Skip to content

Harden GGUF parser and runtime contract#2

Merged
LessUp merged 1 commit into
masterfrom
copilot/harden-gguf-contracts
May 22, 2026
Merged

Harden GGUF parser and runtime contract#2
LessUp merged 1 commit into
masterfrom
copilot/harden-gguf-contracts

Conversation

@LessUp
Copy link
Copy Markdown
Collaborator

@LessUp LessUp commented May 22, 2026

Summary

  • harden GGUF parser overflow/oversized-array behavior and zero-length string handling
  • remove schema-invalid metadata fallback and make GGUF runtime load fail-fast/explicit
  • add OpenSpec change artifacts and regression tests for parser/loader seam

Verification

  • clang-format-18 --dry-run --Werror on touched C++ files
  • cmake configure attempted with session CUDA toolkit; blocked locally by missing static CUDA runtime libs during compiler-id link

- Guard GGUF tensor/array size math against overflow and oversized metadata payloads
- Remove schema-invalid architecture->hidden_dim fallback and harden runtime tensor checks
- Align tests and OpenSpec change artifacts for strict fail-fast GGUF behavior

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
@LessUp LessUp merged commit 8805dec into master May 22, 2026
@LessUp LessUp deleted the copilot/harden-gguf-contracts branch May 22, 2026 06:46
LessUp added a commit that referenced this pull request May 22, 2026
- Guard GGUF tensor/array size math against overflow and oversized metadata payloads
- Remove schema-invalid architecture->hidden_dim fallback and harden runtime tensor checks
- Align tests and OpenSpec change artifacts for strict fail-fast GGUF behavior

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
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