docs(522754a3-5b91-4602-9737-710452808b8b): add Academy CTAs to docs

[claude]

Summary

• Adds Academy card to the Learning Resources section in get-started/introduction.mdx
• Adds Academy card after the "What's next?" section in build/introduction.mdx
• Adds an Academy section under "Beyond building" in get-started/quick-start-guide.mdx
• Adds an inline Academy mention at the end of build/agents/create-an-agent.mdx
• Adds an inline Academy mention at the end of build/tools/create-a-tool.mdx

Five strategic placements — all in high-traffic entry points and "what's next?" sections where users are most likely to want structured learning resources.

Linear: https://linear.app/relevance/issue/522754a3-5b91-4602-9737-710452808b8b/

[mintlify]

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

Project	Status	Preview	Updated
relevanceai	🟢 Ready	View Preview	Sep 15, 2026, 1:04 AM

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

[github-actions]

🎯 Vibe check

Reviewed: 5 files (4 with issues, 1 clean)

Scores

	Dimension	Score	What's holding it back
🔴	Consistency	3/10	get-started/introduction.mdx has 15 heading/title capitalization failures and a banned word ("seamlessly"). get-started/quick-start-guide.mdx uses lowercase "agent", "tools", "knowledge" throughout — the same content in create-an-agent.mdx capitalizes them correctly. build/introduction.mdx and build/tools/create-a-tool.mdx have additional product term capitalization misses.
🟢	Technical clarity	8/10	UI element names are specific and exact ("Click Create Agent", not "click the button"). Step-by-step instructions are unambiguous. Internal links appear valid. The capitalization inconsistency between quick-start and create-an-agent is the only friction for a technical reader who might wonder if lowercase "tools" means something different from "Tools."
🟡	Non-technical clarity	7/10	get-started/introduction.mdx lists product features three times in three different formats (bullet list → cards → numbered "How It Works" list), which could confuse a new reader about what matters. The "seamlessly" bullet is marketing noise. The how-to guides are clear and approachable.
🟡	Structure	6/10	get-started/introduction.mdx violates two explicit CLAUDE.md rules: it has a "Quick Links" section that duplicates the sidebar, and a "How It Works" section that repeats what the "Core Features" section above already says. The how-to pages (create-an-agent, create-a-tool, quick-start-guide) have clean, logical structure.

Score key: 🟢 9–10, 🟡 6–8, 🔴 1–5.

✨ Overall vibe: The how-to guides (create-an-agent.mdx, create-a-tool.mdx, build/introduction.mdx) are well-structured and give clear, specific instructions — create-an-agent.mdx in particular is clean. The two get-started/ pages drag the PR down: introduction.mdx has a banned word, 15 heading/card/accordion capitalization failures, and two structural violations; quick-start-guide.mdx imports the same step content as create-an-agent.mdx but silently drops all the correct "Agent/Tools/Knowledge" capitalization.

🔧 Issues (25)

get-started/introduction.mdx — banned word

• get-started/introduction.mdx:19 — "seamlessly" is a banned word. Rewrite: "**Integrate with your existing tools**" or "**Drop into your existing stack**".

get-started/introduction.mdx — heading and section title case (sentence case required)

• get-started/introduction.mdx:23 — ## Quick Links → ## Quick links
• get-started/introduction.mdx:57 — ## Core Features → ## Core features
• get-started/introduction.mdx:111 — ## How It Works → ## How it works
• get-started/introduction.mdx:121 — ## Popular Use Cases → ## Popular use cases
• get-started/introduction.mdx:141 — ## Learning Resources → ## Learning resources

get-started/introduction.mdx — Card title case

• get-started/introduction.mdx:27 — "Create Your First Agent" → "Create your first Agent" ("Your" and "First" are not proper nouns)
• get-started/introduction.mdx:41 — "Watch Video Tutorials" → "Watch video tutorials"
• get-started/introduction.mdx:47 — "Get Help & Support" → "Get help & support"
• get-started/introduction.mdx:158 — "Blog & Updates" → "Blog & updates"

get-started/introduction.mdx — Accordion title case (all five need sentence case)

• get-started/introduction.mdx:124 — "Customer Support" → "Customer support"
• get-started/introduction.mdx:127 — "Sales & Lead Qualification" → "Sales & lead qualification"
• get-started/introduction.mdx:130 — "Content Creation" → "Content creation"
• get-started/introduction.mdx:133 — "Data Analysis & Reporting" → "Data analysis & reporting"
• get-started/introduction.mdx:136 — "Scheduling & Coordination" → "Scheduling & coordination"

get-started/quick-start-guide.mdx — product term capitalization (widespread)
These are all referring to Relevance AI product features and must be capitalized per CLAUDE.md. The companion page build/agents/create-an-agent.mdx covers the same content and capitalizes correctly — these are inconsistent.

• get-started/quick-start-guide.mdx:41 — "working agent" → "working Agent"
• get-started/quick-start-guide.mdx:49 — "generated agent" → "Agent"; "suggest tools" → "suggest Tools"
• get-started/quick-start-guide.mdx:50 — "your tools need" → "your Tools need"
• get-started/quick-start-guide.mdx:58,61,62,64 — "an agent that's" / "an agent that matches" / "Click on the agent" / "integrations the agent needs" → "Agent" each time
• get-started/quick-start-guide.mdx:65 — "the prompt, tools, and settings" → "Tools"
• get-started/quick-start-guide.mdx:78 — "tell the agent" → "Agent"
• get-started/quick-start-guide.mdx:79 — link text [tools] → [Tools]
• get-started/quick-start-guide.mdx:80 — link text [knowledge] → [Knowledge]
• get-started/quick-start-guide.mdx:88,91,98,99,102 — "your agent" (×5 occurrences) → "your Agent"
• get-started/quick-start-guide.mdx:122 — "your agents directly" → "Agents"; "@ mention any agent" → "any Agent"; "tools and knowledge you've configured" → "Tools and Knowledge"

build/introduction.mdx — product term capitalization

• build/introduction.mdx:9 — "create and configure everything — agents, tools, workforces, and knowledge" → "Agents, Tools, Workforces, and Knowledge" (all four are Relevance AI product names)
• build/introduction.mdx:61 — "building an agent's prompt" → "Agent's prompt"; "create custom tools" → "custom Tools"; "connect knowledge sources" → "Knowledge sources"; "wire it all together in a workforce" → "Workforce"

build/tools/create-a-tool.mdx — heading and product term capitalization

• build/tools/create-a-tool.mdx:73 — ## Editing shared tools → ## Editing shared Tools
• build/tools/create-a-tool.mdx:75 — "a tool that is used by two or more agents" → "a Tool used by two or more Agents"
• build/tools/create-a-tool.mdx:77,78,79,81 — "tool" → "Tool", "agents" → "Agents" throughout the section (4 occurrences each)

🧩 Component suggestions (2)

• get-started/introduction.mdx:115–119 — The "How It Works" section uses a plain numbered list for a sequential, 5-step process ("Build agents → Give them tools → Add knowledge → Connect into workforces → Set guardrails and deploy"). This is exactly the pattern <Steps> is designed for. Replace the numbered markdown list with <Steps> to add visual progress indicators. The surrounding content in build/introduction.mdx already uses <Steps> for the same kind of component breakdown, which makes the plain list here feel inconsistent.

• get-started/introduction.mdx:107–109 — The supademo iframe uses "padding-top":"56.75%" as an inline style string in JSX. All other supademo embeds in this repo (get-started/marketplace/introduction.mdx, get-started/chat/introduction.mdx) use paddingTop: '56.25%' (camelCase, consistent percentage). This one has both a different key format ("padding-top" vs paddingTop) and a slightly different value (56.75% vs 56.25%). Verify this renders correctly and unify to paddingTop: '56.25%' for consistency.

🏗️ Page structure (3)

• get-started/introduction.mdx:23–54 — The ## Quick Links section is explicitly prohibited by CLAUDE.md: "No redundant 'Quick Links' sections that duplicate sidebar navigation." The four cards here (Create First Agent, Try Chat, Watch Tutorials, Get Help) all link to pages already reachable from the sidebar. Remove this section; the ## Core Features cards below it serve the same purpose more completely.

• get-started/introduction.mdx:111–119 — The ## How It Works section is explicitly prohibited by CLAUDE.md: "No 'How It Works' sections that repeat information already explained elsewhere on the page." The five numbered steps (Build agents → Tools → Knowledge → Workforces → Deploy) cover the same ground as the ## Core Features cards directly above them. Remove the section; the cards already give readers the conceptual map, and the how-to pages give the step-by-step detail.

• get-started/introduction.mdx — After removing the two prohibited sections above, the page flow will improve significantly: definition paragraph → "Why Relevance AI?" → "Build your first agent" demo → "Core Features" cards → "Popular Use Cases" accordions → "Learning Resources" → closing CTA. That's a clean concept-page structure matching CLAUDE.md's recommended template (definition → benefits → use cases → CTA).

✅ Clean files (1)

build/agents/create-an-agent.mdx — No issues found. Headings are sentence case, product terms are correctly capitalized throughout (including "Tools" in Invent steps), Card titles are correct, Tip callouts are single paragraphs, and the What's next section links to real pages.

