feat(llmobs): prompt management SDK methods by PROFeNoM · Pull Request #18186 · DataDog/dd-trace-py

PROFeNoM · 2026-05-20T13:09:21Z

Description

Adds prompt management methods to the LLMObs Python SDK, calling the new public API endpoints from MLOB-7523.

Blocked by: https://github.com/DataDog/dd-source/pull/443753 (public CRUD API routes)

New public API

# Write methods (require DD_API_KEY + DD_APP_KEY)
LLMObs.create_prompt(prompt_id, template, *, title, description, user_version, labels)
LLMObs.create_prompt_version(prompt_id, template, *, description, user_version, labels)
LLMObs.update_prompt(prompt_id, *, title, description)
LLMObs.update_prompt_version(prompt_id, version, *, labels, description)
LLMObs.delete_prompt(prompt_id)

# Read methods (require DD_API_KEY only)
LLMObs.list_prompts()
LLMObs.list_prompt_versions(prompt_id)

list_prompts does not take an ml_app filter: the backend list route does not yet honor filter[ml_app] (split into DataDog/dd-source#459316). The parameter will be re-added once that lands, to avoid shipping a silent no-op.

Not covered: GET /prompts/{prompt_id}/versions/{version} is intentionally left out. Prefer to add it once #18127 (hybrid prompt delivery) lands, since it changes the get_prompt signature and it's trivial.

New types (`ddtrace/llmobs/types.py`)

ChatMessage(TypedDict) - template message format (role + content only)
PromptResponse(TypedDict) - returned by create/update/list prompt operations
PromptVersionResponse(TypedDict) - returned by create/update/list version operations
DeletedPromptResponse(TypedDict) - returned by delete

Exception hierarchy

PromptAPIError (base)
  PromptAuthError        - 401/403 (bad API/app key)
  PromptValidationError  - 400 (bad input)
  PromptNotFoundError    - 404
  PromptConflictError    - 409 (duplicate prompt_id)
  PromptServerError      - 5xx

Each method documents which exceptions it can raise.

Cache changes (`cache.py`)

WarmCache now uses per-prompt subdirectories: ~/.cache/datadog/llmobs/prompts/{prompt_id}/{label}.json, with quote()-encoded segments so distinct ids/labels never collide on the same file
Added evict_prompt(prompt_id) - uses shutil.rmtree on the prompt directory

Files changed

File	Change
`types.py`	`ChatMessage`, 3 response TypedDicts, 6 exception classes (`PromptAPIError` base + 5 status-specific)
`_llmobs.py` / `manager.py`	label/labels typed as `str`/`list[str]`
`cache.py`	Per-prompt subdirectory layout with `quote()`-encoded segments, `evict_prompt` method
`manager.py`	`_request` helper, 7 CRUD methods, API key validation on `get_prompt` and `refresh_prompt`
`_llmobs.py`	7 public classmethods delegating to manager; `clear_prompt_cache` reads the manager under its lock
`test_prompts.py`	Exception mapping, cache eviction, path-collision, and read-path guard tests

E2E validation (staging) - 13/13 pass

Tested against datad0g.com using the SDK. Save the script below, install the branch via an editable local checkout, replace keys with your own staging credentials, and run with python test_sdk.py.

#	Journey	SDK method	Expected
1	Create prompt	`create_prompt()`	Returns PromptResponse with matching prompt_id
2	Create duplicate	`create_prompt()`	Raises PromptConflictError (409)
3	List prompts	`list_prompts()`	Our prompt in returned list
4	Create version	`create_prompt_version()`	Returns PromptVersionResponse
5	List versions	`list_prompt_versions()`	>= 2 versions
6	Update prompt	`update_prompt()`	Returns updated PromptResponse
7	Update version	`update_prompt_version()`	Returns updated PromptVersionResponse
8	Get prompt (read path)	`get_prompt()`	source="registry"
9	Get prompt with label	`get_prompt(label="development")`	Matches labeled version
10	Validation error	`update_prompt()` with no fields	Raises PromptValidationError (client-side)
11	Delete prompt	`delete_prompt()`	Returns DeletedPromptResponse
12	Get deleted prompt	`get_prompt()`	Raises ValueError (no fallback)
13	Delete non-existent	`delete_prompt()`	Raises PromptNotFoundError (404)

