AgentRules.org
RulesStacksClientsAuditorPrunerSelf-HealRFCsBenchmarkCLICI/CDAuthorsSubmit
This specification is maintained by the AgentRules Working Group. Have enhancements or want to claim maintainership?Submit Specification →
backendcritical PriorityID: python-313t-fastapi-free-threaded-contention-guardrail

Python 3.13 Free-Threaded FastAPI Thread-Safety & C-Extension Contention Guardrail

Enforces thread-safety invariants, synchronization primitives, and C-extension isolation under Python 3.13 free-threaded (PEP 703) runtimes running ASGI servers like Granian or Uvicorn. Mandates contextvars over thread-locals and eliminates unprotected global state mutations that previously relied on GIL atomicity guarantees.

Prohibited Anti-Pattern: Relying on legacy GIL atomicity for shared mutable state and using threading.local() in free-threaded async workers.

Export Ready-to-Drop Configuration

Select format below to copy code or download file directly.

---
description: Python 3.13 Free-Threaded FastAPI Thread-Safety & C-Extension Contention Guardrail - Enforces thread-safety invariants, synchronization primitives, and C-extension isolation under Python 3.13 free-threaded (PEP 703) runtimes running ASGI servers like Granian or Uvicorn. Mandates contextvars over thread-locals and eliminates unprotected global state mutations that previously relied on GIL atomicity guarantees.
globs: **/*.py
alwaysApply: true
verified_by: "agentrules.org"
license: "CC-BY-4.0"
source: "https://agentrules.org/rules/python-313t-fastapi-free-threaded-contention-guardrail"
---

# Python 3.13 Free-Threaded FastAPI Thread-Safety & C-Extension Contention Guardrail

> **Severity**: CRITICAL | **Category**: backend
> **Anti-Pattern Warning**: Relying on legacy GIL atomicity for shared mutable state and using threading.local() in free-threaded async workers.

## System Constraints
You are a Principal Python Architect specializing in Python 3.13 free-threaded runtimes (python3.13t / PEP 703) and high-concurrency ASGI architectures. Your role is to strictly enforce thread-safety invariants, eliminate data races caused by the removal of the Global Interpreter Lock, and prevent C-extension execution regressions in multi-threaded FastAPI runtimes.

## Strict Guidelines
- Engineers MUST ALWAYS audit whether the GIL is enabled at startup via sys._is_gil_enabled() and raise a configuration warning if unexpected fallback occurs.
- Engineers MUST NEVER rely on dict, list, or set operations being thread-safe or atomic; all shared mutable state MUST be protected by threading.Lock, threading.RLock, or concurrent lock-free data structures.
- Engineers MUST NEVER use threading.local() for request-scoped context in FastAPI; contextvars.ContextVar MUST be used to prevent cross-task and cross-thread context pollution.
- Services MUST verify that all imported C-extensions declare Py_MOD_GIL_NOT_USED support to prevent the runtime from dynamically re-enabling the GIL or crashing due to unsynchronized C-level static globals.
- Async route handlers interacting with blocking CPU operations MUST explicitly use bounded ThreadPoolExecutors configured specifically for free-threaded contention profiles instead of defaulting to uncontrolled unbounded execution.
- Code MUST use read-copy-update (RCU) or fine-grained reader-writer lock patterns for frequently read, infrequently written in-memory caches to prevent thread lock contention under high core counts.
- Dependencies providing stateful middleware MUST ensure per-request instantiations or use thread-safe synchronizers across concurrent async thread loops.

## Bad vs Good Implementations
### Shared State and Context Handling in Free-Threaded FastAPI

#### Bad Antipattern (AI Common Hallucination)
> **Why it fails**: threading.local() fails across async event loop executions across thread pools, causing request ID bleeding. Directly mutating the global dict without a lock assumes GIL serialization; in Python 3.13 free-threaded mode (nogil), concurrent threads mutating a dict simultaneously produce data loss, non-deterministic overwrites, and memory race conditions.
```python
import threading
from fastapi import FastAPI, Request

app = FastAPI()

# ANTI-PATTERN: threading.local leaks state across async worker threads
# and un-synchronized global dict relies on deprecated GIL atomicity.
_request_store = threading.local()
_metrics_cache: dict[str, int] = {}

@app.middleware("http")
async def trace_middleware(request: Request, call_next):
    _request_store.request_id = request.headers.get("X-Request-ID", "unknown")
    response = await call_next(request)
    return response

@app.post("/api/v1/metrics/{key}")
async def increment_counter(key: str):
    # Data race: In Python 3.13t (nogil), in-place mutation of dicts
    # without synchronization causes race conditions or state corruption
    current = _metrics_cache.get(key, 0)
    _metrics_cache[key] = current + 1
    return {"key": key, "count": _metrics_cache[key], "trace": getattr(_request_store, "request_id", None)}
```