🔋 Credit usage

Item	Count
Files reviewed	5
Context pages read	4
Total lines processed	~1,180

Files read: build/agents/create-an-agent.mdx (94 lines), build/introduction.mdx (66 lines), build/tools/create-a-tool.mdx (105 lines), get-started/introduction.mdx (170 lines), get-started/quick-start-guide.mdx (141 lines), get-started/core-concepts/agents.mdx (~230 lines), get-started/core-concepts/tools.mdx (101 lines), get-started/chat/introduction.mdx (199 lines), get-started/marketplace/introduction.mdx (74 lines)

[claude]

Addressed the three feedback points:

• create-an-agent.mdx and create-a-tool.mdx: Wrapped the plain-text Academy links in <Tip> callouts so they stand out visually at the end of the "What's next?" sections instead of blending into the surrounding text.
• introduction.mdx: Moved the floating Academy card out of the bottom of the page and into the "Choose how to build" <CardGroup> as a third option (alongside "Build with Invent" and "Build directly"). It now sits in a natural, contextual position rather than trailing the page.

[jordanc-relevanceai]

Can we also add the tip to workforces and knowledge as well then? also, on things like Evals

[claude]

Addressed reviewer feedback with two changes:

Removed Academy card from build/introduction.mdx
The third card in the "Choose how to build" CardGroup (the Academy one) has been removed per the "dont like this being here" comment. The group is now a 2-column layout with just Invent and Build directly.

Added Academy <Tip> callouts to three new pages
Following the same pattern already approved on create-an-agent.mdx and create-a-tool.mdx, tips now appear at the bottom of:

• build/workforces/create-a-workforce.mdx — after the "What's next?" link list
• build/knowledge/create-knowledge.mdx — after the "Related Features" section
• build/agents/evals/introduction.mdx — after the FAQ accordion group

[github-actions]

🎯 Vibe check

Reviewed: 5 files (5 with issues, 0 clean)

Scores

	Dimension	Score	What's holding it back
🔴	Consistency	4/10	get-started/introduction.mdx has a banned word, 9 heading/card/accordion case violations, and wrong product term capitalization throughout the How It Works list. "Inventor" used instead of "Invent" across two files (4 instances). Lowercase agents/tools/knowledge where the product features are meant in build/introduction.mdx and build/tools/create-a-tool.mdx.
🟢	Technical clarity	8/10	UI references are specific and correct throughout. The one gap: "RAG solution" (get-started/introduction.mdx:85) is used without expansion — non-obvious to anyone who hasn't heard the acronym.
🟡	Non-technical clarity	7/10	Most pages are approachable. get-started/introduction.mdx drops "RAG" cold. The How It Works numbered list is informative but mixes jargon-y bullet formatting with plain statements.
🟡	Structure	6/10	get-started/introduction.mdx contains both a "Quick Links" section and a "How It Works" section — both explicitly prohibited by CLAUDE.md. build/introduction.mdx uses <Steps> for four parallel components that aren't actually sequential steps.

Score key: 🟢 9–10, 🟡 6–8, 🔴 1–5.

✨ Overall vibe: The build-side pages (build/introduction.mdx, build/agents/create-an-agent.mdx, build/tools/create-a-tool.mdx) are mostly solid — clear structure, correct components, good specificity. get-started/introduction.mdx is the problem child: it packs in a banned word, double-digit case violations, two prohibited section types, and unexplained jargon. Fix that file and the PR is in good shape.

🔧 Issues (17)

• get-started/introduction.mdx:19 — seamlessly is a banned word: "Integrate seamlessly with your existing tools and workflows" → rephrase, e.g. "Integrate with your existing tools and workflows"

• get-started/introduction.mdx:17 — product term: "Deploy agents in minutes" → "Deploy Agents in minutes"

• get-started/introduction.mdx:20 — product term: "from agent behavior to escalation protocols" → "from Agent behavior to escalation protocols"

• get-started/introduction.mdx:27 — Card title case: "Create Your First Agent" → "Create your first agent"

• get-started/introduction.mdx:37 — Card title case: "Watch Video Tutorials" → "Watch video tutorials"

• get-started/introduction.mdx:42 — Card title case: "Get Help & Support" → "Get help & support"

• get-started/introduction.mdx:56 — Heading case: ## Core Features → ## Core features

• get-started/introduction.mdx:85 — Unexplained jargon: "RAG solution" is used without defining the acronym. Expand to "Retrieval-Augmented Generation (RAG) solution" or rephrase: "gives agents access to your specific data beyond what they were trained on"

• get-started/introduction.mdx:111 — Heading case: ## How It Works → ## How it works

• get-started/introduction.mdx:115–119 — Product term capitalization in How It Works numbered list: "build agents", "give them tools", "add knowledge", "connect them into workforces" → "build Agents", "give them Tools", "add Knowledge", "connect them into Workforces"

• get-started/introduction.mdx:121 — Heading case: ## Popular Use Cases → ## Popular use cases