"""
Setup:
    # In your dd-trace-py checkout, switch to this branch (run from anywhere via -C):
    git -C <path-to-dd-trace-py> checkout alex/MLOB-7524_prompt-crud-api

    # In your test working dir (separate from the repo), make a venv and
    # editable-install that checkout (compiles native ext once):
    uv venv && source .venv/bin/activate
    uv pip install -e <path-to-dd-trace-py>

Run:
    DD_API_KEY="<your-datad0g-api-key>" \
    DD_APP_KEY="<your-datad0g-app-key>" \
    DD_SITE=datad0g.com \
    python test_sdk.py
"""

import os
import sys
import time
import traceback

os.environ.setdefault("DD_API_KEY", "<your-datad0g-api-key>")
os.environ.setdefault("DD_APP_KEY", "<your-datad0g-app-key>")
os.environ.setdefault("DD_SITE", "datad0g.com")

from ddtrace.llmobs import LLMObs
from ddtrace.llmobs.types import (
    PromptAPIError,
    PromptAuthError,
    PromptConflictError,
    PromptNotFoundError,
    PromptServerError,
    PromptValidationError,
)

RUN_ID = f"e2e-{int(time.time())}"
PROMPT_ID = f"sdk-test-{RUN_ID}"

passed = 0
failed = 0


def test(name):
    def decorator(fn):
        global passed, failed
        try:
            fn()
            print(f"[PASS] {name}")
            passed += 1
        except Exception:
            print(f"[FAIL] {name}")
            traceback.print_exc()
            failed += 1
        return fn
    return decorator


print(f"=== Prompt CRUD E2E - SDK ===")
print(f"Run ID: {RUN_ID}")
print(f"Prompt ID: {PROMPT_ID}")
print()


@test("Create prompt")
def _():
    resp = LLMObs.create_prompt(
        PROMPT_ID,
        [
            {"role": "system", "content": "You are a {{persona}}."},
            {"role": "user", "content": "{{question}}"},
        ],
        title="E2E Test Prompt",
        description="Created by SDK e2e test",
    )
    print(f"  Response: {resp}")
    assert resp["prompt_id"] == PROMPT_ID


@test("Create duplicate raises PromptConflictError")
def _():
    try:
        LLMObs.create_prompt(PROMPT_ID, [{"role": "user", "content": "dup"}])
        assert False, "Should have raised"
    except PromptConflictError as e:
        assert e.status == 409


@test("List prompts")
def _():
    prompts = LLMObs.list_prompts()
    print(f"  Total: {len(prompts)}")
    assert PROMPT_ID in [p["prompt_id"] for p in prompts]


@test("Create prompt version")
def _():
    resp = LLMObs.create_prompt_version(
        PROMPT_ID,
        [
            {"role": "system", "content": "You are a helpful {{persona}}."},
            {"role": "user", "content": "Please answer: {{question}}"},
        ],
        description="v2 - improved",
        user_version="v2",
    )
    print(f"  Response: {resp}")


@test("List prompt versions")
def _():
    versions = LLMObs.list_prompt_versions(PROMPT_ID)
    print(f"  Count: {len(versions)}")
    assert len(versions) >= 2


@test("Update prompt metadata")
def _():
    resp = LLMObs.update_prompt(PROMPT_ID, title="Updated", description="Updated")
    print(f"  Response: {resp}")


@test("Update prompt version")
def _():
    versions = LLMObs.list_prompt_versions(PROMPT_ID)
    ver = str(versions[-1].get("version", 1))
    resp = LLMObs.update_prompt_version(
        PROMPT_ID, ver, description="Updated version", labels=["development"]
    )
    print(f"  Response: {resp}")


@test("Get prompt (read path)")
def _():
    prompt = LLMObs.get_prompt(PROMPT_ID)
    print(f"  id={prompt.id}, version={prompt.version}, source={prompt.source}")
    assert prompt.id == PROMPT_ID
    assert prompt.source == "registry"


