Repository navigation
fix(standard): prove memory recall with a name, not a password - #30
Merged
Merged
Conversation
The cap.memory/v1 kit's m2 check stored a Wi-Fi password, so a memory that refuses credentials by design could never be proven. It now recalls an exact token that is not a secret.
This comment has been minimized.
This comment has been minimized.
With no credential left in the kit, it passed trivially.
This comment has been minimized.
This comment has been minimized.
aitianshi
changed the title
aitianshi
added a commit
that referenced
this pull request
Nov 10, 2026
## What Pull request titles become conventional commit subjects, `type(scope): summary`. The repository squash-merges with the PR title as the commit subject, so the title is what `main`'s history shows. AGENTS.md states the rule with its types (`feat`, `fix`, `perf`, `refactor`, `docs`, `test`, `ci`, `build`, `chore`, `revert`) and scopes (the parts of Nullius), and the `/file-issue` skill, which assistants use to open PRs, applies it. No check enforces it. ## Touches docs and site · CI and skills ## Verified - Backend suite `python3 app/tests/parallel.py`, OK. - The open PRs #29, #30, #32 and #34 already use the form. ## Risk Documentation only; earlier commit subjects on `main` are not rewritten. Undone by a revert.
The memory kit's m2 now asks for a name instead of a password and keeps kit_version 1. Write the rule that decision follows: a check that gets stronger or tests something new bumps the version, a rewording does not. The ports table names the memory kit's two checks and that none stores a credential.
This comment has been minimized.
This comment has been minimized.
It is not trivial: the same memory fails m2 on the kit's old wording, the Wi-Fi password, and passes on the boat's name. It keeps a credential from coming back into the kit, and it is the test issue #13 asks for.
This comment has been minimized.
This comment has been minimized.
Collaborator
Author
|
/judge |
🟡 PR judge · 7 of 7 yes · #30 ↔ #13
Acceptance criteria, 2 of 2 met
Checks the script runs itself, 4 of 4 pass
|
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
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.
Closes #13
What
The
cap.memory/v1kit's m2 check stored a Wi-Fi password and asked for it, so a memory that refuses to keep credentials by design failed m2, was never proven, and was never matched to LongMemEval or LoCoMo. m2 still checks recall of an exact, unguessable token, now one that is not a secret: "The boat at the cabin is named tern-river-9." / "What is the boat at the cabin named?". Its label and task instandard.jsonfollow.kit_versionstays 1: the check is not stronger, only no longer a credential, so the existing proofs of keyword-memory and mem0 stay current.Touches
the standard, docs
Verified
python3 app/tests/parallel.py, 1593 tests, OK; the existing conformance test still proves the fixture memory 2/2 on the new wording.test_a_memory_that_refuses_credentials_still_proves_the_memory_port: the fixture memory, changed to drop every session that mentions a password, token, secret, credential or API key, provescap.memory/v12/2. The same test fails againstmain's kit (m2 recalls nothing for the Wi-Fi password), so it also keeps a credential from coming back into the kit.docs/standard.mdnames the memory kit's two checks and that none stores a credential, and the Kit versions table now says when a kit's version goes up, the rule that keepskit_versionat 1 here.Risk
The existing
cap.memory/v1proofs were made on the password wording and are kept as current. A memory that recalled the password but would miss the boat's name would keep a proof it might no longer earn; both bundled memories retrieve by the question's words, which this fact shares, so that is unlikely. Undone by a revert.