Skip to content

fix(bedrock): validate TagResource ARN, add embed schema, fix profanity and deployments#277

Merged
vieiralucas merged 2 commits intomainfrom
worktree-fix+cubic-batch5-bedrock
Apr 12, 2026
Merged

fix(bedrock): validate TagResource ARN, add embed schema, fix profanity and deployments#277
vieiralucas merged 2 commits intomainfrom
worktree-fix+cubic-batch5-bedrock

Conversation

@vieiralucas
Copy link
Copy Markdown
Member

@vieiralucas vieiralucas commented Apr 12, 2026

Summary

  • Reject empty/missing resourceARN in TagResource with ValidationException instead of storing tags under empty key
  • Return embedding vector schema for amazon.titan-embed-* models instead of text-generation schema
  • Use word-boundary matching for profanity detection to prevent false positives (e.g., "assignment" no longer matches "ass")
  • Serialize non-string hyperparameter values instead of silently converting to empty string
  • Generate unique deployment names when modelDeploymentName is not provided

Addresses unresolved Cubic findings from PRs #260, #266.

Test plan

  • cargo clippy -p fakecloud-bedrock -- -D warnings passes
  • All unit tests pass

Summary by cubic

Fixes correctness and robustness in fakecloud-bedrock: validates TagResource ARN, returns proper Titan embed schema across variants, improves profanity checks, and hardens deployment/hyperparameter handling. Addresses remaining Cubic findings from earlier PRs.

  • Bug Fixes
    • Validate TagResource resourceARN; return ValidationException when missing or empty.
    • Return embedding vector schema for amazon.titan-embed* models (prefix match) instead of text-generation.
    • Use word-boundary matching for profanity to avoid false positives (e.g., "assignment").
    • Serialize non-string hyperparameter values; auto-generate unique deployment names when modelDeploymentName is not provided.

Written for commit 6ef8765. Summary will update on new commits.

…ty matching, deployment names

- Reject empty/missing resourceARN in TagResource with ValidationException
- Return embedding vector schema for amazon.titan-embed models instead of text-gen schema
- Use word-boundary matching for profanity detection to prevent false positives
- Serialize non-string hyperparameter values instead of silently converting to empty string
- Generate unique deployment names when modelDeploymentName is not provided
Copy link
Copy Markdown

@cubic-dev-ai cubic-dev-ai Bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

1 issue found across 5 files

Prompt for AI agents (unresolved issues)

Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="crates/fakecloud-bedrock/src/invoke.rs">

<violation number="1" location="crates/fakecloud-bedrock/src/invoke.rs:181">
P2: Use a Titan-embedding-specific prefix check instead of `contains("embed")` to avoid returning embedding schema for unrelated model IDs.</violation>
</file>

Reply with feedback, questions, or to request a fix. Tag @cubic-dev-ai to re-run a review.

Comment thread crates/fakecloud-bedrock/src/invoke.rs Outdated
@vieiralucas vieiralucas merged commit 2689745 into main Apr 12, 2026
22 checks passed
@vieiralucas vieiralucas deleted the worktree-fix+cubic-batch5-bedrock branch April 12, 2026 13:44
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant