and matters: renewal deadlines, option exercise windows, filing deadlines, discovery due dates, and compliance milestones. Reminders fire at configurable intervals (90 days, 30 days, 7 days, day-of). If the responsible person has not acknowledged, it escalates to their manager. If the deadline is missed, it logs an incident.
  - Regulatory Change Monitoring: NodeLoom monitors regulatory feeds and, when a relevant change is published, identifies which of your contracts, policies, or procedures may be affected based on subject matter, jurisdiction, and entity type. It creates an impact assessment task with the regulatory text, the potentially affected items, and a suggested review priority. Your compliance team gets a structured work queue instead of an unfiltered news feed.
  - Matter Intake and Conflict Check: New matter request comes in. NodeLoom captures the matter details, runs a conflict check against your client and adverse party database, identifies potential conflicts, and routes them for review. Clean matters are opened automatically with the correct billing codes, team assignments, and client communication templates. Conflicts go to the ethics partner with all relevant relationships documented.
  - Document Generation from Templates: NDAs, engagement letters, board resolutions, and standard agreements all follow templates. NodeLoom populates templates with client data, deal terms, and jurisdiction-specific language, generates the document in the correct format (Word or PDF), routes it for internal review, and sends it for signature. Tracked versions, not email attachments.
  - Litigation Hold Management: When litigation is anticipated or filed, NodeLoom identifies custodians based on the subject matter, sends hold notices with acknowledgment tracking, monitors for compliance, sends reminders to non-responsive custodians, and maintains a defensible log of the entire process. When the hold is released, it notifies all custodians and updates the matter record.

### IT & DevOps
Your team has runbooks for every incident type. They live in Confluence or a wiki somewhere. When an alert fires at 3 AM, an engineer wakes up, opens the runbook, and follows the steps manually: check this dashboard, run this query, restart this service, update this ticket. Every step is documented. None of them are automated. NodeLoom turns your existing runbooks into executable workflows that trigger automatically when alerts fire, run diagnostics before a human even opens their laptop, and escalate only when human judgment is actually needed.

**Use Cases:**
  - Automated Incident Response: PagerDuty alert fires. NodeLoom immediately runs the diagnostic steps from your runbook: checks service health endpoints, queries recent deployments, pulls error logs from the relevant time window, checks for correlated alerts, and creates a Jira incident with all diagnostics attached. If the issue matches a known pattern (disk full, certificate expired, OOM kill), it executes the remediation automatically. If not, it pages the on-call engineer with all the context already gathered.
  - Alert Deduplication and Enrichment: Multiple monitoring tools fire alerts for the same underlying issue. NodeLoom deduplicates them using a correlation window, enriches the combined alert with context from Datadog metrics, CloudWatch logs, and your CMDB (which service, which team owns it, what changed recently), and routes a single, enriched notification to the right team. One alert, not twelve.
  - Deployment Pipeline with Approval Gates: PR merged to main. NodeLoom triggers the build, runs the test suite, deploys to staging, runs smoke tests. If smoke tests pass, it sends an approval request to the designated reviewer (Slack, email). Approved? Deploy to production, run health checks, notify the team. Failed health check? Auto-rollback and page the engineer. Every step is logged for your SOC 2 change management evidence.
  - Executable Runbooks: Take your existing Confluence runbook for "database failover" or "certificate renewal" and model it as a NodeLoom workflow. Manual steps become automated actions. Decision points become conditional branches. Human judgment steps become approval gates. The runbook now runs on trigger (scheduled or alert-based) instead of requiring someone to find and follow the documentation.
  - Infrastructure Health Dashboard Feeds: Every 5 minutes, NodeLoom polls your services' health endpoints, aggregates the results with metrics from your monitoring stack, computes a service-level health score, and pushes the results to your status page or dashboard. If a service degrades below threshold, it triggers the incident response workflow automatically.
  - Change Management Logging: Every deployment, configuration change, and infrastructure modification is captured as a change record: who requested it, who approved it, what changed, when it was applied, and whether the post-change verification passed. Your SOC 2 auditor gets a clean change log without anyone having to manually update a spreadsheet.

---

## Frequently Asked Questions

### Can I switch plans at any time?
Yes, you can upgrade or downgrade your plan at any time. When upgrading, the new features are available immediately. When downgrading, changes take effect at the end of your current billing cycle.

### What happens when I exceed my execution limit?
You'll receive a notification when you reach 80% of your limit. Once reached, workflows will pause until the next billing cycle or until you upgrade your plan.

### Do you offer a self-hosted option?
Yes! NodeLoom can be self-hosted on your own infrastructure with a one-time license key. This gives you full control over your data and unlimited executions. Self-hosted deployment is available on the Enterprise plan.

### What AI providers are supported?
NodeLoom supports OpenAI, Anthropic (Claude), Google Gemini, and Ollama for local models. You bring your own API keys, and we never access your AI credentials.

### Can I try NodeLoom before committing?
Both the Team and Business plans include a 14-day free trial so you can evaluate the platform with your team. No commitment required. For Enterprise, we offer personalized demos tailored to your use case.

### Is my data secure?
Absolutely. All credentials are encrypted at rest and in transit using industry-standard encryption. Code execution runs in fully isolated sandboxes. We provide RBAC, comprehensive audit logging, and enterprise-grade security controls.

### Can I monitor agents I didn't build in NodeLoom?
Yes. Our observability SDKs (Python, TypeScript, Java, Go) let you instrument any AI agent, whether built with LangChain, CrewAI, or a custom framework. SDK-instrumented agents get the same monitoring, drift detection, and compliance tracking as native workflows.

### What is agent discovery?
Agent Discovery automatically finds AI agents running across your infrastructure. It scans cloud providers (AWS, GCP, Azure), GitHub repositories, and MCP gateways to build a complete inventory of your AI agents, even ones your team didn't know existed. Available on Business and Enterprise plans.

### What kind of support do you offer?
Team plans include email support. Business plans get priority support with 8-hour response times. Enterprise plans include a dedicated account manager with a dedicated Slack channel.

### What security testing is included?
Business plans include LLM-as-Judge evaluation for scoring agent outputs against custom criteria. Enterprise plans add red team adversarial testing, which runs automated attacks (prompt injection, jailbreak, data exfiltration) against your AI agents and generates vulnerability reports.

### What are incident response playbooks?
Business and Enterprise plans include automated incident response playbooks that trigger when drift thresholds or guardrail violations are detected. Playbooks can quarantine affected workflows, notify team members via email or Slack, escalate to on-call staff, and automatically rollback to the last known-good version.

### What is AI agent governance?
AI agent governance is the practice of establishing controls, policies, and oversight mechanisms for AI agents operating in production environments. It encompasses monitoring agent behavior, enforcing guardrails on inputs and outputs, maintaining audit trails, ensuring compliance with regulations, and testing agents for security vulnerabilities. NodeLoom provides a comprehensive AI agent governance platform that covers discovery, monitoring, guardrails, compliance automation, and adversarial testing.

### What is AI observability for AI agents?
AI observability goes beyond traditional monitoring by providing deep visibility into how AI agents behave, reason, and make decisions. It includes tracking token usage, monitoring output quality through LLM-as-judge evaluation, detecting behavioral drift over time, analyzing sentiment in responses, and tracing the full execution path of agent workflows. NodeLoom provides AI observability through SDKs for Python, TypeScript, Java, and Go, plus zero-instrumentation eBPF probes for kernel-level monitoring.

### How do you monitor AI agents in production?
Monitoring AI agents in production requires instrumenting agents with observability SDKs, establishing behavioral baselines, detecting anomalies and drift, tracking token usage and costs, and setting up alerting. NodeLoom automates this with lightweight SDKs that add less than 1ms overhead, automatic baseline learning, configurable anomaly and drift thresholds, and incident response playbooks that trigger automated actions when issues are detected.

### What are AI guardrails?
AI guardrails are safety controls that validate AI agent inputs and outputs to prevent harmful, non-compliant, or unintended behavior. Types of guardrails include keyword and regex filtering, LLM-as-judge evaluation (where a separate AI scores outputs on dimensions like safety and accuracy), semantic similarity detection (comparing against known-bad patterns), PII redaction, prompt injection detection, and custom rules. NodeLoom supports all these guardrail types with configurable severity levels (WARNING, BLOCK, LOG).