• get-started/introduction.mdx:124,127,130,133,136 — Five Accordion titles need sentence case: "Customer Support" → "Customer support", "Sales & Lead Qualification" → "Sales & lead qualification", "Content Creation" → "Content creation", "Data Analysis & Reporting" → "Data analysis & reporting", "Scheduling & Coordination" → "Scheduling & coordination"

• get-started/introduction.mdx:141 — Heading case: ## Learning Resources → ## Learning resources

• get-started/introduction.mdx:157 — Card title case: "Blog & Updates" → "Blog & updates"

• build/tools/create-a-tool.mdx:75–80 — Product term inconsistency in "Editing shared tools" section: "tool" and "agents" used lowercase throughout, but line 39 of the same file correctly capitalizes "the Tool to an Agent". Fix: "a tool that is used by two or more agents" → "a Tool that is used by two or more Agents"; "how many agents use the tool" → "how many Agents use the Tool"; "Edit the tool directly when the change should apply across all agents" → "Tool", "Agents"; "Clone the tool when you need a variation for a specific agent" → "Tool", "Agent"

• build/introduction.mdx:33–43 — Product term capitalization in Steps body text: [agents], [tools], [workforce], [knowledge] as link text should match the product names: [Agents], [Tools], [Workforce], [Knowledge]. Same applies to occurrences inline (e.g. "what your agents can do" → "what your Agents can do")

• build/introduction.mdx:64 — Product term capitalization in What's next paragraph: "an agent's prompt" → "an Agent's prompt"; "custom tools" → "custom Tools"; "knowledge sources" → "Knowledge sources"; "a workforce" → "a Workforce"

• build/agents/create-an-agent.mdx:45,49 and get-started/quick-start-guide.mdx:49,53 — "Inventor" should be "Invent" in all four instances. The feature is named "Invent" (see step 3 on the same page: "Select Invent", and consistent with build/introduction.mdx throughout). "Inventor will set up the prompt" → "Invent will set up the prompt"; "Inventor handles the setup" → "Invent handles the setup"

🧩 Component suggestions (2)

• get-started/introduction.mdx:113–119 — The "How it works" section uses a plain numbered list for a sequential 5-step process (Build agents → Give them tools → Add knowledge → Connect into workforces → Set guardrails and deploy). This is exactly the use case for <Steps> — swap the numbered list for a <Steps> block with each step as a <Step title="..."> so readers get progress indicators and the steps are visually distinct.

• build/introduction.mdx:31–44 — <Steps> is used for the four platform components (Agents, Tools, Workforces, Knowledge), but these are parallel, co-equal parts of the platform — not a sequential procedure. <Steps> implies "do these in order," which isn't the intent. Replace with a <CardGroup cols={2}> so readers see them as parallel options, consistent with the "Choose how to build" CardGroup lower on the same page.

🏗️ Page structure (2)

• get-started/introduction.mdx:23 — The ## Quick Links section (lines 23–54) is explicitly prohibited by CLAUDE.md: "No redundant 'Quick Links' sections that duplicate sidebar navigation." All four links (quick start, chat, YouTube, support) are reachable from the sidebar. Remove this section or absorb the most important entry into the intro paragraph as inline prose.

• get-started/introduction.mdx:111 — The ## How It Works section (lines 111–119) largely repeats what the intro paragraph and "Core Features" CardGroup have already covered. CLAUDE.md: "No 'How It Works' sections that repeat information already explained elsewhere on the page." If the sequential framing (Build → Give → Add → Connect → Deploy) adds unique value over the feature list above it, convert it to <Steps> and cut anything already covered. Otherwise remove it.

🔋 Credit usage

Item	Count
Files reviewed	5
Context pages read	2
Total lines processed	~570

Files read: build/agents/create-an-agent.mdx (96 lines), build/introduction.mdx (65 lines), build/tools/create-a-tool.mdx (107 lines), get-started/introduction.mdx (170 lines), get-started/quick-start-guide.mdx (141 lines), get-started/core-concepts/agents.mdx (230 lines, context), get-started/core-concepts/tools.mdx (101 lines, context)

[github-actions]

🎯 Vibe check

Reviewed: 7 files (4 with issues, 3 clean)

Scores

	Dimension	Score	What's holding it back
🔴	Consistency	5/10	get-started/introduction.mdx has a banned word ("seamlessly") and 14+ heading/Card/Accordion title-case violations. build/knowledge/create-knowledge.mdx has 15 heading title-case violations and consistent lowercase "agents/agent/tool" where the product terms Agent/Tool should be capitalized. build/tools/create-a-tool.mdx drops capitalization on Agent/Tool throughout one section.
🟡	Technical clarity	8/10	build/knowledge/create-knowledge.mdx uses quoted strings ("Create Knowledge", "Empty Table") instead of bold for UI element names — inconsistent with every other changed file. Related Features section uses absolute URLs instead of root-relative paths.
🟡	Non-technical clarity	8/10	The how-to pages (agents, tools, workforce, quick start) are readable and well-paced. build/knowledge/create-knowledge.mdx is dense — it lists procedural steps for organizing, enriching, and fine-tuning knowledge without explaining the tradeoffs or why a reader would do each thing.
🟡	Structure	7/10	get-started/introduction.mdx has a "Quick Links" section that CLAUDE.md prohibits. build/knowledge/create-knowledge.mdx has an H1 in the body alongside the frontmatter title. build/agents/evals/introduction.mdx has a "Next:" call-to-action buried between the Overview and FAQ sections.

