fix(links): correct Fern-derived slugs in 4 broken internal links - #383
Open
adityabhat-spec wants to merge 2 commits into
Open
fix(links): correct Fern-derived slugs in 4 broken internal links#383adityabhat-spec wants to merge 2 commits into
adityabhat-spec wants to merge 2 commits into
Conversation
The strict full-site link audit has failed on every push to main since #359/#361 (Aug 10). Three distinct slugs were written against file names or old paths instead of the slug Fern derives from the nav page title: - /models/capabilities/voice-cloning/how-to-vc-api -> /models/documentation/voice-cloning/instant-clone-api (page title "Instant Clone (API)") - /models/documentation/llm-electron/migrate-from-openai -> /models/documentation/llm-electron/migrate-from-open-ai (page title "Migrate from OpenAI") - /models/api-reference/llm-chat-completions/electron-chat-completions -> /models/api-reference/llm-chat-completions/complete (api-ref slug comes from the endpoint operation id, not the title) The third was only reported as "pending deploy" by the checker, so it did not fail the build, but it 404s in production today. Also fixes four redirect destinations in docs.yml that pointed at the same two nonexistent slugs (the checker does not validate redirect destinations, so these 404'd silently). Verified: `check_links.py --strict` now exits 0 with 211/211 resolving and 0 pending; `check_nav.py` passes. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Picks up the Deprecation Notices page (models/deprecations/notices), added to the waves nav by #380 without re-running the generator. The drift check runs on pull_request only, so the stale file landed on main and then failed the first PR to touch a page MDX — this one. Output of `python3 scripts/build_llms_txt.py`, no hand edits. The new URL resolves: https://docs.smallest.ai/models/deprecations/notices.md returns 200. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
adityabhat-spec
added a commit
that referenced
this pull request
Aug 20, 2026
Extends this PR to the model card as well as the API reference, per review request. **Third model tag.** `hydra-v1.0` is served and not deprecated, but was documented nowhere. Added between `hydra-v1.1` and the original `hydra` in the server query-param enum, the reference's Model versions table, and the model card's Model versions and endpoint tables. **Voice rosters, verified against the live API** rather than transcribed from prose. Probing `wss://api.smallest.ai/waves/v1/s2s` showed the six voices documented for the original `hydra` tag do not exist: a `session.configure` with `voice: "wren"` is rejected with `invalid_request_error`. They are removed. Rosters are now: - `hydra-v1.1` - ten voices, unchanged - `hydra-v1.0` - fourteen voices: vaughn, brooks, cole, hayes, pierce, sterling, ellis, lane, quinn, arden, rowan, blair, emery, sawyer - `hydra` - deprecated, points at the two rosters above Every one of the 24 documented IDs is in the live API's accepted list. The `SessionConfig.voice` enum is their union. **Two corrections the probe forced:** - The server-side default is `sterling`, not `wren`, on all three tags. - An unrecognised voice is rejected with an `error` frame, not silently defaulted. The old "unknown values silently fall back to the default" claim was wrong; the client-side-validation advice still stands, for a different reason. **`warning` frame shape** is now known from the wire, but is still not added to the schema in this commit - see the PR discussion. Also retargets three spots in the model card that still pointed new users at the deprecated tag: the "How to use it" connect string, the Technical Specifications endpoint row, and the Model ID row. Not changed, deliberately: output sample rate. The probe shows `hydra-v1.1` reporting `output_audio_sample_rate: 24000` where the docs say 48 kHz (`hydra-v1.0` and `hydra` do report 48000). Held pending the S2S team's read on whether that is intended. `fern/llms.txt` is regenerated because this PR now touches a page MDX, which arms the drift check. Byte-identical to the regen in #383, so the two merge cleanly in either order. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
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
1 participant
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.
Why
The Nav & Link Check workflow's strict full-site audit has failed on every push to main since Aug 10 (#359, #361) — 10 consecutive runs, most recently 32414760003 on the #380 merge. The failures were never caused by the PRs being merged; the strict audit only runs on push to main, while PR runs check changed files only, so the two bad links slipped in and then failed every subsequent merge.
What was wrong
Three slugs were written against MDX file names / old paths instead of the slug Fern derives from the nav page title:
/models/capabilities/voice-cloning/how-to-vc-api/models/documentation/voice-cloning/instant-clone-api/models/documentation/llm-electron/migrate-from-openai/models/documentation/llm-electron/migrate-from-open-aiopen-ai/models/api-reference/llm-chat-completions/electron-chat-completions/models/api-reference/llm-chat-completions/completetitle:All three corrected targets verified 200 with the expected page titles ("Instant Voice Clone (REST API)", "Migrate from OpenAI", "Electron — Chat Completions").
The third link (in
electron.mdx) was only reported as pending deploy by the checker, so it never failed the build — but it 404s in production today, so it is fixed here too.Changes
fern/products/waves/pages/text-to-speech/model-cards/lightning-v-3-1-pro.mdx— voice-cloning API linkfern/products/waves/pages/llm/{overview,quickstart,supported-parameters}.mdx— migrate-from-openai linksfern/products/waves/pages/llm/model-cards/electron.mdx— two api-ref linksfern/docs.yml— 4 redirect destinations pointing at the same two nonexistent slugs.check_links.pydoes not validate redirect destinations, so these were 404ing silently for anyone following a legacy URL (/v1/chat/completions,/waves/v1/chat/completions,/waves/v-4-0-0/api-reference/text-to-speech/delete-a-voice-clone,/waves/documentation/voice-cloning/how-to-delete-vc).Redirect sources using the old spellings are intentional legacy entry points and are left alone.
Verification
Ran both CI checks locally against production:
No content changes — link targets only.
🤖 Generated with Claude Code