Update Invent docs: Add text annotation and feedback system

[claude]

Summary

• Adds a new "Review and annotate a plan" accordion entry to the "Set up and use Invent" section, documenting inline comments, overall notes, and structured feedback submission
• Adds a corresponding FAQ entry: "Can I leave comments on a plan Invent has created?"

Addresses https://linear.app/relevance/issue/TSP-1432/

Test plan

• Accordion appears as the second item in the "Set up and use Invent" tab, after "Start a task"
• All headings are in sentence case
• FAQ entry renders correctly in the AccordionGroup
• No broken links introduced

🤖 Generated with Claude Code

[linear]

TSP-1432

[mintlify]

Preview deployment for your docs. Learn more about Mintlify Previews.

Project	Status	Preview	Updated (UTC)
relevanceai	🟢 Ready	View Preview	Sep 1, 2026, 4:35 AM

💡 Tip: Enable Workflows to automatically generate PRs for you.

[github-actions]

🎯 Vibe check

Reviewed: 1 file (1 with issues, 0 clean)

Scores

	Dimension	Score	What's holding it back
🟡	Consistency	7/10	Three distinct capitalization errors: tools/Tools mix in Session approvals, accordion title "About you" vs. UI name "About You", and "Relevance AI tools" → "Relevance AI Tools" in the MCP card.
🟡	Technical clarity	8/10	"Configure an alarm" tab presents the specific auto-diagnosis feature before explaining what alarms are — inverted for a reader encountering this for the first time. Otherwise UI references are exact and instructions are specific.
🟢	Non-technical clarity	9/10	Clear prose throughout, no jargon dumps, good plain-language intro. Minor docking because the FAQ section repeats content the reader just consumed without adding a new angle.
🟡	Structure	7/10	Four enterprise-security FAQ entries are verbatim copies of accordion content 250 lines above. "Invent features" jumps straight into Tabs with no bridging sentence. Alarm tab has inverted content order. No closing CTA for what is effectively the product hub page.

Score key: 🟢 9–10, 🟡 6–8, 🔴 1–5.

✨ Overall vibe: Solid, well-organized hub page that covers Invent's full surface area — access controls, features, Evals integration, and MCP disambiguation are all here and well written. The main weaknesses are a handful of capitalization inconsistencies and a FAQ section that leans too hard on repeating what's already above rather than adding a distinct Q&A value.

🔧 Issues (4)

• build/invent/invent.mdx:54–57 — "tools" and "Tools" are used interchangeably in the same bullet context (Session approvals tab). Lines 55 and 57 capitalize "Tools set to Auto" / "Tools set to Always ask" while lines 54 and 56 use lowercase "tools set to Ask". Because these items are Invent's internal action approvals (not the Relevance AI Tool product — the same section calls them "individual tool" in lowercase on line 64), all four should be lowercase: tools set to **Auto** still run / tools set to **Always ask** still require your approval.

• build/invent/invent.mdx:141 — Accordion title "About you" (lowercase 'y') doesn't match the product UI label. Within the accordion (lines 144, 156) and in the Steps, it's consistently "About You" — the actual name of the setting. The accordion title should match: title="About You".

• build/invent/invent.mdx:330 — "Call your existing Relevance AI tools from external clients…" — these are the Relevance AI Tool product, so: "Relevance AI Tools" (capital T).

• build/invent/invent.mdx:351–361 — Three enterprise FAQ entries ("How does project isolation work…", "Which integrations can Invent access…", "Where can Enterprise Invent data be sent?") are word-for-word copies of the Project isolation, Integration reach, and Data egress accordions from the Enterprise security controls section (lines 91–101). FAQs should add a new angle or reframe the question — verbatim duplication on the same page makes the FAQ section feel like filler and creates a maintenance burden. Either remove these three entries and add anchor links from the FAQ to the relevant accordions, or rewrite them as distinct questions that couldn't be answered by reading the section above.

🏗️ Page structure (4)

• build/invent/invent.mdx:114 — ## Invent features jumps directly into a <Tabs> component with no bridging sentence. One line like "Invent includes the following tools for building and managing your Agent team:" would help readers orient before they see two unlabelled tabs.

• build/invent/invent.mdx:304–316 — "Configure an alarm" tab presents content in the wrong order. The H4 "Turn on auto-diagnosis" and its description appear before the general explanation of what alarms are ("Alarms evaluate the scores collected by a Performance dashboard…"). A reader who hasn't used alarms before won't understand what auto-diagnosis is being turned on for. Put the general alarm description first, then the H4 for the specific setting, then the Note about where alarms are configured.

• build/invent/invent.mdx:343–378 — The FAQ section is structurally sound for most entries, but entries 347 ("Who can use Invent?") and 351–361 (three enterprise security entries) repeat page content verbatim rather than adding a Q&A framing that justifies their presence. See Issues for the three enterprise entries; "Who can use Invent?" is a one-sentence repeat of the Availability section at line 46 — either drop it or rephrase as "Can I turn Invent on for my whole team?" to capture the actionable part.

• build/invent/invent.mdx (whole file) — No closing CTA. This is the hub page for the entire Invent product, so readers who reach the end of the FAQ have just finished a comprehensive orientation. A short ## What's next? section would give them somewhere obvious to go. Natural candidates: /build/agents/build-your-agent/evals (Evals is already linked mid-page via the card at line 248, but repeating it here as a terminal CTA makes sense) and the Workforce or Agent builder as the place to start building.

⚠️ Contradictions (0)

No contradictions found between build/invent/invent.mdx and the context pages. The version-restore behavior (draft landing, (restored) suffix, confirmation required) is consistent with version-history.mdx. The Evals workflow steps are consistent with evals.mdx.

🔋 Credit usage

Item	Count
Files reviewed	1
Context pages read	2
Total lines processed	~915

Files read: build/invent/invent.mdx (379 lines), build/agents/build-your-agent/evals.mdx (447 lines), build/agents/build-your-agent/version-history.mdx (89 lines)