[No QA] Docs updates for E/A#94602 agent auto-added as full copilot of creator

[MelvinBot]

Explanation of Change

This updates the Create and Use Agents help article to match the new behavior shipped in #94602: when you create an agent, it's now automatically added as a full Copilot of your own account so it can act on your behalf — you no longer have to add it as a Copilot by hand.

The article on main already documented adding an agent as a Copilot on your own account, but framed it as an optional/manual step. This PR corrects that framing to reflect that the relationship is now created automatically at agent creation, and notes the message shown above the Create agent button.

Files changed

• docs/articles/new-expensify/ai-agents/Create-and-Use-Custom-Agents.md
  • Intro: agent is now automatically added as a full Copilot on your account (was "You can also add…").
  • How to create an Agent: added a step to review the note above the button explaining the agent will be added as a full Copilot of your account.
  • What happens after you create an agent: changed the optional bullet to state the agent is automatically added as a full Copilot, and added a pointer to Account > Security > Copilot: Delegated access to review it.
  • FAQ: reworked "Can I add an agent as a Copilot on my own account?" → "Is an agent added as a Copilot on my own account?" to reflect the automatic behavior.

Source PR

• [Payment due @cretadn22] Make a new agent a full co-pilot of its creator #94602 — [Payment due @cretadn22] Make a new agent a full co-pilot of its creator

---

@yuwenmemon, please review the files changed and confirm they reflect the current behavior. Then mark this PR Ready for review.

[OSBotify]

A preview of your ExpensifyHelp changes have been deployed to https://6a491336.helpdot.pages.dev ⚡️

Updated articles:

• articles/new-expensify/ai-agents/Create-and-Use-Custom-Agents

[github-actions]

Vague UI reference: This step points to "the note above the button" without naming an exact UI element, which conflicts with the Prohibited Language and Deterministic Writing rules in HELPSITE_NAMING_CONVENTIONS.md (avoid phrases like "the button"). A numbered step should describe a concrete user action referencing exact UI labels. Consider either removing this step (the Copilot behavior is already explained in the body/FAQ) or rewording to reference the exact on-screen text, e.g. "Review the on-screen note explaining that the agent will be added as a full Copilot on your account."

[github-actions]

HelpDot Documentation Review

Overall Assessment
This PR makes content-only edits to a single article, docs/articles/new-expensify/ai-agents/Create-and-Use-Custom-Agents.md. The changes update the Copilot messaging: agents are now described as automatically added as a full Copilot on the creator account (previously optional/manual). The edits touch the intro paragraph, the create-agent steps, the What-happens-after list, and one FAQ. The changes are accurate, consistent, and well-written, with no structural or heading changes.

Scores Summary

• Readability: 9/10 - Sentences are clear, concise, and benefit-focused. The new pointer to Account then Security then Copilot: Delegated access gives readers a concrete follow-up. Minor nit: new step 6 (reviewing the note above the button) describes reading rather than an action and is slightly wordy for a numbered step.
• AI Readiness: 9/10 - Full feature names (Agent, Copilot) are used consistently; the messaging change is applied uniformly across intro, steps, list, and FAQ so retrieval stays coherent. No heading hierarchy or YAML frontmatter was changed by this PR, and existing metadata already reflects search intent.
• Style Compliance: 9/10 - UI labels are correctly bolded (Create agent, Account, Security). The revised FAQ heading is an acceptable question-form heading. The navigation reference Account then Security uses path notation consistent with the existing article style.

Key Findings

• The Copilot behavior change (optional to automatic) is applied consistently in all four locations, avoiding contradictory instructions - a strong point for both readers and AI retrieval.
• Adding the review path (Account then Security, Copilot: Delegated access section) is a helpful, concrete improvement over the prior vague permissions phrasing.
• Minor: new step 6 is not strictly action-oriented per the Step Formatting Standard; consider tightening or folding it into step 7.
• The FAQ answer keeps the anchor Learn more about Copilot - descriptive and compliant with anchor-text rules.

Recommendations

• Optional: tighten new step 6 so each numbered step maps to a discrete user action, or merge it into the Create agent step as a preceding note.
• Otherwise no blocking issues; the diff is compliant with the governance guidelines.

Files Reviewed

• docs/articles/new-expensify/ai-agents/Create-and-Use-Custom-Agents.md - Approved with minor optional suggestion.

Note: This is a summary review based only on the proposed diff.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: dd010b986b

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]

Point users to Account > Copilot

This path sends users to the wrong settings page: the current UI exposes Copilot as its own Account menu item (src/pages/settings/InitialSettingsPage.tsx routes delegate.copilot to ROUTES.SETTINGS_COPILOT), and the dedicated Copilot help article also instructs users to go to Account > Copilot. Users trying to review the delegated access added for an agent won't find it under Security; the same incorrect path appears again in the FAQ below.

Useful? React with 👍 / 👎.

[stephanieelliott]

Approved, but will merge this tomorrow when the change hits prod.