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@1.1.0

Minor Changes

• #8 c067486 Thanks @mosobande! - Make Alaga the builder for one supplied job with one or many delivery units. Alaga now selects useful capabilities from current evidence, composes specialist-owned results through one job envelope, returns overlapping work to an active ancestor instead of invoking it again, preserves explicit authority and candidate identities, continues safe independent work through a confirmed continuation boundary and partial blockers, and closes only after job-level proof and the review required for each candidate type. Ko Skill owns agent-skill verification. Qualifying jobs keep one concise living HTML report under a deterministic structural, continuity, authority, and risk threshold; Alaga owns job-state decisions, the report is their authoritative durable record, and HTML Artifact owns its representation and file lifecycle.

• #3 146048c Thanks @mosobande! - Add Amọ̀ṣẹ́ as the shared owner for domain-model clarification, evidence-backed .learnings, authorized .nongoals, and lightweight ADR lifecycle. Distinguish durable project exclusions from deferrals, implemented behavior, and task-local boundaries before updating .nongoals. Add Irinṣẹ as the shared owner for companion-tool selection, setup, safe bounded use, verification, and removal, with progressive references for tldr-code, ast-grep, Semgrep, and IntelliJ MCP. Replace the superseded Alakowe and QP Setup routes, keep ordinary documentation with its outcome owner, and validate skill behavior through exact-candidate headless and independent sessions. Make Ko Skill gate every author-owned change on behavioral benefit, write descriptions for skill selection, and use examples only when they improve agent behavior or proof.

• #8 c067486 Thanks @mosobande! - Clarify Alarina's build, implementation, and independent-outcome routing. Preserve direct specialist builders when one published owner fully covers the artifact, route skill validation to Ko Skill, keep prose for selection-critical overlap and authority rules, use one complete route table for simple recommendations, make triage supplied-evidence-first, and reconcile the public catalog with the current Irinṣẹ and Olofofo inventory.

• #5 2e18a79 Thanks @mosobande! - Add seda-pr and wo-pr for PR and MR publication and stewardship. seda-pr commits scoped work, pushes the current branch without force, and creates or updates a ready item. wo-pr monitors an open item and handles bounded branch, CI, and feedback work. Neither skill approves or merges.

• #6 d8df677 Thanks @mosobande! - Make ko-skill validation proportionate: check structural surfaces directly, use the smallest fresh-session proof that resolves material behavioral uncertainty, and stop for direction before validation scope expands materially. Require behavior-preserving compression across the complete skill and linked references, semantic rule-owner comparison, compact procedural workflows, and evidence that clarity and safety remain intact. Make seda-pr invocation authorize bounded staging, commit, and non-force push while preserving its zero-context third-party narrative. Require seda-pr and wo-pr to reject custom provider hosts before network contact until the exact normalized host receives separate trust, and prohibit fixed readiness windows in wo-pr. Correct the Claude Code instruction import and keep the README release claim within the changelog's actual scope.

• #5 7645b8b Thanks @mosobande! - Add triage-issue for local-first assessment of one issue or bug report before implementation. Supplied evidence is the default boundary; source inspection, GitHub or GitLab reads, and one evidence-backed provider comment require separate explicit authority. Confirmed handoffs use durable behavioral briefs, matching local artifacts resume established facts without repeated questions, and authorized source reads check existing behavior and prior project decisions by domain concept.

  Remove committed prompt eval suites and the repository skill validator. Verify skill metadata, the public catalog, and Alarina routing directly during exact-candidate review. Material behavior now uses exact-candidate headless sessions and independent scenario review outside repository source.

  Replace the Wo PR state engine with a small observer checkpoint. Remove its deterministic state-engine test only with the deleted watch_core.py owner; retain and run the provider normalization tests. The removed fixed settle window, persistent lease and takeover protocol, stored mutation authority and failure classifications, and claimed-feedback state machine are obsolete. Current lifecycle, authority, provider, retry, receipt, and readiness expectations live in the Wo PR skill and provider references.

• 0b2ab74 - Add olofofo as a quiet default session companion with optional global and project EMI steering, one adaptive global HTML artifact, proportionate evidence-backed quality nudges, curated global OGBON wisdom, explicit title permission, and no workflow-controller role. OGBON stays a compact, selectively loaded, conflict-safe evidence store; EMI remains user-controlled steering and project learning remains with Amọ̀ṣẹ́. Include a narrow portable global activation payload without adding a daemon, hook, or shared setup runtime.

  Replace the unreleased tunmo-pr candidate with salaye, an explicit conversational protocol for exploring, explaining, investigating, researching, analysing, or evaluating ideas, plans, decisions, documents, and code candidates. Ṣàlàyé contributes to Olofofo's active artifact without owning a second session or report lifecycle.

• #8 c067486 Thanks @mosobande! - Clarify that Seda Ticket defaults to outcome-complete vertical slices and permits non-vertical tickets only with a necessary, independently testable exception and a named green integration boundary. Make Atona prefer deep modules at clean seams, apply the deletion test, and reject speculative or pass-through modules without a proven ownership, integration, lifecycle, policy, or testing reason.

• #6 174fc7b Thanks @mosobande! - Refocus html-artifact on translating supplied material and design specifications into checked visual HTML without originating content, product or editorial design, candidates, recommendations, or decisions. Default to an unfamiliar third-party audience with no prior topic context. Add resilient Tailwind and Mermaid enhancements, semantic code diffs, report and supplied-variant patterns, proportionate verification, portability status, and direct-path delivery through concise branch references. Omit default-prompt metadata.

Patch Changes

• #7 73e84f9 Thanks @mosobande! - Deepen internal skill modules without changing their public outcomes: localize QP Code Review provider publication, split Amọ̀ṣẹ́ durable destinations, and move Atona delivery tracking and status-specific handoffs behind direct conditional references. Restore Ko Skill guards for exact reference triggers, cohesive conditional rules, purposeful and compressive examples, exact-candidate proof, and the limits of structural validation. Compress high-value guidance in Alaga, Handoff, HTML Artifact, Simplify, and TDD without weakening its owning rules.

• #10 50cadb3 Thanks @mosobande! - Harden QP Code Review provider operations with explicit custom-host trust, credential isolation, complete GitHub pagination, exact-head refresh before every write, and durable per-write readback receipts. Reconcile the public 1.0 changelog with the current Alaga, Irinṣẹ, and Ko Skill boundaries.

• #4 7426359 Thanks @mosobande! - Add a source-aware global uninstall script with batch removal and post-removal verification, and document its executable command.

• #4 7426359 Thanks @mosobande! - Require TDD workflows to remove temporary smoke proof after green unless it provides distinct durable regression coverage.

• #4 7426359 Thanks @mosobande! - Make Arojinle and Iwadi follow active delegation policy, and give existing repository research conventions precedence over Iwadi's fallback path.

This file records released, user-visible changes to QP Agent Skills. Version 1.0.0 is the clean public baseline. Earlier internal history is intentionally not included.