fix(selfhost): attribute embed/vision/advisory AI usage to a real provider

[JSONbored]

Summary

Follow-up to #5247, found during its own live-deploy verification on edge-nl-01.

#5247 fixed createOpenAiCompatibleAi to return real usage.provider attribution, but
src/server.ts's 3 call sites that actually construct the AI_EMBED/AI_VISION/AI_ADVISORY
bindings never passed providerName at all — confirmed live by triggering a real RAG-indexing
pass on edge-nl-01 right after deploying #5247: the new embeddings rows correctly showed the
real model (bge-m3:latest, not the old hardcoded placeholder) but provider was still blank on
every row, since opts.providerName was undefined at the source.

These 3 bindings deliberately accept any openai-compatible endpoint with no explicit provider
selector (unlike the main review chain's AI_PROVIDER), so there's no existing knob to read a
provider name from. Added providerNameFromBaseUrl() to classify the configured base URL into the
same closed "ollama" | "openai" | "openai-compatible" set createOpenAiCompatibleAi already uses
— recognizing the bundled docker-compose ollama service and OpenAI's own endpoint, and falling
back to the honest generic bucket for anything else (vLLM, LM Studio, a custom hostname) rather than
guessing.

Test plan

• npm run typecheck — clean
• npm run test:ci — full local gate green
• New unit test covers all 4 classification branches (ollama by hostname substring, openai's
  exact hostname, a generic other host, and the undefined/malformed-URL fallback)
• Live-verified on edge-nl-01 after deploying this fix: a manually triggered RAG-index pass
  produced a fresh embeddings row with provider = 'ollama', model = 'bge-m3:latest', and an
  organic screenshot_table_vision row landed with provider = 'ollama', model = 'qwen3-vl:8b-instruct' — both were previously always blank on provider.

[superagent-security]

Superagent didn't find any vulnerabilities or security issues in this PR.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.36%. Comparing base (73a0912) to head (976f874).
⚠️ Report is 3 commits behind head on main.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #5259   +/-   ##
=======================================
  Coverage   94.36%   94.36%           
=======================================
  Files         474      474           
  Lines       40081    40088    +7     
  Branches    14614    14617    +3     
=======================================
+ Hits        37821    37828    +7     
  Misses       1585     1585           
  Partials      675      675           

Flag	Coverage Δ
shard-1	46.24% <0.00%> (-0.29%)	⬇️
shard-2	34.74% <100.00%> (+0.08%)	⬆️
shard-3	32.07% <0.00%> (+0.03%)	⬆️
shard-4	31.84% <0.00%> (-0.29%)	⬇️
shard-5	33.68% <0.00%> (-0.04%)	⬇️
shard-6	44.98% <0.00%> (+0.18%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
src/selfhost/ai.ts	98.15% <100.00%> (+0.02%)	⬆️

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[gittensory-orb]

Warning

🟨🟨🟨🟨🟨🟨🟨🟨🟨🟨🟨🟨

⏸️ Gittensory review result - manual review recommended

_{Review updated: 2026-07-12 12:18:44 UTC}

3 files · 2 AI reviewers · 2 blockers · readiness 93/100 · CI green · unstable

⏸️ Suggested Action - Manual Review

• No linked issue detected — If this PR is intended to solve an issue, link it explicitly in the PR body.
• Maintainer requires a linked issue — Link the relevant issue (for example Closes #123) before opening the PR.

Review summary
This is a small, well-targeted follow-up to #5247: it adds providerNameFromBaseUrl() to classify AI_EMBED/AI_VISION/AI_ADVISORY base URLs into the existing ollama|openai|openai-compatible set, and wires it into the 3 server.ts call sites that were previously leaving providerName undefined. The classification logic is simple and defensively handles malformed/undefined URLs via try/catch returning the generic fallback, and the new unit test exercises all 4 branches. This directly closes the attribution gap described in the PR summary and matches the existing convention in createOpenAiCompatibleAi.