@test("Get prompt with label=development")
def _():
    try:
        prompt = LLMObs.get_prompt(PROMPT_ID, label="development")
        print(f"  id={prompt.id}, version={prompt.version}, label={prompt.label}")
    except ValueError as e:
        print(f"  Label not found (expected if not set): {e}")


@test("Update with no fields raises PromptValidationError")
def _():
    try:
        LLMObs.update_prompt(PROMPT_ID)
        assert False
    except PromptValidationError as e:
        assert e.status == 0


@test("Delete prompt")
def _():
    resp = LLMObs.delete_prompt(PROMPT_ID)
    print(f"  Response: {resp}")


@test("Get deleted prompt raises ValueError")
def _():
    LLMObs.clear_prompt_cache(hot=True, warm=True)
    try:
        LLMObs.get_prompt(PROMPT_ID)
        assert False
    except ValueError as e:
        assert "could not be fetched" in str(e)


@test("Delete non-existent raises PromptNotFoundError")
def _():
    try:
        LLMObs.delete_prompt("nonexistent-" + RUN_ID)
        assert False
    except PromptNotFoundError as e:
        assert e.status == 404


print()
print(f"=== Results: {passed} passed, {failed} failed ===")
sys.exit(1 if failed > 0 else 0)

Expected output

=== Prompt CRUD E2E - SDK ===
Run ID: e2e-1780652134
Prompt ID: sdk-test-e2e-1780652134

  Response: {'prompt_id': 'sdk-test-e2e-1780652134', 'title': 'E2E Test Prompt', 'source': 'registry', 'created_at': '2026-06-05T09:35:34.581341154Z', 'last_version_created_at': '2026-06-05T09:35:34.581341154Z', 'num_versions': 1, 'in_registry': True, 'created_from': 'sdk-registry', 'author': 'c5e19e5a-ad6b-11ed-ab4f-927130d31ef9', 'description': 'Created by SDK e2e test', 'id': 'f590c01f-8075-5c3d-9126-c77ff5003183'}
[PASS] Create prompt
[PASS] Create duplicate prompt raises PromptConflictError
  Total prompts: 78
[PASS] List prompts
  Response: {'prompt_uuid': 'f590c01f-8075-5c3d-9126-c77ff5003183', 'prompt_id': 'sdk-test-e2e-1780652134', 'template': [{'role': 'system', 'content': 'You are a helpful {{persona}}.'}, {'role': 'user', 'content': 'Please answer: {{question}}'}], 'version': 2, 'user_version': 'v2', 'created_at': '2026-06-05T09:35:36.166637357Z', 'version_created_at': '2026-06-05T09:35:36.166637357Z', 'author': 'c5e19e5a-ad6b-11ed-ab4f-927130d31ef9', 'description': 'v2 - improved template', 'id': '6cf36c0a-16cd-5191-bfae-5faf5c65a545'}
[PASS] Create prompt version
  Version count: 2
[PASS] List prompt versions
  Response: {'prompt_id': 'sdk-test-e2e-1780652134', 'title': 'E2E Test Prompt (updated)', 'source': 'registry', 'created_at': '2026-06-05T09:35:34.581341Z', 'last_version_created_at': '0001-01-01T00:00:00Z', 'num_versions': 0, 'in_registry': True, 'created_from': 'sdk-registry', 'author': 'c5e19e5a-ad6b-11ed-ab4f-927130d31ef9', 'description': 'Updated by SDK e2e test', 'id': 'f590c01f-8075-5c3d-9126-c77ff5003183'}
[PASS] Update prompt metadata
  Updating version: 1
  Response: {'prompt_uuid': 'f590c01f-8075-5c3d-9126-c77ff5003183', 'prompt_id': 'sdk-test-e2e-1780652134', 'template': [{'role': 'system', 'content': 'You are a {{persona}}.'}, {'role': 'user', 'content': '{{question}}'}], 'version': 1, 'labels': ['development'], 'created_at': '2026-06-05T09:35:34.581341Z', 'version_created_at': '2026-06-05T09:35:34.581341Z', 'author': 'c5e19e5a-ad6b-11ed-ab4f-927130d31ef9', 'description': 'Updated version description', 'id': 'cf74e1a2-5638-596b-b738-3ae66261577e'}
[PASS] Update prompt version
  Prompt ID: sdk-test-e2e-1780652134, version: v2, source: registry
[PASS] Get prompt (read path)
  Prompt ID: sdk-test-e2e-1780652134, version: 1, label: development
[PASS] Get prompt with label=development
[PASS] Update prompt with no fields raises PromptValidationError
  Response: {'prompt_id': 'sdk-test-e2e-1780652134', 'deleted_at': '2026-06-05T09:35:39.215389Z', 'id': 'f590c01f-8075-5c3d-9126-c77ff5003183'}
[PASS] Delete prompt
[PASS] Get deleted prompt raises ValueError (no fallback)
[PASS] Delete non-existent prompt raises PromptNotFoundError

