Skip to content

feature/Unknown-inline-model-has · L-260926-f3c769 — unknown inline model has no next step - #1275

Closed
lchoquel wants to merge 2 commits into
feature/Strict-method-faultsfrom
feature/Unknown-inline-model-has
Closed

lchoquel wants to merge 2 commits into
feature/Strict-method-faultsfrom
feature/Unknown-inline-model-has

Conversation

@lchoquel

@lchoquel lchoquel commented Sep 26, 2026 •

Copy link
Copy Markdown
Member

A run failing on a model that only the method names, in an inline model setting, was classified as the caller's fault but carried no next step, so every surface fell back to "the message gives the cause". The deck lookup now attaches a change_model user action naming the reference as the method wrote it, which reaches local surfaces and survives STRICT disclosure, and the model panel of pipelex run gives it first with the local-deck remedy after it. A model the deck itself names but does not serve is unchanged: config domain, redacted, no next step of its own.

Closes L-260926-f3c769

🤖 Generated with Claude Code


Summary by cubic

A run failing on a model that only the method itself names, in an inline model setting, is the caller's fault but carried no next step, so every surface fell back to "the message gives the cause". The deck lookup now attaches a change_model user action naming the reference as the method wrote it and the type of model the pipe needs, which reaches local surfaces and survives STRICT disclosure, and the model panel of pipelex run shows it first with the local-deck remedy after it. A model the deck itself names but does not serve is unchanged: config domain, redacted, with no next step.

Closes L-260926-f3c769.

Written for commit a3ebdc8. Summary will update on new commits.

Review in cubic

lchoquel and others added 2 commits September 27, 2026 01:30
A step naming, in an inline model setting, a model the deck neither defines
nor names fails the run as the caller's fault, with no next step: every
surface fell back to "the message gives the cause". The lookup now passes a
change_model user action naming the reference as the method wrote it, which
STRICT disclosure keeps with the message. The model panel of pipelex run
gives that next step first and its local-deck remedy after it, since a local
deck can be what lacks the model; a model the deck names keeps the remedy
alone, and a redacted report with no next step of its own.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A model the deck serves as another type, such as an LLM named by an
image-generation pipe, is also a reference the lookup reports as the
caller's, and "change it to one the model deck serves" pointed at nothing
the caller could find. The detail now names the type the lookup asked for,
from the new ModelType.indefinite_description, and never the type the deck
serves the model as, which would tell a STRICT caller what the deployment
serves.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@lchoquel

Copy link
Copy Markdown
Member Author

Review pass, round 1, profile 4, bar open. Reviewers: cubic, Codex review, Codex adversarial, and the official code-review at level medium. Each of them produced a review of 6f4a7bb.

Fixed: cubic found that the change_model detail named no model type. That was misleading for a model the deck serves as another type, such as an LLM named by an image-generation pipe, and a verifier confirmed the defect. The detail now names the type of model the lookup needs, never the type the deck serves the model as (a3ebdc8). Codex review, Codex adversarial and code-review found nothing.

Deferred: nothing.

round 2 — bar defects — profile 4 (derived) — the round-1 pass at profile 4 fixed under its bar; the ladder puts this pass at defects

@lchoquel

Copy link
Copy Markdown
Member Author

Review pass, round 2, profile 2, bar defects, over the fix commit a3ebdc8. Reviewers: Codex review and the official code-review at level low. Both produced a review, and neither found anything, so nothing was fixed or deferred.

converged — the round-2 pass at profile 2 was clean with full coverage, at or above the required depth of 4; nothing further is warranted

lchoquel added a commit that referenced this pull request Sep 27, 2026
… build

Brings #1275's change_model next step for a model only the method
names. Conflicts: CHANGELOG.md keeps one [Unreleased] above [v0.66.1], with
the STRICT caller-fault entry naming the PipeCondition refusals and the
change_model next step together; error-model.md keeps the PipeCondition row
beside #1275's ModelNotFoundError row; the cli-docs drift ack is re-acked for
the merged tree.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@lchoquel

Copy link
Copy Markdown
Member Author

Superseded by #1276, which carries this branch's commits and landed on dev as a7924f6.

@lchoquel lchoquel closed this Sep 27, 2026
@github-actions github-actions Bot locked and limited conversation to collaborators Sep 27, 2026
@lchoquel
lchoquel deleted the feature/Unknown-inline-model-has branch September 27, 2026 16:46
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant