The Fix
pip install pydantic==2.11.8
Based on closed pydantic/pydantic issue #11705 · PR/commit linked
Production note: This usually shows up under retries/timeouts. Treat it as a side-effect risk until you can verify behavior with a canary + real traffic.
@@ -501,3 +501,49 @@ args, kwargs = val.validate_json('{"args": ["arg1"], "kwargs": {"extra": 1}}')
#> ('arg1',) {'extra': 1}
```
+
+## `MISSING` sentinel
+
repro.py
from typing import TypedDict, NotRequired
from pydantic import TypeAdapter
class RequestDict(TypedDict, total=False):
model: str
temperature: NotRequired[float]
TypeAdapter(RequestDict).validate_python({"model": "gpt-4"})
verify
Re-run the minimal reproduction on your broken version, then apply the fix and re-run.
fix.md
Option A — Upgrade to fixed release\npip install pydantic==2.11.8\nWhen NOT to use: This fix is not suitable for fields where null is a valid input.\n\n
Why This Fix Works in Production
- Trigger: - Optional to include (i.e., the client may omit them entirely),
- Mechanism: Introduced an experimental `MISSING` sentinel to handle optional fields that must not be null if provided
- Why the fix works: Introduced an experimental `MISSING` sentinel in Pydantic to handle optional fields that must not be null if provided. (first fixed release: 2.11.8).
Production impact:
- If left unfixed, retries/timeouts can trigger duplicate external side-effects (double charges, duplicate emails, repeated writes).
Why This Breaks in Prod
- Introduced an experimental `MISSING` sentinel to handle optional fields that must not be null if provided
- Production symptom (often without a traceback): - Optional to include (i.e., the client may omit them entirely),
Proof / Evidence
- GitHub issue: #11705
- Fix PR: https://github.com/pydantic/pydantic/pull/11883
- First fixed release: 2.11.8
- Reproduced locally: No (not executed)
- Last verified: 2026-02-09
- Confidence: 0.95
- Did this fix it?: Yes (upstream fix exists)
- Own content ratio: 0.75
Discussion
High-signal excerpts from the issue thread (symptoms, repros, edge-cases).
“I'll note that in the meanwhile, you can still specify default values with TypedDicts using:”
“Thanks for the well written feature request”
“A new MISSING sentinel was introduced in https://github.com/pydantic/pydantic/pull/11883 as an experimental feature, and available to be tested in the 2.12.0a1 release”
Failure Signature (Search String)
- - Optional to include (i.e., the client may omit them entirely),
- Currently, Pydantic does not provide a clean, expressive way to represent this intent.
Copy-friendly signature
signature.txt
Failure Signature
-----------------
- Optional to include (i.e., the client may omit them entirely),
Currently, Pydantic does not provide a clean, expressive way to represent this intent.
Error Message
Signature-only (no traceback captured)
error.txt
Error Message
-------------
- Optional to include (i.e., the client may omit them entirely),
Currently, Pydantic does not provide a clean, expressive way to represent this intent.
Minimal Reproduction
repro.py
from typing import TypedDict, NotRequired
from pydantic import TypeAdapter
class RequestDict(TypedDict, total=False):
model: str
temperature: NotRequired[float]
TypeAdapter(RequestDict).validate_python({"model": "gpt-4"})
What Broke
Clients could send null values for optional fields, violating API contracts.
Why It Broke
Introduced an experimental `MISSING` sentinel to handle optional fields that must not be null if provided
Fix Options (Details)
Option A — Upgrade to fixed release Safe default (recommended)
pip install pydantic==2.11.8
When NOT to use: This fix is not suitable for fields where null is a valid input.
Use when you can deploy the upstream fix. It is usually lower-risk than long-lived workarounds.
Option D — Guard side-effects with OnceOnly Guardrail for side-effects
Mitigate duplicate external side-effects under retries/timeouts/agent loops by gating the operation before calling external systems.
- Place OnceOnly between your code/agent and real side-effects (Stripe, emails, CRM, APIs).
- Use a stable key per side-effect (e.g., customer_id + action + idempotency_key).
- Fail-safe: configure fail-open vs fail-closed based on blast radius and spend risk.
- This is most useful when retries/timeouts can re-trigger the same external call.
Show example snippet (optional)
onceonly.py
from onceonly import OnceOnly
import os
once = OnceOnly(api_key=os.environ["ONCEONLY_API_KEY"], fail_open=True)
# Stable idempotency key per real side-effect.
# Use a request id / job id / webhook delivery id / Stripe event id, etc.
event_id = "evt_..." # replace
key = f"stripe:webhook:{event_id}"
res = once.check_lock(key=key, ttl=3600)
if res.duplicate:
return {"status": "already_processed"}
# Safe to execute the side-effect exactly once.
handle_event(event_id)
When NOT to use: Do not use this to hide logic bugs or data corruption. Use it to block duplicate external side-effects and enforce tool permissions/spend caps.
Fix reference: https://github.com/pydantic/pydantic/pull/11883
First fixed release: 2.11.8
Last verified: 2026-02-09. Validate in your environment.
When NOT to Use This Fix
- This fix is not suitable for fields where null is a valid input.
- Do not use this to hide logic bugs or data corruption. Use it to block duplicate external side-effects and enforce tool permissions/spend caps.
Verify Fix
verify
Re-run the minimal reproduction on your broken version, then apply the fix and re-run.
Did This Fix Work in Your Case?
Quick signal helps us prioritize which fixes to verify and improve.
Prevention
- Add a CI check that diffs key outputs after upgrades (OpenAPI schema snapshots, JSON payload shapes, CLI output).
- Upgrade behind a canary and run integration tests against the canary before 100% rollout.
- Make timeouts explicit and test them (unit + integration) to avoid silent behavior changes.
- Instrument retries (attempt count + reason) and alert on spikes to catch dependency slowdowns.
Version Compatibility Table
| Version | Status |
|---|---|
| 2.11.8 | Fixed |
Related Issues
No related fixes found.
Sources
We don’t republish the full GitHub discussion text. Use the links above for context.