feat(prompts): Fetch prompts by label - #4163
Merged
Merged
Conversation
jurajmajerik
marked this pull request as ready for review
July 16, 2026 12:54
jurajmajerik
changed the title
Contributor
|
Reviews (1): Last reviewed commit: "feat(ai): fetch prompts by label" | Re-trigger Greptile |
![greptile-apps[bot]](https://avatars.githubusercontent.com/in/867647?s=60&v=4){kind=small}
Comment on lines
280
to
+281
| const versionQuery = version === undefined ? '' : `&version=${encodeURIComponent(String(version))}` | ||
| const promptLabel = this.getPromptLabel(name, version) | ||
| const url = `${this.host}/api/environments/@current/llm_prompts/name/${encodedPromptName}/?token=${encodedProjectApiKey}${versionQuery}` | ||
| const labelQuery = label === undefined ? '' : `&label=${encodeURIComponent(label)}` |
Contributor
There was a problem hiding this comment.
When a configured label resolves to an empty string, this path sends label= and caches the response under the empty-string key. If the API treats an empty label as missing or rejects it, callers either receive the latest prompt instead of a labeled prompt or unexpectedly fall back; reject empty labels before building the request.
Contributor
Author
There was a problem hiding this comment.
No, empty label will be rejected with a 400, and thus never cached
Contributor
There was a problem hiding this comment.
Fair enough — a 400 will throw before any caching happens, so the failure is explicit rather than silent. My concern doesn't apply here.
Tip: You can customize Greptile's behavior for this repo with .greptile/rules.md and .greptile/config.json.
Contributor
Contributor
|
Size Change: +2.68 kB (+0.02%) Total Size: 17.5 MB
ℹ️ View Unchanged
|
Radu-Raicea
approved these changes
Jul 16, 2026
Radu-Raicea
left a comment
Radu-Raicea
left a comment
Member
There was a problem hiding this comment.
Also needs the deploy changeset/label added.
I noticed on both this one and the Python library PR that you don't verify that the response prompt's label matches the requested one. It can be what we want, up to you.
Contributor
Author
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
PostHog prompt management is gaining labels: movable pointers from a name like
productionto exactly one version of a prompt, so a version can be released or rolled back without a code deploy (PostHog/posthog#70862, PostHog/posthog#71509). The@posthog/aiPrompts class can fetch by name or version number, but not by label — and fetching by label is the way labels are meant to be consumed.Changes
Adds a
labeloption toPrompts.get():labelis passed as a query param on the existing fetch endpoint; mutually exclusive withversion(throws if both are given, matching the API's 400)PromptRemoteResultand the cache carry alabelfield, so callers can see which label a prompt resolved throughDraft until the backend PR (PostHog/posthog#71509) is deployed: the API ignores unknown query params, so releasing this first would send
label=productionand silently get the latest version back. Mirrors PostHog/posthog-python#744.How did you test this code?
New jest cases: the label query param is sent and label metadata comes back on the result, version+label together throws before any fetch, and labeled fetches are cached separately from latest fetches (without the cache-key change, a labeled fetch after a latest fetch silently returns the wrong version). Full
prompts.test.tssuite passes (49 tests), eslint clean; the two pre-existing tsc errors incaptureAiGeneration.tsagainst a stale core build are unrelated to this change.