Add a standalone Codex Sales demonstration workflow

[nm-openai]

Summary

• Add one self-contained, portable Codex Sales demonstration skill under examples/codex/standalone-sales-demo/SKILL.md.
• Embed the complete Meridian Cloud scenario, polished interactive executive and seller dashboards, scenario data, state-machine response templates, dynamic dates, and a Python-standard-library-only localhost launcher in that single file.
• Register the example in registry.yaml so it can be discovered by the OpenAI Cookbook.

Motivation

Make the complete Sales demonstration reproducible from one GitHub file URL in a fresh Codex session, without installing the Sales plugin, cloning a repository, configuring connectors, downloading packages, or sacrificing the working dashboards and approval-gated customer-follow-up experience.

Validation

• Validated the skill frontmatter and directory with the Codex skill validator.
• Exercised 11 conversation states, optional Northstar deck and outreach branches, weekend-aware dynamic dates, the exact four-field Salesforce review, simulated-save approval, and three connector-availability fallbacks.
• Opened both dashboards in a visible Playwright browser; verified all four executive sections, live Segment and Manager filters, recalculated forecast scenarios, estimate labels, aggregate-only manager guardrails, and all ten seller accounts with no browser JavaScript errors.
• Confirmed the existing Northstar Google Slides URL remains unchanged and customer emails/CRM updates are never sent or written.
• Parsed and validated the cookbook registry entry and verified its existing nm-openai author metadata.

For new content

• I have added a new entry in registry.yaml so that the content can render on the cookbook website.
• I have conducted a self-review of my content based on the contribution guidelines:
  • Relevance: Demonstrates a complete portable Codex sales workflow.
  • Uniqueness: Existing Codex Cookbook examples do not include this connector-free, executive-to-seller workflow.
  • Spelling and Grammar: Reviewed user-facing scenario instructions and response templates.
  • Clarity: Includes a fast-start command, state-machine guide, scenario contract, and exact response templates.
  • Correctness: Launcher, dashboards, filters, dates, state transitions, and approval boundaries were exercised.
  • Completeness: Includes all dashboards, scenario fixtures, source-shaped evidence, and runtime code in one portable skill file.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 85d9096b2f

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Revalidate cached executable assets before reuse

On a shared host, another local user can pre-create this predictable /tmp/portable-sales-demo-<public hash> directory with the expected marker and modified scripts/start_demo_fast.py; this early return then trusts and executes that script under the victim's account. The same problem occurs if cached files are modified after a legitimate first run—the embedded ZIP hash still passes, but the extracted executables are never checked again. Use a securely created user-private directory or verify/re-extract every cached file before executing it.

Useful? React with 👍 / 👎.