Skip to content

chore(release): bump VERSION to 1.1.0 and sync manifests for v1.1.0 release - #38

Closed
SHcommit wants to merge 32 commits into
masterfrom
release/v1.1.0
Closed

SHcommit wants to merge 32 commits into
masterfrom
release/v1.1.0

Conversation

@SHcommit

@SHcommit SHcommit commented Sep 6, 2026

Copy link
Copy Markdown
Owner

Summary\n\nCut release v1.1.0. The 31 commits accumulated since v1.0.1 include six feat: commits (Cline CLI adapter, GitHub label taxonomy / labeler / dependabot / Issue Forms / auto-triage), so SemVer requires a MINOR bump; no breaking changes were identified, so MAJOR is not warranted and PATCH would understate the surface change.\n\n## SemVer rationale\n\n| Category | Count | Bump type |\n|---|---|---|\n| feat: | 6 | MINOR |\n| fix: | 2 | PATCH |\n| docs/ci/chore: | 23 | — |\n\n→ Result: 1.0.1 → 1.1.0 (MINOR)\n\n## Version sync (scripts/sync_version.py)\n\nPropagated the 1.0.1 → 1.1.0 bump across:\n\n- skills/adr-toolkit/VERSION (source of truth)\n- skills/adr-toolkit/SKILL.md frontmatter version: line\n- .claude-plugin/plugin.json version\n- adapters/gemini-cli/gemini-extension.json version\n- adapters/antigravity/plugin.json version\n\nadapters/codex/.codex-plugin/plugin.json carries description only — no version field to sync. scripts/sync_version.py --check passes (no drift, no untracked manifests).\n\n## Notable changes shipped in v1.1.0\n\n- Cline CLI adapter (adapters/cline/) — README-only, no manifest; verified against Cline CLI 3.0.61.\n- CLINE.md harness entry pointer + AGENTS.md harness-list / per-model-file policy update. Empirically verified Cline auto-injects the repo-root AGENTS.md into workspace context (model cline-pass/glm-5.2)).\n- GitHub governance hardening: structured Issue Forms, source-controlled label taxonomy, path-based PR labels, new-issue triage, weekly grouped Dependabot updates targeting develop, dormant CODEOWNERS draft.\n- CI supply-chain hardening: ruff + strict dependency-audit gates, Python floor raised to 3.10, Antigravity installer replaced with a versioned SHA-512-verified artifact, release Actions pinned to commit SHAs, PR-title shell expression injection removed.\n- ADR-0017: OSS repository governance & supply-chain hardening decision recorded.\n- Backlog items: ReDoS cross-platform safety, supersede 2-phase atomic rollback, ADR overlap/similarity Eval, weekly maintenance workflows, PR significance bot, adr lint --fix, interactive HTML graph viewer, code drift detection, plus the Medium-priority item to fold Cline into the harness-parity CI job.\n\n## Release flow (per AGENTS.md)\n\n1. This PR (release/v1.1.0 → master) merges.\n2. Back-merge master → develop (follow-up PR).\n3. Tag v1.1.0 from master and push — .github/workflows/release.yml runs pytest + sync_version --check + tag == VERSION, then publishes a GitHub Release with the skill tarball + sha256 + Python wheel/sdist, and publishes to PyPI via Trusted Publisher (continue-on-error: true, tracked in improvements.md).\n4. Delete the short-lived release/v1.1.0 branch.\n\n## Verification\n\n- scripts/sync_version.py --check: passes (no drift, no untracked manifests).\n- git diff --stat: 7 files, +91/-50 — all version metadata, changelog, handoff. No source code under skills/adr-toolkit/scripts/ touched, so the existing pytest matrix coverage applies unchanged.\n- Once merged and v1.1.0 is tagged, the release.yml workflow will run the full test suite + manifest-version-vs-tag check before publishing.\n\n## Examples Impact\n\n- [x] N/A — release-wrap PR (chore(release):), no examples/ change required per PR template.\n\n## ADR Impact\n\n- [x] N/A — records ADR-0017 (already merged in develop via PR #35) in the v1.1.0 release notes; no architectural decision changed in this PR.\n

