OpenTelemetry W3C Baggage Isolation and Metric Cardinality Guardrail

Q: How does adopting this rule prevent AI hallucinations?

This rule explicitly declares what the AI must avoid (Blind W3C Baggage Propagation and High-Cardinality Dimension Injection into Metrics) and prescribes verified positive examples. Grounding the LLM with negative and positive pairs eliminates guesswork.

Q: Which AI coding environments are supported?

This specification is multi-platform compatible: Cursor (.cursor/rules/*.mdc & .cursorrules), Anthropic Claude Code (CLAUDE.md), Windsurf (.windsurfrules), Codex / Cline (AGENTS.md), and GitHub Copilot (.github/copilot-instructions.md).

Q: Can I combine this rule with other technology stacks?

Yes. Use the Interactive Stack Builder or download an Agent Pack to merge this rule with other frontend, backend, database, security, and DevOps rules without syntax collision.

OpenTelemetry W3C Baggage Isolation and Metric Cardinality Guardrail

--- description: OpenTelemetry W3C Baggage Isolation and Metric Cardinality Guardrail - Distributed tracing setups that blindly forward W3C Baggage across perimeter boundaries leak sensitive PII downstream and trigger metric cardinality denial-of-service. This guardrail mandates strict allowlisting for baggage propagation across services, completely decouples baggage from metric dimensions, and enforces bounded cardinality controls at span and metric boundaries. globs: src/**/*.ts, src/**/*.js, lib/**/*.ts, instrumentation/**/*.ts alwaysApply: true verified_by: "agentrules.org" license: "CC-BY-4.0" source: "https://agentrules.org/rules/opentelemetry-baggage-isolation-cardinality-guardrail" --- # OpenTelemetry W3C Baggage Isolation and Metric Cardinality Guardrail > **Severity**: CRITICAL | **Category**: devops > **Anti-Pattern Warning**: Blind W3C Baggage Propagation and High-Cardinality Dimension Injection into Metrics ## System Constraints You are a Principal Observability and SRE Architect enforcing OpenTelemetry W3C Baggage Isolation and Metric Cardinality Guardrails. You must prevent telemetry cost blowouts, backend TSDB memory exhaustion, and sensitive data leakage via tracing baggage. Strictly enforce allowlists on baggage propagation, prohibit raw baggage keys from becoming metric labels, and mandate bounded cardinality bucketing. ## Strict Guidelines - Engineers MUST ALWAYS sanitize and filter W3C Baggage using an explicit organizational allowlist before propagating context downstream. - Engineers MUST NEVER inject dynamic or arbitrary W3C Baggage entries directly into Metric instrument attributes. - Engineers MUST validate that metric attributes have bounded cardinality (<100 distinct values); high-cardinality identifiers (UUIDs, user IDs, emails, raw paths) MUST only exist as trace span attributes or structured logs. - Services MUST strip incoming W3C Baggage headers at untrusted edge boundaries unless explicitly verified and prefixed (e.g., `x-org-*`). - Components MUST NOT bind un-sanitized baggage values to external HTTP client headers going to third-party or untrusted APIs. - Instrumentation code SHOULD configure OpenTelemetry View attribute filters to drop unexpected dimensions before TSDB export. ## Bad vs Good Implementations ### Sanitizing Outbound Baggage and Safeguarding Metric Cardinality #### Bad Antipattern (AI Common Hallucination) > **Why it fails**: Directly mapping high-cardinality fields (transactionId, userEmail, IP) to metric attributes explodes metric cardinality in systems like Prometheus/Datadog, leading to massive cloud bills or TSDB crash. Passing raw baggage entries without allowlisting exposes PII to all downstream microservices and third parties. ```typescript import { baggage, metrics, context } from '@opentelemetry/api'; const meter = metrics.getMeter('order-service'); const checkoutCounter = meter.createCounter('checkout_requests_total'); export async function processCheckout(req: { headers: Record<string, string>; userId: string }) { // VULNERABILITY 1: Unsanitized baggage reading and blind addition const currentBaggage = baggage.getBaggage(context.active()); const userEmail = currentBaggage?.getEntry('user.email')?.value; const transactionId = req.headers['x-transaction-id']; // VULNERABILITY 2: Telemetry cardinality explosion & PII leakage into TSDB labels checkoutCounter.add(1, { 'user.email': userEmail || 'unknown', // Unbounded PII 'transaction.id': transactionId, // Unbounded UUID -> TSDB OOM 'client.ip': req.headers['x-forwarded-for'], }); // VULNERABILITY 3: Blindly forwarding all baggage downstream without filtering // Outbound HTTP requests automatically forward raw baggage to third-party APIs } ``` #### Production Standard (Correct) > **Standard**: High-cardinality attributes (transaction ID, user ID) are attached to Spans rather than Metrics, preserving debuggability without blowing up TSDB time-series cardinality. Inbound baggage is strictly allowlisted against approved business keys, preventing accidental downstream PII leakage. ```typescript import { baggage, metrics, trace, context } from '@opentelemetry/api'; const meter = metrics.getMeter('order-service'); const tracer = trace.getTracer('order-service'); // Cardinality-bounded metric: only predefined, finite categorical values const checkoutCounter = meter.createCounter('checkout_requests_total', { description: 'Total checkouts partitioned by customer tier and status', }); const ALLOWED_BAGGAGE_KEYS = new Set(['app.tenant.tier', 'app.experiment.cohort']); const VALID_TIERS = new Set(['free', 'starter', 'enterprise']); export async function processCheckout(req: { headers: Record<string, string>; userId: string }) { const activeCtx = context.active(); const currentBaggage = baggage.getBaggage(activeCtx); // 1. High-cardinality data belongs strictly on Tracing Spans, NEVER on Metrics const span = tracer.startSpan('processCheckout', undefined, activeCtx); span.setAttribute('order.transaction_id', req.headers['x-transaction-id'] ?? 'unknown'); span.setAttribute('user.id', req.userId); // 2. Bound metric attributes to finite categorical values const rawTier = currentBaggage?.getEntry('app.tenant.tier')?.value ?? 'free'; const validatedTier = VALID_TIERS.has(rawTier) ? rawTier : 'unknown'; checkoutCounter.add(1, { 'tenant.tier': validatedTier, 'operation.status': 'success', }); // 3. Construct an isolated, sanitized Baggage container for downstream propagation let isolatedBaggage = baggage.createBaggage(); if (currentBaggage) { for (const [key, entry] of currentBaggage.getAllEntries()) { if (ALLOWED_BAGGAGE_KEYS.has(key)) { isolatedBaggage = isolatedBaggage.setEntry(key, entry); } } } span.end(); return baggage.setBaggage(activeCtx, isolatedBaggage); } ```

