OmniNode-ai /onex:changelog_audit
Fetch, parse, and classify changelog entries for a target dependency since the last audit date. For each ADOPT_NOW or BREAKING_CHANGE entry, create a Linear ticket automatically. Generate / refresh the unadopted features dashboard.
Usage
/onex:changelog_audit --target claude-code
/onex:changelog_audit --target anthropic-sdk-python --since-date 2026-03-01
/onex:changelog_audit --target github-cli --dry-run
Targets and Changelog URLs
| Target | URL |
|---|---|
claude-code |
https://code.claude.com/docs/en/changelog |
anthropic-sdk-python |
https://github.com/anthropics/anthropic-sdk-python/blob/main/CHANGELOG.md |
github-cli |
https://github.com/cli/cli/releases |
uv |
https://github.com/astral-sh/uv/blob/main/CHANGELOG.md |
kafka-python |
https://github.com/dpkp/kafka-python/blob/master/CHANGES.md |
custom-url |
Provide via --url arg |
Classification Rules
| Label | Meaning | Action |
|---|---|---|
BREAKING_CHANGE |
Removes, renames, or incompatibly changes a feature | Create Linear ticket; grep workspace for usage |
ADOPT_NOW |
New feature / flag / hook we should adopt immediately | Create Linear ticket |
ADOPT_SOON |
Useful improvement worth scheduling | Log to report only |
EVALUATE |
Potentially useful, needs review | Log to report only |
SKIP |
Irrelevant (infra-only, other-platform) | Omit from report |
Keywords that trigger BREAKING_CHANGE: removed, deprecated, renamed, breaking, incompatible, migration required
Keywords that trigger ADOPT_NOW: new flag, new hook, new command, new env var, new tool, new permission
State Files
- Last-audit timestamp:
.onex_state/changelog_audit/<target>.last_audit.json - Audit report:
.onex_state/changelog_audit/<target>-<date>.md - Dashboard:
.onex_state/changelog_audit/DASHBOARD.md
Implementation
Step 1 — Resolve target URL and last-audit-date
import sys, os
plugin_root = os.environ.get("CLAUDE_PLUGIN_ROOT", "")
lib_path = f"{plugin_root}/skills/changelog_audit/_lib"
if lib_path not in sys.path:
sys.path.insert(0, lib_path)
from dispatch import dispatch, AuditTarget
result = dispatch(
target="{{--target}}",
since_date="{{--since-date}}" or None,
dry_run="{{--dry-run}}" == "true",
)
Step 2 — Display result
import json
print(json.dumps(result, indent=2, default=str))
if not result["success"]:
raise SystemExit(1)
Print the audit summary:
- Target + date range audited
- Count per classification label
- Linear tickets created (IDs + titles)
- Workspace usages found for BREAKING_CHANGE entries (file:line list)
- Dashboard location