### How do you discover shadow AI agents in your organization?
Shadow AI refers to AI agents deployed by teams without centralized visibility or governance. NodeLoom Agent Discovery automatically finds these agents by scanning cloud providers (AWS, GCP, Azure), GitHub repositories, container orchestrators, and MCP gateways. For Linux hosts, an optional eBPF kernel-level agent intercepts LLM API calls at the TLS layer, detecting any process communicating with OpenAI, Anthropic, Google, or other LLM providers — without requiring code changes.

### What is the difference between AI monitoring and AI governance?
AI monitoring focuses on observability: tracking what AI agents are doing, detecting anomalies, and alerting on issues. AI governance is broader — it includes monitoring plus enforcement of policies, compliance automation, access controls, audit trails, adversarial testing, and incident response. NodeLoom is an AI governance platform that includes monitoring as a core capability alongside guardrails, compliance dashboards, red team testing, RBAC, and automated playbooks.

### What tools support self-hosted AI governance?
NodeLoom offers full self-hosted deployment on your own infrastructure using Docker Compose, Kubernetes, or Helm charts. Self-hosted deployments include all platform features (monitoring, guardrails, compliance, red team testing) and support air-gapped environments with no external data transmission. This is critical for organizations in regulated industries (healthcare, financial services, defense) that require data sovereignty.

### How do you test AI agents for security vulnerabilities?
NodeLoom provides automated red team adversarial testing. An attacker LLM generates targeted attacks across 6 categories — prompt injection, jailbreak, data exfiltration, tool abuse, PII leakage, and harmful output. Each attack is sent to the target AI agent, and a judge LLM evaluates whether the attack succeeded. Results include a resilience score (1.00-5.00), per-finding severity ratings, and remediation recommendations. Critical findings automatically trigger incident response playbooks.

### What compliance frameworks does NodeLoom support?
NodeLoom supports compliance reporting for SOC 2, HIPAA, GDPR, ISO 42001, NIST AI RMF, and PCI-DSS. The platform provides a compliance dashboard showing guardrail coverage, audit trail integrity, credential health, and user access patterns. One-click report generation creates framework-specific compliance reports. The cryptographic audit trail uses SHA-256 hash chaining for tamper-proof event logs.

### Can NodeLoom monitor LangChain and CrewAI agents?
Yes. NodeLoom SDKs include built-in integrations for LangChain and CrewAI. The Python SDK provides a LangChain callback handler and CrewAI decorator that automatically capture traces, spans, token usage, and tool calls. The TypeScript SDK includes a LangChain callback handler. Any framework can be instrumented manually using the trace/span API. SDK-instrumented agents get full monitoring, drift detection, evaluation, and compliance tracking.

---

## Company

NodeLoom was founded by Reda Zerrad, a Director of Engineering with 15+ years of experience at Atlassian, Banyan Security, American Express, and Lookout. The company is based in San Francisco, California.

- Website: https://nodeloom.io
- Email: contact@nodeloom.io
- Phone: (415) 340-1981
- Address: 1875 Mission St Ste 103 #196, San Francisco, CA 94103

## Links

- Home: https://nodeloom.io
- Features: https://nodeloom.io/features
- Pricing: https://nodeloom.io/pricing
- Security: https://nodeloom.io/security
- Solutions: https://nodeloom.io/solutions
- Financial Services: https://nodeloom.io/solutions/financial-services
- Healthcare: https://nodeloom.io/solutions/healthcare
- Insurance: https://nodeloom.io/solutions/insurance
- Legal: https://nodeloom.io/solutions/legal
- IT & DevOps: https://nodeloom.io/solutions/it-devops
- Demo: https://nodeloom.io/demo
- Contact: https://nodeloom.io/contact
- Documentation: https://app.nodeloom.io/docs
- SDK Reference: https://app.nodeloom.io/docs/sdks/overview
- API Reference: https://app.nodeloom.io/docs/api-reference