SHcommit and others added 30 commits September 5, 2026 22:50
Back-merges release/1.0.0 and release/1.0.1 changes (PyPI publishing
pipeline, antigravity harness parity CI, mypy --strict fix, Windows lock
metadata fix) into develop per AGENTS.md git-flow policy (release
branches merge into master and back into develop).
docs: expand improvements backlog with high/medium items
Pins pytest/pytest-cov/mypy/ruff so Dependabot's pip ecosystem has
real versions to track, and adds an E9+F ruff selection scoped to
what already passes today (broader rule sets are a follow-up).

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Clears the codebase for the new lint CI job added in a later task.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TAZPZwzSutmru6kSCEogu6
Records the audit findings (repo public since 2026-08-29 with v1.0.1
released but zero branch protection, develop/master release drift,
missing label taxonomy/dependabot/issue forms/lint job) and the
NOW-tier implementation plan those findings led to.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TAZPZwzSutmru6kSCEogu6
type:/area:/priority:/size: labels plus needs-triage and blocked,
synced from .github/labels.yml by EndBug/label-sync on push to
develop or manual dispatch. Keeps existing default labels rather
than deleting them.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TAZPZwzSutmru6kSCEogu6
…iage

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Lint is scoped to E9+F (already clean, prior commit); broader ruff
rule sets are a follow-up, not bundled here. Dependency audit installs
the pinned dev toolchain directly, not this project's own unpublished
package (pip-audit's --strict and --skip-editable can never coexist,
so the audited environment simply never includes anything editable),
and upgrades pip first to clear pip's own known CVEs.

pytest==8.4.2 carries a real CVE (CVE-2025-71176 / PYSEC-2026-1845,
local /tmp directory hijack) whose fix requires pytest>=9.0.3, which
in turn requires Python>=3.10. Decided (human call, project has no
users yet) to drop Python 3.9 support rather than carry a permanent
ignore-vuln exception: requires-python, classifiers, ruff
target-version, and the CI matrix all move to 3.10 as the floor.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TAZPZwzSutmru6kSCEogu6
The dependency-audit job can't install via `.[dev]` (it must not
install this project's own unpublished package -- see the existing
comment), so its pytest/pytest-cov/mypy/ruff versions are a manual
copy of pyproject.toml's dev extra. Dependabot only updates the
extra, not this workflow file, so flag the duplication for whoever
bumps one and not the other.

Also verified the full CI pytest invocation (with --cov-fail-under=85)
under the pytest==9.1.1 pin, not just the bare test run from the
prior commit's verification.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TAZPZwzSutmru6kSCEogu6
…iew)

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Cline CLI installs skills via the open Agent Skills standard (SKILL.md),
so the adapter is README-only - no manifest. Documents
`cline skill add SHcommit/ADR-toolkit` plus a manual symlink fallback.

Manually verified against Cline CLI 3.0.61: discovery, install, and the
installed script layer (preflight/init/validate) all return ok:true.

Closes #19
The public transition happened 2026-08-29 with v1.0.0/v1.0.1 already
released, but branch protection was never applied -- update the docs
to reflect actual re-verified state instead of the old plan.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TAZPZwzSutmru6kSCEogu6
feat(adapters): add Cline CLI adapter
§4/§8/§9 already state the repo has been public since 2026-08-29 with
branch protection still missing; §2's "확인된 사실" bullets still said
the repo "is private" and that PR template doesn't exist, directly
contradicting those sections. The missing-branch-protection fact itself
was still accurate and is kept -- only the private-repo framing and the
now-false "no PR template" claim needed correcting.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TAZPZwzSutmru6kSCEogu6
# Conflicts:
#	changelog.md
#	handoff.md
#	improvements.md
chore: sync develop with v1.0.1 release work
…ance-hardening

