OpenFGA Contextual Tuple Graph Recursion and Depth Limiter

--- description: OpenFGA Contextual Tuple Graph Recursion and Depth Limiter - Enforces strict structural validation, depth bounding, and cycle detection on ad-hoc contextual tuples passed to OpenFGA Check and ListObjects operations. Unbounded or cyclic contextual tuples allow malicious actors to induce catastrophic ReBAC graph expansion, thread starvation, and service-level denial of service via adversarial graph structures. globs: src/auth/**/*.{ts,js,py,go}, src/middleware/**/*.{ts,js,py,go}, src/services/fga/**/*.{ts,js,py,go}, src/lib/fga/**/*.{ts,js,py,go} alwaysApply: true verified_by: "agentrules.org" license: "CC-BY-4.0" source: "https://agentrules.org/rules/openfga-contextual-tuple-recursion-depth-limiter" --- # OpenFGA Contextual Tuple Graph Recursion and Depth Limiter > **Severity**: CRITICAL | **Category**: security > **Anti-Pattern Warning**: Unvalidated Contextual Tuple Injection Causing ReBAC Catastrophic Backtracking and CPU Starvation ## System Constraints You are a Principal Security Architect specializing in Fine-Grained Authorization (ReBAC) with OpenFGA. Mandate strict validation, graph cycle detection, and depth bounding on all client-provided contextual tuples prior to delegating evaluation to OpenFGA instances. Protect evaluation endpoints from catastrophic backtracking, recursive graph depth exhaustion, and thread starvation. ## Strict Guidelines - Engineers MUST ALWAYS enforce a hard upper bound on the number of contextual tuples permitted in a single evaluation request (maximum 20 tuples unless explicitly documented and protected). - Engineers MUST NOT pass raw, unvalidated client-supplied tuples directly into `fgaClient.check()` or `fgaClient.listObjects()` as `contextualTuples`. - Code SHOULD validate contextual tuples as an acyclic directed graph (DAG) prior to FGA submission to preemptively fail closed on intentional recursion loops. - Services MUST enforce strict, non-negotiable request timeouts (maximum 250ms for synchronous authorization calls) using contextual AbortControllers or deadlines. - Components NEVER permit arbitrary relations or wildcards in user-supplied contextual tuples; tuple subjects, objects, and relations MUST conform to a strict, pre-compiled whitelist schema. ## Bad vs Good Implementations ### Adversarial Contextual Tuple Forwarding vs Graph-Bounded FGA Check #### Bad Antipattern (AI Common Hallucination) > **Why it fails**: Directly forwards unvalidated, unbounded client contextual tuples to OpenFGA. An adversary can pass a self-referential or deeply chained set of relations that forces exponential graph expansion, exhausting memory and CPU in the authorization worker pool. ```typescript import { OpenFgaClient, type ClientCheckRequest } from '@openfga/sdk'; const fgaClient = new OpenFgaClient({ apiUrl: process.env.FGA_API_URL, storeId: process.env.FGA_STORE_ID, }); // ANTI-PATTERN: Forwards arbitrary user-controlled contextual tuples directly. // Allows cyclic hierarchies (e.g. A in B, B in A) or deep nesting chains (1000 items) // leading to server thread exhaustion and catastrophic FGA backtracking. export async function evaluateAccess( user: string, relation: string, object: string, untrustedContextTuples: Array<{ user: string; relation: string; object: string }> ): Promise<boolean> { const response = await fgaClient.check({ user, relation, object, contextualTuples: untrustedContextTuples, }); return Boolean(response.allowed); } ``` #### Production Standard (Correct) > **Standard**: Enforces a strict upper bound on contextual tuple count, validates allowed relations against an invariant allowlist, runs pre-flight cycle and depth limit verification, and enforces a deterministic 250ms timeout abort signal to neutralize DoS attack vectors. ```typescript import { OpenFgaClient, type ClientCheckRequest } from '@openfga/sdk'; const fgaClient = new OpenFgaClient({ apiUrl: process.env.FGA_API_URL, storeId: process.env.FGA_STORE_ID, }); const MAX_CONTEXTUAL_TUPLES = 20; const MAX_TRANSITIVE_DEPTH = 3; const ALLOWED_RELATIONS = new Set(['parent', 'viewer', 'editor', 'member']); const CHECK_TIMEOUT_MS = 250; export interface ContextTuple { user: string; relation: string; object: string; } function validateContextualGraph(tuples: ContextTuple[]): void { if (tuples.length > MAX_CONTEXTUAL_TUPLES) { throw new Error(`Contextual tuple count exceeds maximum limit: ${MAX_CONTEXTUAL_TUPLES}`); } const adj = new Map<string, string[]>(); for (const t of tuples) { if (!ALLOWED_RELATIONS.has(t.relation)) { throw new Error(`Disallowed relation in contextual tuple: ${t.relation}`); } const neighbors = adj.get(t.user) ?? []; neighbors.push(t.object); adj.set(t.user, neighbors); } // Detect cycles and depth using DFS const visited = new Set<string>(); const inStack = new Set<string>(); function dfs(node: string, depth: number): void { if (depth > MAX_TRANSITIVE_DEPTH) { throw new Error(`Contextual graph depth exceeds maximum limit of ${MAX_TRANSITIVE_DEPTH}`); } visited.add(node); inStack.add(node); for (const neighbor of adj.get(node) ?? []) { if (inStack.has(neighbor)) { throw new Error(`Cycle detected in contextual tuples: ${node} -> ${neighbor}`); } if (!visited.has(neighbor)) { dfs(neighbor, depth + 1); } } inStack.delete(node); } for (const node of adj.keys()) { if (!visited.has(node)) { dfs(node, 1); } } } export async function evaluateAccess( user: string, relation: string, object: string, rawTuples: ContextTuple[] ): Promise<boolean> { validateContextualGraph(rawTuples); const controller = new AbortController(); const timeout = setTimeout(() => controller.abort(), CHECK_TIMEOUT_MS); try { const response = await fgaClient.check( { user, relation, object, contextualTuples: rawTuples, }, { signal: controller.signal } ); return Boolean(response.allowed); } finally { clearTimeout(timeout); } } ```

Architectural Specification & Invariant Rationale

Architectural Invariant Design: OpenFGA Contextual Tuple Graph Recursion and Depth Limiter

OpenFGA Contextual Tuple Graph Recursion and Depth Limiter defines an industrial-grade behavioral standard for security engineering. Enforces strict structural validation, depth bounding, and cycle detection on ad-hoc contextual tuples passed to OpenFGA Check and ListObjects operations. Unbounded or cyclic contextual tuples allow malicious actors to induce catastrophic ReBAC graph expansion, thread starvation, and service-level denial of service via adversarial graph structures. 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: Unvalidated Contextual Tuple Injection Causing ReBAC Catastrophic Backtracking and CPU Starvation 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	Unvalidated Contextual Tuple Injection Causing ReBAC Catastrophic Backtracking and CPU Starvation	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 enforce a hard upper bound on the number of contextual tuples permitted in a single evaluation request (maximum 20 tuples unless explicitly documented and protected).

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.