Skip to content

Unify template instructions around the actual plugin contract #15

Closed

Description

@cchacs

Problem

README describes a complete working plugin, AGENTS.md omits the required screen manifest gate, and CLAUDE.md separately corrects AGENTS.md. The manifest also advertises visualization without implementing that renderer. An agent following only one guide can make incorrect decisions.

Acceptance criteria

  • Make AGENTS.md the concise authoritative development guide, with other agent files referring to it without contradictory corrections.
  • Align README and per-plugin documentation with the corrected runnable example.
  • Distinguish normative specification requirements from observed host behavior, including folder/id handling.
  • Identify the applicable host/spec repositories and a reproducible reference revision for contract verification.
  • Include exact setup/test commands and task completion criteria.
  • Explain what must be renamed when creating a plugin.

Coordinate with the runnable-example issue; avoid broad documentation expansion or speculative upstream workflows.

Activity

  1. pullfrog commented on Oct 13, 2026

    @pullfrog
    Contributor
  2. pullfrog commented on Oct 17, 2026

    @pullfrog
    Contributor

    Unified the template docs around the actual plugin contract. PR: #18

    AGENTS.md is now the authoritative guide: it names the spec and Host repos with pinned revisions (get-flashbacks/feedBack-plugin-spec@3d653d1, get-flashbacks/feedBack@ff8984a), separates normative spec rules from observed Host behavior (folder/id handling and the screen-mount gate), adds the copy/rename checklist, and states exact setup/completion criteria. CLAUDE.md no longer carries corrections to it; the READMEs drop the unsupported folder/id "will not load" claim and correct the settings-store path.

    Verification: pytest -q → 21 passed; node --check and the custom-checks.yml manifest validation pass; every cited contract claim was read off the pinned revisions. Docs-only diff, so no behavioral surface to review beyond citation accuracy.

    Task list (7/7 completed)
    • Rewrite AGENTS.md as the authoritative guide: contract sources + pinned revisions, spec-vs-Host distinction, rename checklist, completion criteria
    • Fold CLAUDE.md corrections into AGENTS.md and make CLAUDE.md defer without contradicting
    • Align root README folder/id section and spec links with the corrected contract
    • Align my-plugin/README folder/id note and spec links
    • Update CHANGELOG.md entries
    • Run tests + local custom-checks validation
    • Self-review with reviewfrog, commit, push, open PR

    Pullfrog  | View workflow run | via Pullfrog | Using big-pickle (free) | 𝕏

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

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions