Make the two docs a reviewer opens next tell the truth - #13
Merged
Conversation
ARCHITECTURE.md still described a sixth layer of specialist agents exchanging typed proposals under a supervisor, with a decision ledger. That layer was cut weeks ago and does not exist. A reviewer opening the architecture document was being shown a multi-agent system the repository does not contain — the same class of overclaim already fixed three times in the served surface, surviving in the one document nobody had re-read. Rewritten to describe the five layers that exist, with the cut stated explicitly rather than silently deleted. DATA.md omitted Materna entirely. Materna supplies three of the seven workloads in the diagnostic study and the materna-2 exception that keeps Q13 honest, and its code has carried LICENCE = "unverified-host-unreachable" the whole time. The doctrine is that an unverified licence is a stop condition rather than permission to omit the record, and the record was omitted. Bitbrains had a one-line mention with no detail. Both now have a full entry: the canonical Grid Workloads Archive host is unreachable, the traces come from the at-large mirror, checksums are committed, nothing is redistributed, and five of seven workloads depend on them. The distinction that justifies using them at all — depending on a checksummed fetch is weaker exposure than redistributing bytes — is stated rather than assumed, and Alibaba is contrasted as the case that was dropped instead. Also sets the repository description, homepage and topics on GitHub. The description still advertised the original thesis rather than what the project became, and the live URL appeared only in the README, not in the About panel a reviewer reads first.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Final pre-submission read of the documents a reviewer opens after the README. Two of them
were wrong.
ARCHITECTURE.md described a system that doesn't exist
It still listed a sixth layer of specialist agents exchanging typed proposals under a
supervisor, with a decision ledger. That layer was cut weeks ago. A reviewer opening the
architecture document was being shown a multi-agent system the repository does not contain —
the same class of overclaim already fixed three times in the served surface, surviving in the
one document nobody had re-read.
Rewritten around the five layers that exist, with the cut stated explicitly rather than
silently deleted, and with the diagnostic promoted to its own layer since it is the first
thing the product actually does.
DATA.md omitted Materna entirely
Materna supplies three of the seven workloads in the diagnostic study — including
materna-2, the exception that keeps Q13 honest — and its code has carriedLICENCE = "unverified-host-unreachable"the whole time. The project's own doctrine is thatan unverified licence is a stop condition, not permission to omit the record. The record was
omitted. Bitbrains had a one-line mention with no detail.
Both now carry a full entry: canonical Grid Workloads Archive host unreachable (verified
twice), traces fetched from the at-large mirror, checksums committed, nothing redistributed,
and an explicit statement that five of seven workloads depend on them.
The distinction that justifies using them at all — depending on a checksummed fetch is weaker
exposure than redistributing bytes — is now stated rather than assumed, with Alibaba
contrasted as the case that was dropped instead.
Repository presentation
Description still advertised the original thesis rather than what the project became, the live
URL appeared only in the README and not in the About panel, and there were no topics. All
three set.
Gate
ruff, format, mypy, 201 tests — verified by exit code.