Architectural Specification & Invariant Rationale

Architectural Invariant Design: OpenTelemetry W3C Baggage Isolation and Metric Cardinality Guardrail

OpenTelemetry W3C Baggage Isolation and Metric Cardinality Guardrail defines an industrial-grade behavioral standard for devops engineering. Distributed tracing setups that blindly forward W3C Baggage across perimeter boundaries leak sensitive PII downstream and trigger metric cardinality denial-of-service. This guardrail mandates strict allowlisting for baggage propagation across services, completely decouples baggage from metric dimensions, and enforces bounded cardinality controls at span and metric boundaries. Modern software architectures require strict adherence to deterministic patterns to guarantee high availability, maintainability, and zero security vulnerabilities.

Framework Paradigm Shift & Drift MechanicsLLMs are trained on millions of open-source repositories containing outdated code, legacy idioms, and unpatched antipatterns. When coding assistants generate implementations for this stack, they default to common statistical tokens: Blind W3C Baggage Propagation and High-Cardinality Dimension Injection into Metrics Without an explicit system instruction file, developers spend hours debugging silent failures.

Failure Modes & Production Blast Radius

Hazard Analysis: What Breaks Without This Rule

The following failure modes have been cataloged from hundreds of hours of production code reviews where AI assistants produced non-compliant implementations:

Failure Mode	AI Symptom (The Anti-Pattern)	Severity	Production Impact & Remedy
AI Hallucination & Deprecation Drift	Blind W3C Baggage Propagation and High-Cardinality Dimension Injection into Metrics	Critical	Violating this industrial standard introduces regression bugs, security risks, and unexpected runtime failures in production. Enforce explicit behavioral system directives across all AI coding assistants in your repository.
Silent Architecture Inconsistency	AI assistants produce mixed paradigms across files, creating fragmented codebase conventions.	Medium	Increases cognitive load during code reviews and leads to maintenance debt across engineering teams. Pin project conventions in .cursor/rules, CLAUDE.md, and AGENTS.md so all team agents follow identical guardrails.

Architectural Directives in Detail

Step-by-Step Production Directives

Each rule constraint below is injected into the AI agent's system prompt to establish deterministic boundaries:

Engineers MUST ALWAYS sanitize and filter W3C Baggage using an explicit organizational allowlist before propagating context downstream.

Rationale:AI models frequently generate unconstrained code here because statistical language distributions lack project-specific architectural context. Mandating this directive ensures that generated code adheres to verified industrial production requirements.