What this guide is
This guide defines a prompt pack that developers can use to enforce API security intent when working with coding agents, IDE assistants, and automated test generators.
As APIs are increasingly designed, refactored, and tested by agentic systems, security controls must be expressed in a way that machines can consistently apply. Prompts serve as executable security intent. They shape how agents generate code, review changes, and produce tests.
This prompt pack maps directly to the OWASP API Security Top 10 (2023) and is designed to prevent the most common and most exploitable API failures in an automated development environment.
How to use this prompt pack
These prompts can be used in several ways:
- As a system prompt for an IDE coding agent
- As a repository-level security prompt (for example SECURITY_PROMPT.md)
- As input to CI-based code review agents
- As guidance for test-generation agents
- As guardrails for refactoring and migration tasks
Each prompt is intentionally explicit. Do not paraphrase unless you understand the security implications.
Prompt pack structure
Each section includes:
- Security intent
- Recommended usage
- Copy-paste agent prompt
API1: Broken Object Level Authorization (BOLA)
Security intent
Prevent unauthorized access to objects by enforcing authorization at the data access layer.
Recommended usage
Apply this prompt when generating:
- CRUD endpoints
- Repository or ORM access logic
- Authorization refactors
- Multi-tenant APIs
Agent prompt
Always enforce object-level authorization before accessing or modifying any resource.
Authorization checks must occur at the data access layer and must verify ownership or access rights using authenticated user or tenant context.
Do not rely on request paths, client-supplied identifiers, or post-query filtering for authorization decisions.
API2: Broken Authentication
Security intent
Ensure authentication logic is centralized and token validation is complete and consistent.
Recommended usage
Apply when:
- Implementing authentication middleware
- Integrating OAuth or JWT
- Refactoring auth logic across services
Agent prompt
Use centralized authentication and token validation logic.
Validate issuer, audience, expiration, signature, and required claims for every token.
Reject tokens that are expired, malformed, missing claims, or intended for a different audience.
Do not duplicate authentication logic across services.
API3: Broken Object Property Level Authorization (BOPLA)
Security intent
Prevent exposure or modification of unauthorized object properties.
Recommended usage
Apply when:
- Defining request and response schemas
- Implementing serializers
- Handling partial updates or PATCH requests
Agent prompt
Do not expose all object properties by default.
Use role-specific response models and explicit allowlists for writable fields.
Never bind request bodies directly to domain or persistence models.
Separate read and write schemas to enforce property-level authorization.
API4: Unrestricted Resource Consumption
Security intent
Prevent denial of service and cost abuse caused by automated callers.
Recommended usage
Apply when:
- Designing endpoints with heavy computation
- Implementing pagination or batch operations
- Integrating external services
Agent prompt
Assume all APIs will be called by automated systems.
Add rate limits, quotas, payload size limits, pagination constraints, and timeouts to all endpoints.
Guard expensive operations with stricter limits and defensive defaults.
API5: Broken Function Level Authorization (BFLA)
Security intent
Ensure privileged functionality is not accessible without explicit permission checks.
Recommended usage
Apply when:
- Creating admin or internal endpoints
- Adding operational or support APIs
- Refactoring routing logic
Agent prompt
Treat every endpoint as inaccessible by default.
Apply explicit role and permission checks to all functions.
Do not assume access based on route naming, folder structure, or documentation.
Include tests that invoke privileged endpoints as non-privileged users.
API6: Unrestricted Access to Sensitive Business Flows
Security intent
Protect business logic from automation, abuse, and replay.
Recommended usage
Apply when:
- Implementing onboarding, payments, promotions, or reset flows
- Designing multi-step workflows
- Adding new business-critical endpoints
Agent prompt
Identify sensitive business flows and protect them against automation and abuse.
Add flow-specific rate limits, velocity controls, checkpoints, and step-up authentication where appropriate.
Design for abuse scenarios, not just functional correctness.
API7: Server-Side Request Forgery (SSRF)
Security intent
Prevent APIs from making unsafe outbound requests.
Recommended usage
Apply when:
- Fetching URLs
- Integrating webhooks
- Building proxy or connector services
Agent prompt
Never fetch user-supplied URLs without strict validation.
Enforce allowlists for outbound destinations.
Block internal IP ranges and metadata endpoints.
Disable redirect following unless explicitly required and validated.
API8: Security Misconfiguration
Security intent
Eliminate insecure defaults and prevent debug features from leaking into production.
Recommended usage
Apply when:
- Managing environment configuration
- Building deployment pipelines
- Adding feature flags or debug modes
Agent prompt
Use secure configuration defaults.
Do not expose debug, test, or admin features in production.
Treat configuration as code and flag permissive settings during generation and review.
API9: Improper Inventory Management
Security intent
Prevent forgotten or deprecated APIs from remaining exposed.
Recommended usage
Apply when:
- Introducing new APIs
- Versioning endpoints
- Decommissioning features
Agent prompt
Do not introduce new APIs without ownership, documentation, and lifecycle controls.
Track API versions and environments explicitly.
Flag deprecated or unused endpoints for removal and do not leave legacy versions exposed.
API10: Unsafe Consumption of APIs
Security intent
Prevent cascading failures caused by trusting third-party APIs.
Recommended usage
Apply when:
- Integrating SaaS APIs
- Consuming partner services
- Handling external data ingestion
Agent prompt
Treat all external API responses as untrusted input.
Validate response schemas, enforce timeouts, retries, and throttling.
Monitor for abnormal response patterns and handle unexpected data safely.
Recommended baseline system prompt (optional)
You can prepend this to any coding agent session:
You are generating and reviewing API code in an environment where APIs are consumed by automated and agentic systems.
Prioritize security correctness over convenience.
Follow the API Security Prompt Pack and enforce OWASP API Top 10 controls by default.
If a design choice creates ambiguity in authorization, validation, or exposure, choose the safer option and explain the tradeoff.
Final note for developers
Agentic workflows do not remove responsibility from developers. They compress time and amplify impact. Prompts are how developers encode judgment, constraints, and security intent so automation works with them rather than against them.
This prompt pack is not about trusting agents more.
It is about making sure they cannot accidentally do the wrong thing faster.
Take control of your Application and API security
See how Aptori’s award-winning, AI-driven platform uncovers hidden business logic risks across your code, applications, and APIs. Aptori prioritizes the risks that matter and automates remediation, helping teams move from reactive security to continuous assurance.
Request your personalized demo today.
.svg){kind=small}
.svg){kind=small}