Nits — 5 non-blocking

• providerNameFromBaseUrl's `hostname.includes("ollama")` substring match would misclassify a hostname like `myollama-alternative.example.com` as ollama even though it's an unrelated third-party host; consider anchoring to a suffix/exact match if false positives matter in practice.
• The long comment block above providerNameFromBaseUrl (src/selfhost/ai.ts) is quite dense for a doc comment; could be trimmed to the essential WHY (no explicit provider knob exists) without the full enumeration of every scenario.
• Consider extracting the ollama-hostname heuristic into a named constant/regex if it's likely to need adjustment as more self-hosted provider names are supported.
• nit: src/selfhost/ai.ts:612 uses a broad substring match for `ollama`, so a hostname like `not-ollama.example` is attributed as `ollama`; this is acceptable for conservative telemetry only if that looseness is intentional.
• nit: src/selfhost/ai.ts:612 has a long explanatory block comment that reads more like PR context than durable code documentation; trim it to the invariant the helper enforces.

Concerns raised — review before merging

• No linked issue detected — If this PR is intended to solve an issue, link it explicitly in the PR body.
• Maintainer requires a linked issue — Link the relevant issue (for example Closes #123) before opening the PR.

Signal	Result	Evidence
Code review	❌ 2 blockers	2 reviewers, synthesized
Linked issue	⚠️ Missing	No linked issue or no-issue rationale found.
Related work	✅ No active overlap found	No same-issue or scoped active PR overlap found.
Change scope	✅ 20/20	Low review scope from cached public metadata (no linked issue context).
Validation posture	✅ 25/25	PR body includes validation/test evidence.
Contributor workload	✅ 10/10	Author activity: 44 registered-repo PR(s), 36 merged, 470 issue(s).
Contributor context	✅ Confirmed Gittensor contributor	JSONbored; Gittensor profile; 44 PR(s), 470 issue(s).
Gate result	❌ Blocking	Repo-configured hard blocker found.
Improvement	✅ Minor	risk: clean · value: minor — Code changes are accompanied by test evidence. LLM value judgment: minor — The change is a narrow, correct bug-fix follow-up that wires an already-existing classification convention into three call sites, fixing real blank-provider data without expanding scope.

Review context

• Author: JSONbored
• Role context: owner (maintainer lane)
• Public audience mode: oss maintainer
• Lane context: Repository is configured for direct PR review.
• Public profile languages: not available
• Official Gittensor activity: 44 PR(s), 470 issue(s).
• PR-specific overlap: none found.

Contributor next steps

• Treat this as maintainer-lane context rather than normal contributor-lane activity.
• Explain no-issue PR.
• Link the issue being solved, or explicitly explain why this is a no-issue PR.

Signal definitions

• Related work = same linked issue, overlapping active PRs, or title/path similarity.
• Change scope = cached public metadata such as size labels, draft state, and review-burden hints.
• Validation posture = whether the PR provides enough public validation/test evidence for maintainer review.
• Contributor workload = public contributor activity and cleanup pressure, not a repo-wide quality failure.
• Contributor context = public GitHub/Gittensor identity context; non-Gittensor status is not a blocker.

[BETA] Chat with Gittensory

Ask Gittensory a question about this PR directly in a comment — grounded only in the same cached, public-safe facts shown above, never a new claim.

• @gittensory ask <question> answers contribution-quality Q&A with source citations and freshness.
• @gittensory chat <question> answers in natural prose from cached decision-pack facts via local inference (maintainer/collaborator; read-only).
• A plain-language @gittensory mention with a real question is routed to the closest matching read-only command automatically -- no exact syntax required.

Full command reference: https://gittensory.aethereal.dev/docs/gittensory-commands

_{🟩 Safe / merged · 🟦 Advisory · 🟨 Held for review · 🟥 Blocked / closed}

---

💰 Earn for open-source contributions like this. Gittensor lets GitHub contributors earn for the work they already do — register to start earning →.

Checked by Gittensory, a quiet PR intelligence layer for OSS maintainers.

• Re-run Gittensory review