=== Results: 13 passed, 0 failed ===

Add CRUD operations for LLM Observability prompt registry to the Python SDK.

datadog-prod-us1-6 · 2026-05-20T13:10:16Z

Tests

✨ Fix all issues with BitsAI

⚠️ Warnings

🚦 8 Pipeline jobs failed

DataDog/apm-reliability/dd-trace-py | build linux serverless: [amd64, cp315-cp315, v113741238-d2b8243-manylinux2014_x86_64, 1]

See error
Failed to generate the artifact due to a package version mismatch for 'bytecode'.

DataDog/apm-reliability/dd-trace-py | build linux serverless: [amd64, cp315-cp315, v113741491-d2b8243-musllinux_1_2_x86_64, 1]

See error
NotImplementedError: This version of CPython is not supported yet

DataDog/apm-reliability/dd-trace-py | build linux serverless: [arm64, cp315-cp315, v113741357-d2b8243-manylinux2014_aarch64, 1]

See error
NotImplementedError: This version of CPython is not supported yet

View all 8 failed jobs.

ℹ️ Info

No other issues found (see more)

🧪 All tests passed
❄️ No new flaky tests detected

Useful? React with 👍 / 👎

_{This comment will be updated automatically if new data arrives.

🔗 Commit SHA: 241ccdd | Docs | Datadog PR Page | Give us feedback!}

cit-pr-commenter-54b7da · 2026-05-20T13:14:59Z

Codeowners resolved as

tests/llmobs/test_prompts.py                                            @DataDog/ml-observability

list_prompts and list_prompt_versions only need DD_API_KEY since the backend uses ValidReportingAPIUser for read endpoints.

… [MLOB-7524]

…ons [MLOB-7524]

_request() passes body= to conn.request(), mock signature needed to match.

…[MLOB-7524] Handler expects filter[ml_app], SDK was sending ml_app.

…B-7524] Plain JSON API returns bare arrays, not JSONAPI {"data": [...]} wrappers.

PROFeNoM · 2026-05-27T10:09:02Z

@codex review

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 17efe7f82a

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

…ecycle [MLOB-7524] Both _fetch_from_registry and _request shared identical connection setup/teardown logic. Extract low-level _http_request(method, path, body, headers, timeout) -> (status, response_body) that handles transport only. Each caller keeps its own error handling strategy.

…-7524] Map HTTP status codes to exception classes via _STATUS_EXCEPTIONS dict instead of chained if/raise statements. Cleaner, data-driven, easier to extend.

…-crud-api

…-7524]

create_prompt, create_prompt_version, update_prompt, and update_prompt_version now evict hot+warm cache entries so subsequent get_prompt calls reflect the latest state instead of serving stale data until TTL expiry.

…[MLOB-7524] The API expects a numeric auto-increment version number in the URL path (parsed via PathInt on the Go side). Changing from str to int makes the contract explicit and prevents misuse with user_version strings.

- Fix cache key parsing: use rsplit instead of partition to handle prompt_ids containing colons - Wrap json.loads in _request with try-except for malformed 2xx responses - Extract _evict_prompt_caches helper to deduplicate 5 eviction sites

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: bffe1315a5

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

A read path such as get_prompt() or list_prompts() can build and cache the PromptManager before enable(app_key=...) runs. enable() updated LLMObs._app_key but left the cached manager untouched, so it kept its then-empty app key and every write API (create/update/delete) raised PromptAuthError even though a valid app key had been provided. Invalidate the cached manager in enable() when an app key is supplied so it rebuilds with the new key on next use.