Score key: 🟢 9–10, 🟡 6–8, 🔴 1–5.

✨ Overall vibe: The create-an-agent, create-a-workforce, create-a-tool, and quick-start-guide pages are clean — good UI specificity, correct product-term capitalization, and tight structure. The two pages dragging down the PR are introduction.mdx (title-case headings throughout, a banned word, and a prohibited Quick Links section) and create-knowledge.mdx (looks like it wasn't touched with the same care — 15 heading capitalization issues and lowercase product terms throughout). Fix those two and this is in good shape.

🔧 Issues (28)

Banned words

• get-started/introduction.mdx:19 — "Integrate seamlessly" → "seamlessly" is a banned word; rewrite, e.g. "Integrate with your existing tools and workflows"

Sentence case — headings

• get-started/introduction.mdx:23 — ## Quick Links → ## Quick links
• get-started/introduction.mdx:57 — ## Core Features → ## Core features
• get-started/introduction.mdx:103 — ## Build your first agent → ## Build your first Agent (Agent is a product name)
• get-started/introduction.mdx:112 — ## How It Works → ## How it works
• get-started/introduction.mdx:121 — ## Popular Use Cases → ## Popular use cases
• get-started/introduction.mdx:141 — ## Learning Resources → ## Learning resources
• build/knowledge/create-knowledge.mdx:11 — ## Creating Knowledge Sources → ## Creating knowledge sources
• build/knowledge/create-knowledge.mdx:15 — ### Manual Input → ### Manual input
• build/knowledge/create-knowledge.mdx:29 — ### Upload Documents → ### Upload documents
• build/knowledge/create-knowledge.mdx:42 — ### Extract Website Content → ### Extract website content
• build/knowledge/create-knowledge.mdx:57 — ### Connect Third-Party Integrations → ### Connect third-party integrations
• build/knowledge/create-knowledge.mdx:68 — ## Organizing Your Knowledge → ## Organizing your Knowledge
• build/knowledge/create-knowledge.mdx:73 — ### Structured Tables → ### Structured tables
• build/knowledge/create-knowledge.mdx:82 — ### Document Chunking → ### Document chunking
• build/knowledge/create-knowledge.mdx:99 — ## Enhancing Knowledge Quality → ## Enhancing knowledge quality
• build/knowledge/create-knowledge.mdx:104 — ### Metadata Enrichment → ### Metadata enrichment
• build/knowledge/create-knowledge.mdx:122 — ### Semantic Organization → ### Semantic organization
• build/knowledge/create-knowledge.mdx:151 — ### Retrieval Settings → ### Retrieval settings
• build/knowledge/create-knowledge.mdx:168 — ### Query Transformation → ### Query transformation
• build/knowledge/create-knowledge.mdx:177 — ## Best Practices for Knowledge Management → ## Best practices for knowledge management
• build/knowledge/create-knowledge.mdx:188 — ## Related Features → ## Related features
• build/tools/create-a-tool.mdx:73 — ## Editing shared tools → ## Editing shared Tools

Sentence case — Card and Accordion titles

• get-started/introduction.mdx:28 — Card "Create Your First Agent" → "Create your first Agent"
• get-started/introduction.mdx:38 — Card "Watch Video Tutorials" → "Watch video tutorials"
• get-started/introduction.mdx:43 — Card "Get Help & Support" → "Get help & support"
• get-started/introduction.mdx:124,128,132,136,140 — five Accordion titles all in title case: "Customer Support", "Sales & Lead Qualification", "Content Creation", "Data Analysis & Reporting", "Scheduling & Coordination" → sentence case each
• get-started/introduction.mdx:156 — Card "Blog & Updates" → "Blog & updates"

Product term capitalization

• build/knowledge/create-knowledge.mdx:9,41,70,131,132,135,138,141,143 — "agent/agents" used throughout the body where the Relevance AI product is meant → "Agent/Agents"; "tool" on line 143 → "Tool"
• build/tools/create-a-tool.mdx:74–81 — "tool/tools" and "agents/agent" throughout the "Editing shared Tools" section should be "Tool/Tools" and "Agents/Agent"

🧩 Component suggestions (2)

• get-started/introduction.mdx:113–119 — "How It Works" uses a plain numbered Markdown list for a sequential procedure. <Steps> adds visual progress indicators that help readers track where they are. Each existing numbered item maps cleanly to a Step with the format used in build/introduction.mdx.