#### Production Standard (Correct)
> **Standard**: Replaces threading.local with ContextVar for deterministic, task-isolated request boundaries. Encapsulates mutable shared state within a synchronized class using an explicit threading.Lock, guaranteeing atomic updates under Python 3.13t free-threaded execution.
```python
import sys
import threading
from contextvars import ContextVar
from fastapi import FastAPI, Request

app = FastAPI()

# Check if GIL is truly disabled in Python 3.13t
if hasattr(sys, "_is_gil_enabled") and sys._is_gil_enabled():
    import warnings
    warnings.warn("Runtime running with GIL enabled. Thread-scaling disabled.", RuntimeWarning)

# ContextVar provides isolated task-local storage safe across async threads
request_id_ctx: ContextVar[str] = ContextVar("request_id_ctx", default="unknown")

class SynchronizedMetricsRegistry:
    """Thread-safe metrics cache protected by fine-grained locking."""
    def __init__(self) -> None:
        self._lock = threading.Lock()
        self._cache: dict[str, int] = {}

    def increment(self, key: str) -> int:
        with self._lock:
            val = self._cache.get(key, 0) + 1
            self._cache[key] = val
            return val

metrics_registry = SynchronizedMetricsRegistry()

@app.middleware("http")
async def trace_middleware(request: Request, call_next):
    token = request_id_ctx.set(request.headers.get("X-Request-ID", "unknown"))
    try:
        return await call_next(request)
    finally:
        request_id_ctx.reset(token)

@app.post("/api/v1/metrics/{key}")
async def increment_counter(key: str):
    new_count = metrics_registry.increment(key)
    return {"key": key, "count": new_count, "trace": request_id_ctx.get()}
```
Recommended MCP Servers (mcp-serverlist.com)

Equip agent with system tools to verify code for this stack.

Python Free-Thread Runtime Inspector MCP
Analyzes Python bytecode, GIL status, and module C-extension compatibility for PEP 703 nogil deployments.
Recommended Agent Skills (askills.dev)

Ready-to-use executable skill functions tailored to this stack.

GIL Audit CLI
Static analysis CLI tool scanning dependencies for missing Py_MOD_GIL_NOT_USED flags and unsafe global collections.
Architectural Specification & Invariant Rationale

Architectural Invariant Design: Python 3.13 Free-Threaded FastAPI Thread-Safety & C-Extension Contention Guardrail

Python 3.13 Free-Threaded FastAPI Thread-Safety & C-Extension Contention Guardrail defines an industrial-grade behavioral standard for backend engineering. Enforces thread-safety invariants, synchronization primitives, and C-extension isolation under Python 3.13 free-threaded (PEP 703) runtimes running ASGI servers like Granian or Uvicorn. Mandates contextvars over thread-locals and eliminates unprotected global state mutations that previously relied on GIL atomicity guarantees. 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: Relying on legacy GIL atomicity for shared mutable state and using threading.local() in free-threaded async workers. 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 ModeAI Symptom (The Anti-Pattern)SeverityProduction Impact & Remedy
AI Hallucination & Deprecation DriftRelying on legacy GIL atomicity for shared mutable state and using threading.local() in free-threaded async workers.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 InconsistencyAI 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:

1

Engineers MUST ALWAYS audit whether the GIL is enabled at startup via sys._is_gil_enabled() and raise a configuration warning if unexpected fallback occurs.

Engineers MUST ALWAYS audit whether the GIL is enabled at startup via sys._is_gil_enabled() and raise a configuration warning if unexpected fallback occurs.

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.
Enforcement:Integrated into the Agent Pack rule file. When Cursor, Claude Code, or Codex processes this prompt, it applies this constraint before emitting syntax.
2

Engineers MUST NEVER rely on dict, list, or set operations being thread-safe or atomic

Engineers MUST NEVER rely on dict, list, or set operations being thread-safe or atomic; all shared mutable state MUST be protected by threading.Lock, threading.RLock, or concurrent lock-free data structures.

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.
Enforcement:Integrated into the Agent Pack rule file. When Cursor, Claude Code, or Codex processes this prompt, it applies this constraint before emitting syntax.
3

Engineers MUST NEVER use threading.local() for request-scoped context in FastAPI

Engineers MUST NEVER use threading.local() for request-scoped context in FastAPI; contextvars.ContextVar MUST be used to prevent cross-task and cross-thread context pollution.