- refresh_prompt: guard DD-API-KEY like get_prompt instead of relying on server reject - clear_prompt_cache: snapshot prompt manager under lock to avoid TOCTOU null-deref vs enable() - HotCache.evict_prompt: match exact prompt_id so colon-bearing ids no longer over-evict - WarmCache: encode path segments with quote() (injective) so distinct ids no longer collide on the same cache file

…label param Labels are deployment markers (typically DD_ENV values) and the registry/DB already store and match them as arbitrary strings. Drop the PromptLabel = Literal["development","production"] alias entirely and type label/labels as str / list[str] directly. Deprecate the read 'label' parameter of get_prompt()/refresh_prompt() via debtcollector: the going-forward mechanism is to set DD_ENV and let the SDK resolve the version for that environment. The write 'labels' param (create/ version) is unchanged.

emmettbutler

release note approved

emmettbutler · 2026-06-03T19:59:33Z

+deprecations:
+  - |
+    LLM Observability: The ``label`` parameter of ``LLMObs.get_prompt()`` and ``LLMObs.refresh_prompt()`` is
+    deprecated. Set ``DD_ENV`` instead; the prompt version is resolved for that environment.


Suggested change

deprecated. Set ``DD_ENV`` instead; the prompt version is resolved for that environment.

deprecated. Set ``DD_ENV`` instead; the prompt version is resolved for that environment. #18186

Also, when will the deprecated parameter be removed?

The backend list route does not yet honor filter[ml_app] (the fix is split into a separate dd-source bugfix PR). Remove the SDK ml_app parameter so it does not advertise a silent no-op; it will be re-added once the backend filter lands.

MockHTTPResponse.read() only handled dict and str bodies; a list body (e.g. the list_prompts response) fell through to b"", so _request saw an empty body and returned {}. test_request_normalizes_backend_id_key[response1] (list_prompts) asserted a list and failed. Serialize any non-str, non-None body as JSON.

feat(llmobs): add prompt management write methods [MLOB-7524]

7577f2e

Add CRUD operations for LLM Observability prompt registry to the Python SDK.

chore(llmobs): add release note for prompt CRUD methods [MLOB-7524]

d79cc3a

PROFeNoM changed the title ~~feat(llmobs): prompt management write methods [MLOB-7524]~~ feat(llmobs): prompt management write methods May 20, 2026

PROFeNoM added 3 commits May 20, 2026 15:15

fix(llmobs): don't require DD_APP_KEY for read methods [MLOB-7524]

b73a398

list_prompts and list_prompt_versions only need DD_API_KEY since the backend uses ValidReportingAPIUser for read endpoints.

fix(llmobs): validate DD_API_KEY in _request, clarify auth docstrings…

b72675d

… [MLOB-7524]

fix(llmobs): validate DD_API_KEY in get_prompt [MLOB-7524]

f0ec2a3

PROFeNoM changed the title ~~feat(llmobs): prompt management write methods~~ feat(llmobs): prompt management SDK methods [MLOB-7524] May 20, 2026

PROFeNoM added 2 commits May 20, 2026 15:52

fix(llmobs): use Any return on _request with typed call-site annotati…

d23e3b5

…ons [MLOB-7524]

style(llmobs): format types.py [MLOB-7524]

9faa87a

PROFeNoM changed the title ~~feat(llmobs): prompt management SDK methods [MLOB-7524]~~ feat(llmobs): prompt management SDK methods May 21, 2026

PROFeNoM added 3 commits May 21, 2026 09:22

fix(llmobs): add body kwarg to MockHTTPConnection.request [MLOB-7524]

7dc0e5c

_request() passes body= to conn.request(), mock signature needed to match.

fix(llmobs): use filter[ml_app] query param to match API expectation …

4bbd190

…[MLOB-7524] Handler expects filter[ml_app], SDK was sending ml_app.

fix(llmobs): remove data envelope unwrapping from list endpoints [MLO…

17efe7f

…B-7524] Plain JSON API returns bare arrays, not JSONAPI {"data": [...]} wrappers.