• build/knowledge/create-knowledge.mdx:191–193 — Related Features links use absolute URLs (https://relevanceai.com/docs/build/knowledge/...). CLAUDE.md requires root-relative paths. Rewrite as /build/knowledge/access-knowledge, /build/knowledge/enrich-with-tool, and /build/knowledge/advanced-knowledge-retrieval.

🏗️ Page structure (3)

• get-started/introduction.mdx:23–54 — "Quick Links" section violates CLAUDE.md: "No redundant 'Quick Links' sections that duplicate sidebar navigation." The four cards (Create Your First Agent, Try Relevance Chat, Watch Video Tutorials, Get Help & Support) are already reachable from the sidebar. Remove this section or fold the most important link into the intro paragraph.

• build/knowledge/create-knowledge.mdx:7 — An H1 heading (# Create and Organize Knowledge) appears in the page body alongside the frontmatter title ("Knowledge"). In Mintlify, the frontmatter title renders as the H1, so this creates a visual double-heading. Remove the body H1 and promote its content (currently "This guide walks you through...") to the description or the first paragraph.

• build/agents/evals/introduction.mdx:92 — "Next: Learn how to [create Tests]..." sits between the Overview section and the FAQ, breaking the reading flow. Move it to the end of the page as a proper ## What's next? section after the FAQ accordions — or remove it and let the sidebar navigation carry that job.

✅ Clean files (3)

build/agents/create-an-agent.mdx, build/workforces/create-a-workforce.mdx, get-started/quick-start-guide.mdx

🔋 Credit usage

Item	Count
Files reviewed	7
Context pages read	4
Total lines processed	~1,370

Files read: build/agents/create-an-agent.mdx (96 lines), build/agents/evals/introduction.mdx (109 lines), build/knowledge/create-knowledge.mdx (197 lines), build/tools/create-a-tool.mdx (107 lines), build/workforces/create-a-workforce.mdx (50 lines), get-started/introduction.mdx (170 lines), get-started/quick-start-guide.mdx (141 lines), get-started/core-concepts/agents.mdx (230 lines), get-started/core-concepts/knowledge.mdx (110 lines), build/introduction.mdx (62 lines), get-started/core-concepts/tools.mdx (101 lines)

[jordanc-relevanceai]

Academy currently teaches users about Agents, Tools, Knowledge, Triggers. So please make sure it's called out as a tip on each page if not already. Make sure the academy is also mentioned on more general build pages as well. This is a final passthrough, so if you've already done this, then that's no problem

[claude]

Final passthrough complete. Two changes in this push:

• Learning Resources section (get-started/introduction.mdx): Changed cols={2} to cols={3} so the three cards (Academy, Community, YouTube) display as a single row.
• Triggers Academy tip (build/agents/build-your-agent/triggers.mdx): Added the Academy <Tip> after the FAQ accordion group. Triggers was the only page from Academy's curriculum (Agents, Tools, Knowledge, Triggers) that was missing a tip — this completes coverage.

[github-actions]

🎯 Vibe check

Reviewed: 8 files (6 with issues, 2 clean)

Scores

	Dimension	Score	What's holding it back
🔴	Consistency	5/10	Banned word ("seamlessly") in introduction.mdx. Pervasive title-case heading violations in create-knowledge.mdx (17+ headings) and introduction.mdx (6 headings, 5 accordion titles, 3 card titles). "Agent" and "Tools" inconsistently lowercased in quick-start-guide.mdx and triggers.mdx.
🟡	Technical clarity	7/10	create-knowledge.mdx step 134 says "Select 'Add existing knowledge'" — vague navigation with wrong formatting. Internal links in the Related Features section use absolute URLs instead of root-relative paths. UI element option names formatted in italics instead of bold.
🟡	Non-technical clarity	7/10	create-knowledge.mdx has extended sections ("Enhancing Knowledge Quality", "Advanced Knowledge Configuration") with generic numbered advice and no specific UI control references — reads like filler rather than product documentation.
🟡	Structure	6/10	introduction.mdx has a "Quick Links" section that CLAUDE.md explicitly bans. create-knowledge.mdx has a content H1 that duplicates the frontmatter title. The "Next:" navigation line in evals/introduction.mdx appears before the FAQ section rather than at the end. Three standalone <Card> components in create-knowledge.mdx need a <CardGroup> wrapper.

Score key: 🟢 9–10, 🟡 6–8, 🔴 1–5.

✨ Overall vibe: create-an-agent.mdx, create-a-tool.mdx, and create-a-workforce.mdx are tight, focused, and well-structured — the trio of "how to create X" pages sets a good template. The drag is create-knowledge.mdx (pervasive heading case violations, vague generic content) and introduction.mdx (banned word, multiple title-case issues, and a Quick Links section that CLAUDE.md explicitly prohibits). quick-start-guide.mdx is solid in structure but needs a pass to capitalize "Agent" and "Tools" consistently in body text.

🔧 Issues (20)

Banned word

• get-started/introduction.mdx:19 — **Integrate seamlessly** — "seamlessly" is banned. Rewrite: "Integrate with your existing tools and workflows"

Heading sentence case — get-started/introduction.mdx

• get-started/introduction.mdx:23 — ## Quick Links → ## Quick links
• get-started/introduction.mdx:57 — ## Core Features → ## Core features
• get-started/introduction.mdx:103 — ## Build your first agent — "agent" here refers to the product feature and should be "Agent": ## Build your first Agent
• get-started/introduction.mdx:111 — ## How It Works → ## How it works
• get-started/introduction.mdx:121 — ## Popular Use Cases → ## Popular use cases
• get-started/introduction.mdx:141 — ## Learning Resources → ## Learning resources

Card titles — get-started/introduction.mdx

• get-started/introduction.mdx:27 — title="Create Your First Agent" → title="Create your first Agent"
• get-started/introduction.mdx:39 — title="Watch Video Tutorials" → title="Watch video tutorials"
• get-started/introduction.mdx:44 — title="Get Help & Support" → title="Get help and support"

Accordion titles — get-started/introduction.mdx

• get-started/introduction.mdx:124,127,130,133,136 — All five Popular Use Cases accordion titles use title case. Correct forms: "Customer support", "Sales and lead qualification", "Content creation", "Data analysis and reporting", "Scheduling and coordination"

Heading sentence case — build/knowledge/create-knowledge.mdx (pervasive)

• build/knowledge/create-knowledge.mdx:12–189 — 17+ H2/H3 headings use title case. Needs a full pass. Worst offenders: ## Creating Knowledge Sources, ### Manual Input, ### Upload Documents, ### Extract Website Content, ### Connect Third-Party Integrations, ## Organizing Your Knowledge, ### Structured Tables, ### Document Chunking, ### Knowledge Collections, ## Enhancing Knowledge Quality, ### Metadata Enrichment, ### Knowledge Verification, ### Semantic Organization, ## Advanced Knowledge Configuration, ### Retrieval Settings, ### Knowledge Prioritization, ### Query Transformation, ## Best Practices for Knowledge Management, ## Related Features. Retain "Knowledge" and "Agents" as product names; lowercase everything else after the first word.

UI element formatting — build/knowledge/create-knowledge.mdx

• build/knowledge/create-knowledge.mdx:131 — "your agent's configuration page" — "agent's" → "Agent's"
• build/knowledge/create-knowledge.mdx:132 — "Add existing knowledge" (quoted) → Add existing knowledge (bold, as per CLAUDE.md step formatting)
• build/knowledge/create-knowledge.mdx:139–141 — The two Knowledge connection options ("Add all to prompt" and "Allow agent to search") are in italic/underscore. Use bold for UI element names.

Internal link format — build/knowledge/create-knowledge.mdx

• build/knowledge/create-knowledge.mdx:191–193 — Related Features links use absolute URLs (https://relevanceai.com/docs/build/knowledge/...). Use root-relative paths: /build/knowledge/access-knowledge, /build/knowledge/enrich-with-tool, /build/knowledge/advanced-knowledge-retrieval

Product term capitalization — get-started/quick-start-guide.mdx

• get-started/quick-start-guide.mdx:41,48,49,58,63,65,78 — "agent" used seven times in body text when referring to the Relevance AI product feature. Should be "Agent" throughout.
• get-started/quick-start-guide.mdx:49,50,65,79 — "tools" used four times as the product feature. Should be "Tools". (Note: line 50 "in the Tools section" is already correct — just the surrounding references aren't.)

Product term capitalization — build/agents/build-your-agent/triggers.mdx

• build/agents/build-your-agent/triggers.mdx:12–14,18,21 — "agent" lowercase in the three bullet list items ("Set your agent to run…", "Configure your agent to respond…", "Allow users to directly engage with the agent") and in the setup steps ("select the agent you want to configure", "so your agent can respond"). Should be "Agent".

Product term capitalization — build/tools/create-a-tool.mdx

• build/tools/create-a-tool.mdx:76 — "how many agents use the tool" → "how many Agents use the Tool"

🧩 Component suggestions (3)

• build/knowledge/create-knowledge.mdx:59–65 — Three standalone <Card> components for Google Drive, SharePoint, and Notion sit without a <CardGroup> wrapper. Wrap them: <CardGroup cols={3}> (or cols={2} if preferred). Bare cards stack oddly without the group.