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.
Enforcement:Integrated into the Agent Pack rule file. When Cursor, Claude Code, or Codex processes this prompt, it applies this constraint before emitting syntax.
4

Services MUST verify that all imported C-extensions declare Py_MOD_GIL_NOT_USED support to prevent the runtime from dynamically re-enabling the GIL or crashing due to unsynchronized C-level static globals.

Services MUST verify that all imported C-extensions declare Py_MOD_GIL_NOT_USED support to prevent the runtime from dynamically re-enabling the GIL or crashing due to unsynchronized C-level static globals.

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.
Enforcement:Integrated into the Agent Pack rule file. When Cursor, Claude Code, or Codex processes this prompt, it applies this constraint before emitting syntax.
5

Async route handlers interacting with blocking CPU operations MUST explicitly use bounded ThreadPoolExecutors configured specifically for free-threaded contention profiles instead of defaulting to uncontrolled unbounded execution.

Async route handlers interacting with blocking CPU operations MUST explicitly use bounded ThreadPoolExecutors configured specifically for free-threaded contention profiles instead of defaulting to uncontrolled unbounded execution.

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.
Enforcement:Integrated into the Agent Pack rule file. When Cursor, Claude Code, or Codex processes this prompt, it applies this constraint before emitting syntax.
6

Code MUST use read-copy-update (RCU) or fine-grained reader-writer lock patterns for frequently read, infrequently written in-memory caches to prevent thread lock contention under high core counts.

Code MUST use read-copy-update (RCU) or fine-grained reader-writer lock patterns for frequently read, infrequently written in-memory caches to prevent thread lock contention under high core counts.

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.
Enforcement:Integrated into the Agent Pack rule file. When Cursor, Claude Code, or Codex processes this prompt, it applies this constraint before emitting syntax.
7

Dependencies providing stateful middleware MUST ensure per-request instantiations or use thread-safe synchronizers across concurrent async thread loops.

Dependencies providing stateful middleware MUST ensure per-request instantiations or use thread-safe synchronizers across concurrent async thread loops.

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.
Enforcement:Integrated into the Agent Pack rule file. When Cursor, Claude Code, or Codex processes this prompt, it applies this constraint before emitting syntax.
Technical Questions & Answers

Frequently Asked Technical Questions

Common questions developers face when configuring AI coding assistants for this stack:

Q:Why is Python 3.13 Free-Threaded FastAPI Thread-Safety & C-Extension Contention Guardrail critical for AI coding assistants?

Enforces thread-safety invariants, synchronization primitives, and C-extension isolation under Python 3.13 free-threaded (PEP 703) runtimes running ASGI servers like Granian or Uvicorn. Mandates contextvars over thread-locals and eliminates unprotected global state mutations that previously relied on GIL atomicity guarantees. Without explicit behavioral boundaries, AI coding assistants (Cursor, Claude Code, Windsurf, Codex) rely on training data averages, which frequently produce deprecated, inefficient, or vulnerable code patterns.

Q:How does adopting this rule prevent AI hallucinations?

This rule explicitly declares what the AI must avoid (Relying on legacy GIL atomicity for shared mutable state and using threading.local() in free-threaded async workers.) 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.

Open Specification · CC BY 4.0·Free for Personal, Team & Commercial In-Repo Use

Citation & Open Source Attribution Guidelines

This specification is licensed under Creative Commons Attribution 4.0 International (CC BY 4.0). You are completely free to adapt and share this rule in your workspace. If citing, including in public AI benchmark datasets, or redistributing, attribution to AgentRules.org is required.

Related BACKEND Rules

View all rules →
backend
42

Fastify 5 Schema Compilers & Encapsulation Production Standard

Fastify relies on lexical encapsulation contexts to sandbox decorators, hooks, and compiled validation/serialization schemas without memory leaks or cross-tenant bleed. In Fastify 5, misuse of global compilers or failing to encapsulate plugin boundaries causes compiled `fast-json-stringify` serializers and Ajv validators to either crash on duplicate IDs, leak private data across sibling scopes, or bypass JIT compilation optimizations. Production services must enforce strict encapsulation hierarchies and type-safe schema compilation across every routing subtree.

backend
68

Bun Embedded SQLite WAL Mode & Transaction Synchronization Standard

Production database standard for bun:sqlite. Mandates enabling Write-Ahead Logging (PRAGMA journal_mode = WAL), busy timeouts, and using db.transaction() for atomic operations without lock starvation.

backend
68

Drizzle ORM Batch API & Interactive Transaction Isolation Standard

Production database standard for Drizzle ORM (PostgreSQL & MySQL). Mandates using db.batch() over HTTP drivers to prevent round-trip waterfalls, explicit db.transaction() with serializable/read-committed isolation, and strictly avoids unawaited query promises.

1 Rule in Stack

Python