# Conflicts:
#	changelog.md
#	handoff.md
#	improvements.md
…ance-hardening

# Conflicts:
#	changelog.md
#	handoff.md
feat(github): add OSS repository governance & supply-chain hardening
docs: add ADR-0017 for OSS governance & supply-chain hardening
… backlog item

The Cline adapter (adapters/cline/) is manually verified against Cline CLI 3.0.61 only. Unlike the Codex, Gemini, and Antigravity adapters, it is not exercised by the harness-parity CI job on push/PR, so Cline/ClinePass version drift goes undetected. Cline CLI is npm-installable, so the precondition (public package registry) is already met; this change records the gap and prerequisite checks (npm package name/version pin, install-path verification, preflight/init/validate ok:true) in improvements.md and project-roadmap.md, and notes it in changelog.md and handoff.md.

No code or CI workflow changes; ruff/mypy unaffected. test_cline_adapter.py assertions re-run with Python 3.13 (stdlib-only) and pass. Closes the follow-up backlog item identified while reviewing the Cline adapter install state.
docs(backlog): log Cline CLI harness-parity CI coverage gap as Medium backlog item
Cline joins Codex/Claude/Gemini as a listed harness with a thin entry
file that delegates to AGENTS.md, matching the existing CODEX.md /
CLAUDE.md / GEMINI.md shape. AGENTS.md's harness enumeration (line 4)
and the Harness entry files paragraph (line 69) now both name Cline.

The entry file also records an empirical finding: an isolated Cline run
(CLI 3.0.61, model cline-pass/glm-5.2) showed Cline auto-injects the
repo-root AGENTS.md into workspace context at session start — the
reasoning trace explicitly says the full content of AGENTS.md was
provided in the workspace context. So AGENTS.md already reaches every
model routed through ClinePass (DeepSeek, GLM, Kimi, Qwen, ...) without
extra plumbing.

To make the harness-vs-model distinction explicit, AGENTS.md also
states that per-model entry files (DEEPSEEK.md, GLM.md, KIMI.md,
QWEN.md) are intentionally not created: those are model/API providers
reached through a harness (Cline, Cursor, Aider, ...), not harnesses
themselves, so the harness entry file already covers them. This
preempts the recurring false question of whether each new model needs
its own .md.

Tests: added test_cline_entry_file_is_thin_pointer_to_agents_md and
test_agents_md_lists_cline_among_harnesses to tests/unit/test_cline_adapter.py.
All 5 cases re-run with Python 3.13 standalone (stdlib-only) and pass.
ruff/mypy scope: test_cline_adapter.py only.
SHcommit and others added 2 commits September 6, 2026 17:40
docs(harness): add CLINE.md thin pointer and list Cline in AGENTS.md
…elease

Cut release v1.1.0 from develop. The 31 commits accumulated since v1.0.1
include six feat: commits (Cline CLI adapter, GitHub label taxonomy /
labeler / dependabot / Issue Forms / auto-triage), so SemVer requires a
MINOR bump; no breaking changes were identified.

sync_version.py propagated the bump across:
- skills/adr-toolkit/VERSION
- skills/adr-toolkit/SKILL.md (frontmatter version line)
- .claude-plugin/plugin.json
- adapters/gemini-cli/gemini-extension.json
- adapters/antigravity/plugin.json

(adapters/codex/.codex-plugin/plugin.json carries description only; no
version field to sync.)

changelog.md moves the Unreleased block under a new ## v1.1.0 (2026-09-06)
heading summarizing the 31 commits. handoff.md describes the release flow
per AGENTS.md: release/v1.1.0 merges into master, then back-merges into
develop, then v1.1.0 is tagged from master.

Verification: scripts/sync_version.py --check passes (no drift, no untracked
manifests).
@SHcommit SHcommit changed the title release(v1.1.0): bump VERSION, sync manifests, cut release chore(release): bump VERSION to 1.1.0 and sync manifests for v1.1.0 release Sep 6, 2026
@SHcommit SHcommit closed this Sep 6, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant