fix(search): point jina-code alias at published HF repo

[carlos-alm]

Summary

• Re-applies the fix from Fix jina-code model alias #1026 since that author has not signed the CLA.
• Maps jina-code to jinaai/jina-embeddings-v2-base-code (the published code embedding model). The previous alias Xenova/jina-embeddings-v2-base-code 404s on Hugging Face.
• Drops the stale "requires HF token" note from the README — the published repo is public.
• Adds a regression test asserting the alias resolves to the correct repo.

Fixes #1025.

Test plan

• npx vitest run tests/search/embedding-strategy.test.ts (26/26 pass, including the new test)
• npx tsc --noEmit
• npx biome check src/domain/search/models.ts tests/search/embedding-strategy.test.ts

[claude]

Claude finished @carlos-alm's task —— View job

---

PR Review Complete

Reviewing PR #1053: fix(search): point jina-code alias at published HF repo

Todo List:

• Read and analyze changed files
• Phase 1: Root cause analysis
• Phase 2: Configuration change detection
• Phase 3: Test integrity assessment
• Phase 4: Code quality review
• Phase 5: Backlog compliance rating
• Provide final assessment and recommendation

---

Root Cause Analysis

Problem Being Solved: The jina-code embedding model alias was pointing to Xenova/jina-embeddings-v2-base-code, which returns a 404 error on Hugging Face Hub. This breaks semantic search functionality when users try to use the code-specific embedding model.

Why This Approach: The fix correctly redirects the alias to jinaai/jina-embeddings-v2-base-code, which is the actual published repository for Jina's code embedding model. The approach is surgical and targeted - it fixes the broken URL without changing any other behavior.

Risk Assessment: Very low risk. This is a simple string replacement that fixes a broken external dependency reference. The change is well-tested with a regression test.

Backlog Compliance

• Zero-dep: ✓ — No new runtime dependencies added
• Foundation-aligned: ✓ — Maintains local-first approach, fixes broken external model reference
• Problem-fit: 4/5 — Directly fixes broken functionality that affects AI agents using semantic search for code understanding
• Breaking: No — Purely fixes broken functionality, no API changes
• Tier: 1 (zero-dep ✓ + foundation-aligned ✓)

Code Quality Assessment

Changes Analysis:

1. models.ts:45 - Changes alias from Xenova/jina-embeddings-v2-base-code → jinaai/jina-embeddings-v2-base-code
2. README.md:431 - Removes "(requires HF token)" note since the new repo is public
3. Tests - Adds regression test asserting the correct alias mapping

Strengths:

• Precise fix: Only changes the broken URL, nothing else
• Well-tested: Includes regression test to prevent future breakage
• Documentation sync: Updates README to reflect that HF token is no longer needed
• Clean commit: Single concern, good commit message, references the issue

Test Quality:
The regression test is simple but effective:

test('jina-code points to the published code embedding model', () => {
  expect(MODELS['jina-code'].name).toBe('jinaai/jina-embeddings-v2-base-code');
});

This will catch any future regressions where the alias gets pointed back to the wrong repository.

Technical Verification:

• I verified the old URL (Xenova/jina-embeddings-v2-base-code) would indeed 404
• The new URL (jinaai/jina-embeddings-v2-base-code) points to the official Jina AI repository
• The model dimensions and other config remain unchanged, maintaining compatibility

Minor Observations

• The fix correctly maintains all other model metadata (dim: 768, contextWindow: 8192, etc.)
• No version bump needed since this fixes broken functionality rather than changing behavior
• The change follows the existing patterns in the codebase

Final Recommendation

• Rating: ⭐⭐⭐⭐ (4/5)
• Action: APPROVE
• Reasoning: This is a clean, well-tested fix for broken functionality. The change is minimal and surgical, includes appropriate test coverage, and maintains all compatibility. The only reason it's not 5/5 is that it's a relatively straightforward bug fix rather than an exceptional contribution.

The PR successfully resolves issue #1025 and restores working semantic search functionality for code-specific embeddings.

[greptile-apps]

Greptile Summary

This PR fixes the broken jina-code model alias by pointing it at the canonical jinaai/jina-embeddings-v2-base-code HuggingFace repo instead of the now-404 Xenova mirror, drops the stale HF-token requirement note from the README, and adds a regression test to guard against future regressions.

Confidence Score: 5/5

Safe to merge — minimal, targeted fix with no logic changes beyond the broken alias

Single-line model-name fix backed by a regression test and a matching README cleanup; no logic, API, or schema changes; no issues found

No files require special attention

Important Files Changed

Filename	Overview
src/domain/search/models.ts	Updates jina-code alias from the 404-ing Xenova mirror to the canonical jinaai/jina-embeddings-v2-base-code HF repo
tests/search/embedding-strategy.test.ts	Adds a targeted regression test asserting the jina-code alias resolves to the correct HF repo name
README.md	Removes the stale "requires HF token" note from the jina-code row, consistent with the model now pointing to a public repo

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A["User requests 'jina-code' model"] --> B["getModelConfig('jina-code')"]
    B --> C["MODELS['jina-code'].name"]
    C -->|Before PR| D["Xenova/jina-embeddings-v2-base-code\n(404 on HuggingFace)"]
    C -->|After PR| E["jinaai/jina-embeddings-v2-base-code\n(public, published repo)"]
    E --> F["loadModel → HuggingFace download"]
    F --> G["Embeddings generated"]

Loading

_{Reviews (1): Last reviewed commit: "fix(search): point jina-code alias at pu..." | Re-trigger Greptile}