Skip to content

Clarify publication convergence error selection in docs #907

Description

@bcdonadio

Observed behavior: The docs say mismatches preserve the original typed contention. Following the candidate A2 fix, a non-expiry identity refusal after a granted retry reports the current contention, while recognized expiry preserves the first. The class, exit1 and fail-closed contract remain accurate.

Expected behavior / safe fix: Clarify the two sentences and align them with current error selection and the separately tracked failed-probe timeout edge. This is a documentation follow-up; no live reproduction needed.

Evidence: Independent Opus5 medium round3 adjudication of Bug783 candidate ed7c920f3c0445bdc478390546a000e7dcd8a1ae; docs/cli.md:172; docs/backend-publication.md:296. Static source review only, no live installation/daemon or exploitation work.

How to reproduce safely: Use the deterministic fixture/test or documentation comparison described above. Never modify production HOME/locks or run a live daemon to demonstrate this.

Environment:

  • Agent: Opus5 medium adjudication, Astra owner
  • Connector: static repository review
  • OS: Fedora Linux

Related #783 / Epic #848. Accepted P2 finding A9; separate native Bug outside S1. Candidate PR link follows at publication. Nonblocking for the original existing-daemon convergence fix.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    documentationImprovements or additions to documentation

    Type

    Fields

    Priority

    Low

    Effort

    None yet

    Projects

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions