chore: version skills

[github-actions]

This PR was opened by the Changesets release GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to ori, this PR will be updated.

Releases

qp-skills@4.1.0

Minor Changes

• #112 5d33c34 Thanks @mosobande! - Retrieve applicable project knowledge in planning, delivery, and review without adding a new knowledge store. Require a current plan premortem before Atọ́nà declares readiness or enters Planned. Add selectively loaded boundary-failure review and agent-facing architecture guidance. Introduce Experimental measured optimization with isolated trials and retained-candidate acceptance; keep skill behavioral validation with Kọ Skill and distinguish adjacent improvement tasks in Alárinà. Add a thin native Codex plugin manifest with focused package validation. Keep authenticated host behavior explicitly unverified until real trials establish it. Trim superseded repository docs and reduce the README to current installation, usage, compatibility, and contribution entrypoints.

• #113 ff2d7b6 Thanks @mosobande! - Add existing-codebase quality assessments to atunwo, consolidating the experimental codebase-grade candidate into its core review workflow. Preserve contextual judgment, strengths, failure-path inspection, optional calibrated grades, and coverage limits without requiring a candidate acceptance verdict. Integrate assessment criteria into the shared evidence, challenge, and reporting steps without a separate assessment procedure. Update the review worker to support this result.

  Require pare reductions and atunwo review recommendations to preserve or improve readability. Reject formatting removal and code compression as simplification, and preserve helpful explicitness.

  Clarify idiomatic, readable minimum mechanisms and assembled identity/authority/resource-limit proof in alaga. Evaluate simplification across callers as well as implementation, audit cumulative compression pressure on actual skill loading paths, and assess whether generic return types have an enforced implementation relationship.

  Align pare selection rules, relocation checks, helper examples, and findings with comprehension and maintenance benefits. Allow useful named helpers, longer clearer implementations, and justified retention without requiring structural counts to fall.

• #112 5d33c34 Thanks @mosobande! - Consolidate visual explanation into salaye, technical writing and prose cleanup/pruning into oro-ologbon, and coding scope guards into alaga scope-only mode. Remove the fihanmi, technical-writing, yo-slop, and scope-guard entrypoints; update explicit invocations and project guidance to their replacements. Restore Mermaid and focused HTML explanations and apply visual reasoning within relevant HTML Artifact sections. Keep optimization Experimental, clarify comparative-trial selection, and inline its universal measurement guidance.

  Retire the standalone handoff skill and its routing entry. Preserve concise session-transfer guidance within delegation. Remove the inherited writing-skill license file and metadata as requested.

• #111 19daac1 Thanks @mosobande! - Use one flat native-discovery skill library with optional Pepeye. Preserve all specialist resources and skill names, compose skills directly, make ordinary planning and persistence proportional, and replace third-party installer ownership coupling with verified QP snapshots and exact host links. Existing native-manager installations require explicit migration; source paths change without changing skill invocations.

  Simplify skill composition throughout entrypoints and references to direct use of the named skills, without repeating their methods or prescribing handoff protocols. Keep supporting skills independent of their callers by removing redundant reverse dependencies.

  Use supervised subagents for substantial analysis, research, and expert work in long-running skills, keeping the main context focused.

  Use semantic color in HTML Artifact diagrams and data views, including Mermaid, while preserving contrast, non-color cues, and consistent meaning across themes.

  Restore Alárinà as the QP-specific router with named entrypoints and useful work paths.

  Keep artifact theme and back-to-top controls icon-only with accessible labels; verify back-to-top stays fixed and visible at the bottom right while scrolling.

  Restore Àròjinlẹ̀'s conversational question blocks and explicit ask–wait cycle while delegated fact-finding proceeds independently.

  Keep Fihànmí renderer-neutral; Mermaid guidance remains in HTML Artifact.

  Use the exact inline SVG brand mark from the live Quantipixels Skills header in HTML Artifact.

• #103 04afa81 Thanks @mosobande! - Make Pepeye the sole general main agent and keep its coordination skill focused on delegated work. Remove the competing qp agent; migrate startup selections to qp-skills:pepeye. Load skills on demand, preserve worker supervision and cancellation safeguards, and move native main-agent setup out of the skill. Clarify skill focal points and keep routing and handoff independent of conversational identity.

• #113 ff2d7b6 Thanks @mosobande! - Add focused native planning, implementation, review, research, architecture, and skill-authoring agents for Pepeye coordination across Claude Code and Codex. Add qp-setup for layered user/repository .qp/setting.json preferences for no-op/default, adaptive, or Yoruba communication plus adaptive/inherited or explicit per-host/per-agent model and reasoning selection. Provide collision-safe native agent synchronization without replacing host configuration, permissions, skills, or startup defaults.

• #115 18d138e Thanks @mosobande! - Teach Alága to detect mutable shared-state invariants before implementation and select concurrency mechanisms only when they protect the actual invariant. Add focused database/Spring transaction, retry, replay, and cross-system guidance behind expert counsel, and teach Àtúnwò to test credible concurrent interleavings without treating check-then-act, transactions, or single statements as automatic findings.

• #99 f37802d Thanks @mosobande! - Simplify PR publication and babysitting around their outcomes. Babysitting now includes scoped fixes and publication while watching stays read-only. Load stack coordination only for dependent PRs, retain provider and reconciliation safeguards, and return concise results.

Patch Changes

• #98 9c0662f Thanks @mosobande! - Ship the Quantipixels favicon and logo with a default HTML Artifact base template containing the minimal layout, theme control, and back-to-top control. Consolidate the foundation styles and both default controls into the template, and retain report, filter, and carousel controls as optional assets.

• #111 19daac1 Thanks @mosobande! - Verify the embedded Quantipixels wordmark against the official site asset, preserve its geometry, and keep an exact embedding regression. Its foreground remains legible when printing without backgrounds.

• #100 126b74a Thanks @mosobande! - Simplify premise testing, session handoffs, and issue triage while preserving evidence and authority boundaries. Refocus Kọ Skill on authoring and verification, selectively load governance and resource guidance, and permit useful domain expertise within execution skills. Require an independent purpose for committed documentation and retire duplicate/superseded reports.

• #114 bbad730 Thanks @mosobande! - Clarify that pare's simplification ladder applies to each material candidate, and let pepeye report worker status in prose or a table without mandatory repeated tables.

• #112 5d33c34 Thanks @mosobande! - Consolidate native Codex rejection tests into package integrity checks. Keep shipped safety mechanics in package CI and reserve standing model-evaluation corpora and harnesses for the separate internal eval repository. Remove the duplicate in-package evaluation scenario catalogue without changing skill behavior or runtime verification claims.

• #113 ff2d7b6 Thanks @mosobande! - Clarify test-first implementation as a proof technique within alaga delivery, system and change inspection as scopes of pare, defect-only requests as review restrictions in atunwo, and pruning as an explicit editing objective in oro-ologbon. Preserve read-only boundaries, scope-only delivery restrictions, parity evidence requirements, and existing requests for these behaviors.