Repository navigation
[No QA] Add new Agent explanatory articles and restructure AI Agents hub - #103119
Conversation
Ports the AI Agents docs refactor from claude-docs-agentRuleSuggestions onto this branch, keeping this branch's content wherever the two conflict: - Rename Create-Agent-Rules -> Automate-Workflows-with-Agent-Rules, Create-and-Use-Custom-Agents -> Create-and-Use-Agents, How-to-Write-Agent-Rules -> How-to-Write-Agent-Instructions - Use Learn-About-Agents as the overview article - Add the RuleBot-as-approver section and FAQ, and generalize the writing guide to cover both Agents and Agent rules - Update cross-links across the AI Agents set and inbound links from other articles - Add redirects for the three renamed articles - Leave AI-Agent-Capabilities.md and AI-Agent-Triggers.md unchanged Co-authored-by: Stephanie Elliott <[email protected]>
Add scope and setup columns, Concierge workspace setup and bulk updates, agent messaging points, and fix links, casing, and frontmatter. Co-authored-by: Stephanie Elliott <[email protected]>
Updated the order of the article to reflect new structure.
Updated the link to the article on how agents work.
Added order metadata for article organization.
Updated the link text for learning about agents to improve clarity.
Updated the link to the section on how agents use instructions for better clarity.
Added order field to the Agent Capability Reference.
Added order field to the Agent Trigger Reference documentation.
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
Concierge reviewer checklist:
For more detailed instructions on completing this checklist, see How do I review a HelpDot PR as a Concierge Team member? |
|
@mkhutornyi @VictoriaExpensify One of you needs to copy/paste the Reviewer Checklist from here into a new comment on this PR and complete it. If you have the K2 extension, you can simply click: [this button] NOTE: It looks like |
Added contentType and order metadata to the Agent Trigger reference.
|
@MelvinBot can you apply the rules defined in |
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
|
I reviewed all 12 edited articles against
Fixes that apply to most articles:
Changes needed in each articleLearn-About-AI-in-Expensify.md
How-to-Write-Agent-Instructions.md
Automate-Workflows-with-Agent-Rules.md
AI-Agent-Capabilities.md and AI-Agent-Triggers.md
All articles already meet the guidelines on these points:
|
|
@MelvinBot please push the changes suggested here: #103119 (comment) |
stephanieelliott
left a comment
There was a problem hiding this comment.
Clarity pass on the difference between Agents, Agent rules, and RuleBot, and between Agent templates, Agent instructions, and suggested Agent rules. AI-Agent-Capabilities.md and AI-Agent-Triggers.md are intentionally left alone.
Main themes in the inline suggestions:
- Instructions are the common thread. All AI Agents work from instructions, and Agent rules are RuleBot's instructions. Templates and suggested rules are pre-written instructions for each kind of Agent.
- Observer vs. approver. As an observer, RuleBot flags issues and asks questions. As an approver, it approves, rejects, edits and forwards. "Reject" was missing from RuleBot's action list.
- Personal Agents work on your own account until they're invited to a workspace as an approver.
- New FAQ: "Should I use a personal Agent or RuleBot as an approver?"
Outside the diff, so I couldn't add suggestions:
- Create-and-Use-Agents.md: The first FAQ has a broken link:
[...]([url](/articles/...)). - Use-Agent-Templates.md:
- The keywords list "Splitter Sam", but the article covers "HotelSplitBot".
- The heading "What Description Dan Does" is capitalized differently from the other headings.
- The intro says "prebuilt personal Agents". Consider "starting points for creating a personal Agent".
- Automate-Workflows-with-Agent-Rules.md:
- The FAQ "Do I need to create or manage RuleBot? … managed by Expensify" conflicts with the remove/demote/delete section.
- Step 5 says "select Edit to write your own rule". Please check the button label.
- UI path mismatch: This PR says Rules > Add rule > Describe agent rule, but Workspace-Rules.md says Agents tab > Add AI rule.
- Terminology: The articles mix "Agent rules", "agent rules", "Agent Rules" and "AI rule", plus "suggested Agent rule" and "Suggested Agent rule", plus "Agents" and "agents". Pick one form of each.
Lowercase agents, agent rules, and roles in running text; use select instead of click/tap; use bolded navigation paths with lowercase placeholders; fix heading casing and a broken FAQ link. Co-authored-by: Stephanie Elliott <[email protected]>
|
I pushed the changes from my earlier comment in 9f2aacc. It updates all 12 articles. These changes apply across the articles:
Changes in specific articles
I left these unchanged:
No automated checks apply to Markdown-only changes, so please spot-check the HelpDot preview that OSBotify posts for this commit. |
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
|
|
||
| --- | ||
|
|
||
| ## What are suggested Agent rules |
There was a problem hiding this comment.
Inconsistent use of captalisation for Agent/agent
There was a problem hiding this comment.
this should be fixed now!
|
|
||
| Both give you pre-written instructions, but for different kinds of Agents: | ||
|
|
||
| - An **Agent template** gives you instructions for a personal Agent that works on your own expenses. |
There was a problem hiding this comment.
https://8ca159ac.helpdot.pages.dev/articles/new-expensify/ai-agents/Learn-About-AI-in-Expensify feels like a pre-curser to this article, but it doesn't make any reference to agent templates. It was a bit jarring to go from that article to this one and see this reference for the first time. I think it might be good to reference Tamplates in the About AI article, even if it's just to say "here are some existing templates that can be applied..."
There was a problem hiding this comment.
Good call! I added this to Learn About AI in Expensify, under ## How do agents automate expense tasks?:
You don't have to create an agent from scratch. You can use agent templates, which come with pre-written instructions for common tasks. You can choose a template and customize its instructions to fit your workflow.
VictoriaExpensify
left a comment
There was a problem hiding this comment.
I've left a couple of minor comments but this is looking good!
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
|
🚀 Deployed to staging by https://github.com/VictoriaExpensify in version: 9.5.6-0 🚀
|
|
🚀 Deployed to production by https://github.com/puneetlath in version: 9.5.6-6 🚀
Bundle Size Analysis (Sentry): |
Explanation of Change
Created 2 new articles:
Also updates the remaining articles to tighten scope and reinforce cross-linking,
Fixed Issues
$ https://github.com/Expensify/Expensify/issues/686511
PROPOSAL:
Tests
Offline tests
QA Steps
// TODO: These must be filled out, or the issue title must include "[No QA]."
PR Author Checklist
### Fixed Issuessection aboveTestssectionOffline stepssectionQA stepssectionAvatar, I verified the components usingAvatarare working as expected)StyleUtils.getBackgroundAndBorderStyle(theme.componentBG))npm run compress-svg)Avataris modified, I verified thatAvataris working as expected in all cases)Designlabel and/or tagged@Expensify/designso the design team can review the changes.mainbranch was merged into this PR after a review, I tested again and verified the outcome was still expected according to theTeststeps.Screenshots/Videos
Android: Native
Android: mWeb Chrome
iOS: Native
iOS: mWeb Safari
MacOS: Chrome / Safari