PROFeNoM force-pushed the alex/MLOB-7524_prompt-crud-api branch from cfd1734 to 17efe7f Compare May 21, 2026 16:45

chatgpt-codex-connector Bot reviewed May 27, 2026

View reviewed changes

Comment thread ddtrace/llmobs/_prompts/manager.py

Comment thread ddtrace/llmobs/_prompts/manager.py

PROFeNoM added 8 commits May 27, 2026 12:20

refactor(llmobs): replace status code if-chain with dict lookup [MLOB…

264148b

…-7524] Map HTTP status codes to exception classes via _STATUS_EXCEPTIONS dict instead of chained if/raise statements. Cleaner, data-driven, easier to extend.

Merge remote-tracking branch 'origin/main' into alex/MLOB-7524_prompt…

60ba43f

…-crud-api

fix(llmobs): rename section comment to reflect read+write scope [MLOB…

4850f04

…-7524]

fix(llmobs): evict prompt cache on all write operations [MLOB-7524]

102c43c

create_prompt, create_prompt_version, update_prompt, and update_prompt_version now evict hot+warm cache entries so subsequent get_prompt calls reflect the latest state instead of serving stale data until TTL expiry.

update release note

6d3c83a

PROFeNoM marked this pull request as ready for review May 28, 2026 15:30

PROFeNoM requested review from a team as code owners May 28, 2026 15:30

PROFeNoM requested a review from gnufede May 28, 2026 15:30

PROFeNoM requested a review from P403n1x87 May 28, 2026 15:30

chatgpt-codex-connector Bot reviewed May 28, 2026

View reviewed changes

Comment thread ddtrace/llmobs/_llmobs.py

PROFeNoM added 3 commits May 29, 2026 11:11

PROFeNoM force-pushed the alex/MLOB-7524_prompt-crud-api branch from 3224971 to 2a56014 Compare June 3, 2026 07:42

emmettbutler approved these changes Jun 3, 2026

View reviewed changes

PROFeNoM added 4 commits June 4, 2026 11:12

fix(llmobs): keep label deprecation out of prompt CRUD

1d2acd5

fix(llmobs): harden prompt CRUD responses and errors

7591cdb

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat(llmobs): prompt management SDK methods#18186

feat(llmobs): prompt management SDK methods#18186
PROFeNoM wants to merge 25 commits into
mainfrom
alex/MLOB-7524_prompt-crud-api

PROFeNoM commented May 20, 2026 •

edited

Loading

Uh oh!

datadog-prod-us1-6 Bot commented May 20, 2026 •

edited by datadog-datadog-prod-us1-2 Bot

Loading

Uh oh!

cit-pr-commenter-54b7da Bot commented May 20, 2026 •

edited

Loading

Uh oh!

PROFeNoM commented May 27, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

emmettbutler left a comment

Uh oh!

emmettbutler Jun 3, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

	deprecated. Set ``DD_ENV`` instead; the prompt version is resolved for that environment.
	deprecated. Set ``DD_ENV`` instead; the prompt version is resolved for that environment. #18186

Conversation

PROFeNoM commented May 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

New public API

New types (ddtrace/llmobs/types.py)

Exception hierarchy

Cache changes (cache.py)

Files changed

Expected output

Uh oh!

datadog-prod-us1-6 Bot commented May 20, 2026 • edited by datadog-datadog-prod-us1-2 Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

⚠️ Warnings

ℹ️ Info

Uh oh!

cit-pr-commenter-54b7da Bot commented May 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codeowners resolved as

Uh oh!

PROFeNoM commented May 27, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

emmettbutler left a comment

Choose a reason for hiding this comment

Uh oh!

emmettbutler Jun 3, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

PROFeNoM commented May 20, 2026 •

edited

Loading

New types (`ddtrace/llmobs/types.py`)

Cache changes (`cache.py`)

datadog-prod-us1-6 Bot commented May 20, 2026 •

edited by datadog-datadog-prod-us1-2 Bot

Loading

cit-pr-commenter-54b7da Bot commented May 20, 2026 •

edited

Loading