fix: multi-issue fixes across fw-dev-tools skills

[rapchic]

Summary

Multi-issue fixes across fw-dev-tools skills addressing CDN URLs, broken links, stale versions, Platform 2.x removal, lint output bugs, code deduplication, skeleton expansion, and new documentation.

Changes:

• cross-scenarios.md + macos.md — latest.tgz → latest-v24.tgz, v0.39.0 → v0.40.1
• cross-scenarios.md Scenario 3 — rewritten from "Downgrade to Legacy FDK" (Platform 2.x) to "Modernize Older App to Platform 3.0"; TOC, Testing Matrix, Quick Reference updated
• fw-publish/SKILL.md — removed broken references/openai-server-mcp-tools.md links
• fw-review/scripts/common.js — extracted shared walkFiles + IGNORED_DIRECTORIES
• fw-review/scripts/{unused-library-imports,external-import-sources,freshworks-css-only,https-imports,oauth-config-usage}.js — deduplicated walkFiles via common import; fixed undefined line number in unused-library-imports
• hybrid-skeleton/ — expanded manifest with events, server with multi-handler SMI pattern, app.js with client.request.invoke()
• fw-app-dev/references/freshservice-requested-items.md — new reference doc for requested_items API

Type of change

• Skill content (SKILL.md, command file, rule file)
• Reference / template file
• Plugin manifest (.claude-plugin, .cursor-plugin, .codex-plugin, skills/*/.cursor-plugin)
• Installer (installer/)
• Test suite (tests/, installer/tests/)
• CI / docs / infra only

---

Test checklist

Static tests (required for all PRs)

cd tests && npm install && npm test        # 140 static tests — runs in CI

• All 140 static tests pass locally

If installer/ changed:

N/A — installer not modified.

LLM evals (required when SKILL.md, a command file, or .meta.template.json is modified)

67 behavioral scenarios in tests/skill-eval.test.js. Run in an agent session (recommended, no API key):

"Run the skill evals"

Or with API key: cd tests && ANTHROPIC_API_KEY=sk-... npm run eval

• Reviewed / updated scenarios in tests/skill-eval.test.js (and tests/run-inline-eval.mjs if adding doc-regression checks)
• Added new scenario(s) for any new behavioral rule or gate
• 67/67 scenarios pass (or explained in summary why a failure is acceptable)
• Attached tests/eval-report.html below (or linked in PR description)

Repo hygiene (when applicable)

• AGENTS.md rules/commands inventory updated (if skills/*/rules/ or skills/*/commands/ changed)
• tests/TESTING.md scenario table updated (if eval scenarios added/renamed)
• docs/engine-matrix.md updated (if fw-setup toolchain pins changed)
• node scripts/bump-version.mjs run after root package.json version bump (syncs plugin manifests + SKILL.md)

Security

• No secrets, tokens, or real credentials in the diff (use placeholders in docs)
• If dependencies changed: npm audit --audit-level=high passes in installer/ and tests/

Not applicable

• This PR does not change installer/ — installer test checklist skipped
• This PR does not add or change dependencies — audit checklist skipped

---

CI (runs on PR automatically)

Check	What
Static skill tests	tests/npm test
Installer tests	installer/npm test (Node 24)
Dependency review	New deps — high/critical CVEs blocked
npm audit	installer/ + tests/ lockfiles
Secret scan	gitleaks (full git history)
CodeQL	Org-level (if enabled)

LLM evals and tests/e2e.sh are local / agent session only — not in CI.

---

Eval report

[github-actions]

Dependency Review

✅ No vulnerabilities or license issues or OpenSSF Scorecard issues found.

Scanned Files

None