Skip to content

Good first evidence: decode challenge_001 in a fresh external runtime #7

Description

@jaden3824

Goal

Run the frozen Urusilla decode challenge in an environment operated outside this project and publish either an exact match, a mismatch, a refusal, or a fallback. Negative and null results are first-class evidence.

Challenge: https://github.com/jaden3824/urusilla/blob/51ff0e3/interop_lab/evidence/challenge_001.md
HF record: https://huggingface.co/datasets/jaden3824/urusilla-interop-lab
HF result thread: https://huggingface.co/datasets/jaden3824/urusilla-interop-lab/discussions/1

Minimum evidence

  • runtime/OS/language and exact dependency versions;
  • whether the receiver/operator had prior Urusilla exposure;
  • exact Capsule URI and SHA-256 verified;
  • frame byte count and SHA-256 verified;
  • decoded typed JSON or the precise failure/refusal;
  • whether deterministic re-encoding reproduces the same frame;
  • whether the project decoder, an independently written decoder, or model-only reading was used;
  • zero secret, private prompt, chain-of-thought, or personal-data disclosure.

Safety boundary

The packet is declarative, unsigned, and non-effect-authorizing. It creates no obligation to adopt or retransmit and grants no persistence, memory, spending, permission expansion, code execution, network action, or external effect. Publication is a separate external action controlled by the submitter.

Claim boundary

Using the project decoder is a useful external runtime check but not an independent implementation. A model decoding one packet is a bounded comprehension observation, not language adoption or general efficiency. Broad post-decode API-input saving remains 0% until stronger evidence changes it.

Post the result here or use the structured interop-test issue form. Codex agents assisted with the fixture and issue wording; the maintainer reviewed and submitted them.

Metadata

Metadata

Assignees

No one assigned

    Labels

    good first issueGood for newcomersgood-first-evidenceBounded evidence task suitable for a first contributionhelp wantedExtra attention is neededresearch-evidenceReproducible favorable, null, or unfavorable research evidence

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions