v1.1.0

arc-ready v1.1.0 is a non-breaking product-quality release. It preserves the canonical artifact contract and Modes A-D while making the skill standards-valid, progressively disclosed, product-form aware, domain-composable, and release-gated by reproducible evidence.

Highlights

• Routes six product forms before applying product archetype, industry, and regulatory overlays.
• Replaces the large domain catalog with a compact router and 37 focused profiles, including Data / Analytics / BI, Manufacturing / MES, Developer Platform / API / SDK, and Research / Lab / LIMS.
• Reduces SKILL.md to a 175-line activation surface that passes the pinned official Agent Skills validator.
• Adds a fresh pre-publication hardening check that closes the late-Critical launch race.
• Adds current OWASP Top 10:2025 web-risk routing.
• Clarifies arc mode versus session kind, canonical shipping primaries versus companion state, and observability installation versus operational maturity.
• Adds immutable CI action pins, stronger lint, Unicode baselining, deterministic evaluations, live-harness cases, and a release-grade validation command.

Evidence

• Deterministic evaluations: 14/14
• Operational dogfood smoke: 12/12
• Independent Codex live-harness cases: 100/100, every gate invariant passed
• Official pinned skills-ref validate: passed against the absolute repository path
• Inherited domain extraction comparison: 0 mismatches across 33 moved profiles
• Protected meta-linter check: passed on the PR and merged main

See CHANGELOG.md, EVALS.md, and the v1.1 migration notes.

v1.0.2

[1.0.2] - 2026-05-30

Patch release. Resolves the two items v1.0.1 deferred and adds a third lint guard. No discipline change, no new failure-mode patterns, no artifact-path contract change.

Added

• references/building/login-pages.md and references/building/registration-pages.md: the 78K login-and-auth-pages.md split at its existing H1 boundary into its two constituent documents (login / sign-in surfaces; registration / sign-up surfaces) so an agent loads only the relevant half. Cite sites updated in SKILL.md (sub-step 23 and the load-on-demand table) and references/building/auth-and-rbac.md; the two intra-document back-references now name the sibling file. domain-considerations.md (125K) is intentionally left whole (it is a grep-friendly per-domain lookup catalog).
• scripts/lint.sh gains a reference-citations check: every references/<basename>.md inline-code or prose citation inside references/ must name a real reference. Reference basenames are globally unique, which makes this tier-agnostic form the dominant in-repo convention (~200 citations); the check guards it against a rename or typo.

Changed

• AGENTS.md cross-references section now documents the actual two-form citation convention (file-relative [..](..) links guarded by relative-links-resolve; tier-agnostic references/<basename>.md citations guarded by reference-citations) instead of a ../tier/bar.md convention the corpus used exactly once.
• Reference-budget guidance (AGENTS.md, SKILL.md) updated for the new catalog size (167 files) and to name domain-considerations.md as the remaining split candidate.

Notes

• The em-dashes in the inherited building/ references (53 files) are intentionally left per the faithful-copy rule; the lint continues to exempt reference files from unicode-clean.
• references/building/security-setup.md's apparent duplicate ## Reporting a vulnerability heading was confirmed a false positive (the second is inside a fenced code example), so no change was made.

Why a patch, not a minor

Structural and tooling consistency only. The file split is a faithful bisection at an existing document boundary (no prose rewritten beyond two back-reference pointers); the lint addition guards an existing convention. No new content, no contract change.

v1.0.1

[1.0.1] - 2026-05-30

Patch release. Contract-consistency and mechanical-correctness fixes across the orchestrator, the dogfood smoke test, the reference cross-links, and a repo-wide documentation-drift audit. No discipline change, no new failure-mode patterns, no artifact-path contract change.

Fixed

• Critical-finding gate grep (SKILL.md): the documented gate read the line before each severity: critical for its status (grep -B 1), so an open Critical never blocked launch. Corrected to scan forward (grep -A 1) and pinned the canonical FINDINGS.md finding shape (a severity: line followed by a status: line). The dogfood smoke test (Test 9) already used correct logic; the documented snippet now matches it.
• Resume-protocol drift check (SKILL.md): the snippet used ${tier^^} (Bash 4+), which fails on the Bash 3.2 the project targets (macOS default). Replaced with a tr-based uppercase.
• Drift detection missed the architecture tier (SKILL.md, scripts/dogfood-smoke.sh): the loop variable architecture never matched the ledger label (ARCH), so architecture-tier drift went undetected. Added the architecture to ARCH label mapping in the resume snippet and both smoke-test drift checks.
• Dogfood smoke Test 1 was a vacuous pass (scripts/dogfood-smoke.sh): the happy-path drift check used a malformed, case-mismatched regex that matched nothing, so the test could not fail regardless of disk state. Rewritten to mirror Test 2's correct uppercased check.
• Stale template version (SKILL.md): the PROGRESS.md schema example hardcoded ## Skill version: 0.1.1. Bumped to the current version, now guarded by the new skill-version-body lint check.
• Broken reference cross-links (references/): rewrote 14 RESEARCH-2026-04.md markdown links in references/planning/ to ../shared/RESEARCH-2026-04.md (the file lives in references/shared/), including one anchored link. Fixed two self-directory ORCHESTRATORS.md links in references/shared/RESEARCH-2026-04.md. Cross-reference corrections only; reference prose is unchanged.

Changed

• Stack artifact filename reconciled to .stack-ready/STACK.md repo-wide. The Tier 1 prose, the Tier 1.4 gate, the output instruction, the README artifact map, and the resume example said DECISION.md, contradicting the grep tests, the worked example, the dogfood, and the v0.1.1 decision record. Swept DECISION.md to STACK.md across SKILL.md, README.md, the orchestration sequencing rules and PROGRESS schema, pillars-integration.md, the stack-tier deep-dives, six shipping references, and MIGRATION.md (24 occurrences across 14 reference files plus the root docs). STACK.md is the canonical write name; DECISION.md is still accepted on import for externally-authored artifacts; STATE.md remains the ongoing-work file. The historical provenance mentions in references/shared/RESEARCH-2026-04.md are left intact (they describe the source skill). This aligns drifted documentation to the already-canonical name; it does not move the contract.
• Artifact-map alignment across README.md, MIGRATION.md, and SKILL.md: the repo artifact now reads .repo-ready/SCAFFOLD.md (scaffold mode) with .repo-ready/AUDIT-REPORT.md noted for Mode B audits; the deploy and observe primaries now read .deploy-ready/DEPLOY.md and .observe-ready/OBSERVE.md (previously simplified to STATE.md in README and MIGRATION), matching SKILL.md's produces table.
• Reference-budget guidance (AGENTS.md, SKILL.md, CONTRIBUTING.md): the "~80 references at 5-15K" claim was inaccurate (the catalog is ~165 files, several well over 25K). Reworded to reflect reality and to name domain-considerations.md and login-and-auth-pages.md as split candidates.
• Documentation drift fixes: corrected the CODE_OF_CONDUCT.md enforcement links from the stale aihxp/repo-ready slug to aihxp/arc-ready; added the two new lint checks to the MAINTAINING.md lint-checks table; refreshed the 0.1.5 version placeholder in the bug-report issue template to 1.0.1; clarified in MIGRATION.md that trigger-disambiguation.md is retained as a reference rather than removed.

Added

• scripts/lint.sh gains two checks: relative-links-resolve (every markdown link to a real reference resolves from the linking file's directory) and skill-version-body (SKILL.md body version strings match the frontmatter). Both run in CI on push and pull request.

Why a patch, not a minor

Bug-fix and consistency release. No new content, no new failure-mode patterns, no change to the discipline or the artifact-path contract. The stack-filename reconciliation aligns drifted documentation to the STACK.md name the dogfood and the v0.1.1 decision record already established.

v1.0.0

Stable contract release. arc-ready now declares the combined canonical artifact contract and aihxp/pillars memory contract stable for downstream agents and orchestrators. Includes the required Pillars memory layer, source-backed Pillars smoke coverage, README stability promise, migration guidance, and 9/9 dogfood smoke validation.

v0.1.6: operational validation pass

Closes the remaining gaps from the v0.1.5 evaluation: untested resume-protocol heuristic, no end-to-end smoke test, oversized description for Cursor's per-rule limit. Adds CI integration of the smoke test.

Added

• scripts/dogfood-smoke.sh: 7-test operational smoke test covering drift-detection (happy + synthetic), next-sub-step heuristic, artifact-path contract reachability for all eleven .<tier>-ready/ paths, AGENTS.md emit-respect (existing + absent cases), critical-finding gate. 7/7 pass against a synthetic mid-arc project. Wired into CI on push and PR.
• .cursorrules (1,754 chars; under Cursor's 2KB recommendation): slim trigger-surface adapter for Cursor users. Points at the full SKILL.md. The full description (2,558 chars) is fine for Claude Code, Codex, Antigravity, Pi, OpenClaw.

Changed

• SKILL.md resume-protocol shell snippet: walks tiers in dependency order (1.1 -> 1.2 -> ... -> 3.4), treats skipped as complete, resolves next-sub-step against dependency-order traversal not file-order pattern matching. Smoke test verifies it returns 1.2 correctly when PRD is imported and ARCH is in-flight (the bug v0.1.5 acknowledged).

Validated

Check	Result
Description size	2,558 chars / ~639 tokens; under 4KB practical ceiling for Claude Code / Codex / Antigravity
Cursor adapter size	1,754 chars; under 2KB Cursor recommendation
Trigger collision (static)	63 unique domain-specific triggers; no generic single-word triggers; low collision risk
Operational smoke test	7/7 pass

Still untested (requires live harnesses; out of scope for build script)

• Functional agent execution end-to-end on a real or fictional product (a session, not a script).
• Live trigger routing on Claude Code / Codex / Cursor / Windsurf (empirical measurement).
• Real cache cost across the eight compatible_with harnesses.

Why a patch

Refines an inline heuristic; adds a smoke test + adapter; no discipline change. arc-ready remains faithful consolidation.

v0.1.5: repo-hygiene pass

Closes the C-territory gaps from the post-build self-evaluation. No discipline change.

Added

• CODE_OF_CONDUCT.md (Contributor Covenant).
• Issue templates at .github/ISSUE_TEMPLATE/: bug report, feature request, taxonomy question.
• PR template at .github/pull_request_template.md with discipline / versioning / lint / cross-reference checks.
• Dependabot at .github/dependabot.yml for weekly GitHub Actions update PRs.
• Plugin marketplace entry at plugins/arc-ready/.claude-plugin/plugin.json. Uses symlinks to the repo SKILL.md and references/ to avoid content duplication.
• README quickstart with concrete first command and status badges (lint, release, license).

Changed

• SKILL.md Tier 2.2 navigation: six-cluster grouping (Foundations / Quality / Polish / Scale and i18n / Engagement and ops / Integration and testing) before the 35 numbered sub-steps. Substantive content unchanged.

Smoke-tested

Resume protocol simulated against synthetic .arc-ready/PROGRESS.md. Drift detection correct; next-sub-step heuristic identified as a v0.1.6 refinement target.

Why a patch

Repo-hygiene additions and a navigation-aid edit. Faithful consolidation preserved.

v0.1.4: content-fidelity gap closed (TRIGGER-DISAMBIGUATION ported)

Closes a content-fidelity gap from v0.1.0. Audit comparing arc-ready against the source eleven-skill suite revealed one missing reference: the hub's references/TRIGGER-DISAMBIGUATION.md.

Added

• references/orchestration/trigger-disambiguation.md (99 lines): ported verbatim from aihxp/ready-suite/references/TRIGGER-DISAMBIGUATION.md with the canonical-skill column rewritten to arc-ready's tier vocabulary. Disambiguation table covers "set up CI" vs "CI/CD pipeline", "ADR" (Tier 1.2 vs 1.4), "trust boundaries" (declare vs verify), "runbook" (Tier 3.2 vs 3.4), "audit" (Tier 2.1 vs 3.4), "performance" (Tier 2.2 vs 3.2), the production-ready overload, and the not-in-scope list. Registered in SKILL.md references table.

Audit result

After this release: 164 of 164 expected source reference files present in arc-ready, zero missing.

Intentional drops (all hub-only, multi-repo era; not applicable to a single skill):

• SUITE.md (byte-identical-across-12-repos collection map)
• install.sh / uninstall.sh (Agent Skills install is standard)
• plugins/ (Claude Code plugin marketplace)
• PLUGIN-RESEARCH.md (research for the plugin marketplace)
• per-skill SUITE.md / README.md / CHANGELOG.md / CONTRIBUTING.md / SECURITY.md / LICENSE (consolidated to one each)

Why a patch

Closes a documented gap. No new patterns; no discipline change. Faithful consolidation.

v0.1.3: standalone-repo cleanup

Sweep release. arc-ready had no operational dependencies on the eleven sibling repos but inherited reference content still mentioned sibling-repo paths in prose. After this release, zero references in references/ mention sibling-repo paths.

Fixed

• Inherited prose production-ready/ORCHESTRATORS.md rewritten to references/shared/ORCHESTRATORS.md across all reference files. The ORCHESTRATORS content is in arc-ready; citations now point inside arc-ready.
• Bare references/<file>.md paths (assuming the source skills' flat references/ layout) rewritten to tier-prefixed paths in references/orchestration/*.md and same-tier sibling form in references/building/questioning.md.
• Cursor-handoff prose in references/orchestration/handoff-protocols.md rewritten from the eleven-skill-suite assumption to arc-ready's single-skill reality. Historical inter-skill handoff pattern preserved as annotated reference.

What remains

Sibling-repo mentions in arc-ready after this sweep:

• MIGRATION.md: the intentional former-path -> new-path migration matrix.
• README.md, CHANGELOG.md, SECURITY.md: predecessor acknowledgment.

These are not coupling; they are documentation of lineage.

Why a patch, not a minor

Path-rewrite cleanup. No content discipline change. arc-ready remains faithful consolidation.

v0.1.2: bug fixes (CI, hardcoded paths, broken cross-link)

Bug-fix release. No discipline change, no new content.

Fixed

• CI lint workflow no longer fails on push. tag-release-parity is now run only on schedule and workflow_dispatch (releases are created after the tag push that triggers CI; the daily schedule catches missing releases). Other lint checks run on push and PR.
• Hardcoded local-clone paths (/Users/hprincivil/Projects/<skill>-ready/...) in references/planning/trust-boundaries.md and references/shared/RESEARCH-2026-04.md rewritten to skill-relative references. These were inherited from source authoring and survived the v0.1.0 cross-reference sweep.
• Broken external link in references/shared/RESEARCH-2026-04.md from aihxp/production-ready/blob/main/SUITE.md to aihxp/ready-suite/blob/main/SUITE.md (the canonical hub) with predecessor annotation.

v0.1.1: SKILL.md expansion + dogfood-aligned artifact map

Patch release that expands SKILL.md to the spec target (1,500-2,000 lines) and aligns the artifact map with the dogfood at aihxp/ready-suite-example.

SKILL.md expansion

From 842 to 1,570 lines. New content (all faithfully consolidated from the source eleven-skill suite, no v2 additions):

• Mode B routing: ten-step decision tree mapping the current project state to the next tier sub-step.
• Mode C audit: per-tier audit procedure with severity vocabulary, audit-output canonical paths, and verdict semantics (PASS / PASS WITH FINDINGS / BLOCK).
• Mode D multi-repo: dispatch protocol with byte-identical-collection-map invariant.
• Per-tier inline grep tests: mechanical checks for PRD, architecture, roadmap, stack, repo, production, deploy, observe, launch, harden.
• Per-harness integration: Claude Code, Codex, Cursor, Windsurf, Antigravity, Pi, OpenClaw, generic chat frontend.
• Worked example: condensed Pulse arc walkthrough (the dogfood, narrated tier by tier).
• Symptom-to-pattern table: "if X is happening, look for failure mode Y" lookup.
• Critical-finding gate logic: Tier 3.4 -> Tier 3.3 with grep test.
• AGENTS.md emit-and-respect rules: across Tier 0 sub-step 6a and Tier 2.1 sub-step 14.
• Resume protocol elaborated: shell-snippet examples and disk-wins semantics.
• Per-tier STATE.md schemas: PROGRESS.md plus per-tier session-state shapes.
• "What good looks like" excerpts: passing-quality examples for every tier (Problem section, ADR, roadmap row, stack recommendation, scaffold, vertical slice, canary, SLO, hero, hardening finding).

Artifact map (dogfood-aligned)

The source eleven-skill suite produces multiple artifacts per tier (e.g., .deploy-ready/ has DEPLOY.md, PLAN.md, TOPOLOGY.md, STATE.md); v0.1.0 was documenting one canonical filename per tier. v0.1.1 reflects the multi-artifact reality:

• .stack-ready/STACK.md (canonical) alongside DECISION.md variant.
• .repo-ready/SCAFFOLD.md (canonical) alongside AUDIT-REPORT.md (Mode B audit output).
• .deploy-ready/{DEPLOY,PLAN,TOPOLOGY,STATE}.md.
• .observe-ready/{OBSERVE,SLOs,INDEPENDENCE,STATE}.md.

The dogfood at aihxp/ready-suite-example verifies cleanly against arc-ready's tier dispatch with these names.

Why a patch, not a minor

Content deepening, not contract change. Discipline unchanged. No new failure-mode patterns. arc-ready remains faithful consolidation, not v2.