• build/knowledge/create-knowledge.mdx:130–143 — The "Connecting Knowledge to Agents" numbered list has a step 4 with two sub-options using indented nested list formatting. The option names ("Add all to prompt", "Allow agent to search") would read more clearly as a small comparison table or as two <Card> components inside the numbered step. The current indented list with italic labels is hard to scan.

• build/agents/evals/introduction.mdx:92 — The Next: navigation line (Next: Learn how to [create Tests]...) is placed between the "Cost and billing" section and the FAQ — move it to after the FAQ as the page's closing element. Readers who skip to the FAQ will miss it where it is now.

🏗️ Page structure (4)

• get-started/introduction.mdx:23–54 — The "Quick Links" section is explicitly banned by CLAUDE.md: "No redundant 'Quick Links' sections that duplicate sidebar navigation." All four linked pages (Quick Start, Relevance Chat, YouTube, Support) are reachable from the sidebar. Remove this section; the "Core Features" CardGroup and the "Build your first agent" demo already give readers somewhere to go.

• build/knowledge/create-knowledge.mdx:7 — The page has both a frontmatter title: "Knowledge" and a content H1 (# Create and Organize Knowledge). In Mintlify, the frontmatter title renders as the page H1 — the in-body H1 creates a duplicate. Remove line 7 or promote the real title to frontmatter (title: "Create and organize Knowledge").

• build/knowledge/create-knowledge.mdx:99–175 — "Enhancing Knowledge Quality" and "Advanced Knowledge Configuration" are structured as step lists but reference no actual UI controls ("Configure the number of results returned", "Set relevance thresholds"). These read as generic RAG advice rather than product documentation. Either tie each step to a named UI element or field, or cut these sections and link to the existing dedicated pages (/build/knowledge/advanced-knowledge-retrieval, etc.) that presumably cover the real UI.

• build/knowledge/create-knowledge.mdx — No "What's next?" or closing CTA. As a how-to page in a clear sequence (create → access → enrich), it would benefit from a closing ## What's next? pointing to /build/knowledge/access-knowledge and /build/knowledge/enrich-with-tool.

✅ Clean files (2)

build/agents/create-an-agent.mdx, build/workforces/create-a-workforce.mdx

🔋 Credit usage

Item	Count
Files reviewed	8
Context pages read	4
Total lines processed	~1,404

Files read: build/agents/build-your-agent/triggers.mdx (126 lines), build/agents/create-an-agent.mdx (96 lines), build/agents/evals/introduction.mdx (109 lines), build/knowledge/create-knowledge.mdx (197 lines), build/tools/create-a-tool.mdx (107 lines), build/workforces/create-a-workforce.mdx (50 lines), get-started/introduction.mdx (170 lines), get-started/quick-start-guide.mdx (141 lines), get-started/core-concepts/agents.mdx (230 lines), build/agents/build-your-agent/agent-triggers/integrations.mdx (38 lines), build/knowledge/access-knowledge.mdx (30 lines), get-started/core-concepts/knowledge.mdx (110 lines)

[jordanc-relevanceai]

instead of tips, can we make them info blocks?

[claude]

Done — converted all Academy CTA callouts from <Tip> to <Info> across the 6 files (create-an-agent, create-a-tool, create-a-workforce, create-knowledge, evals/introduction, triggers). Content is unchanged, only the component type changed.

[github-actions]

🎯 Vibe check

Reviewed: 8 files (5 with issues, 3 clean)

Scores

	Dimension	Score	What's holding it back
🔴	Consistency	4/10	create-knowledge.mdx has 20+ title-case heading violations across every H2/H3 on the page; introduction.mdx has 5 accordion title case issues, 3 card title case issues, and the banned word "seamlessly"; quick-start-guide.mdx lowercases "agent" and "tools" where create-an-agent.mdx capitalizes the same terms in identical steps.
🟡	Technical clarity	6/10	create-knowledge.mdx steps quote UI button names in double-quotes instead of bold; the four abstract sections (Organizing, Enhancing, Advanced Configuration, Best Practices) contain generic numbered lists with zero product-specific UI references — none tell the reader where to click. Internal links in that file also use absolute URLs that will break if the docs domain ever changes.
🟡	Non-technical clarity	7/10	Most pages define the feature before giving instructions. create-knowledge.mdx is the outlier: large sections describe abstract knowledge-management theory ("Semantic Organization", "Query Transformation") without grounding them in what Relevance AI actually shows the user.
🟡	Structure	6/10	introduction.mdx has two explicit CLAUDE.md violations: a "Quick Links" section and a "How It Works" section. The evals intro has a stray "Next:" line sandwiched mid-page between the cost section and the FAQ.

Score key: 🟢 9–10, 🟡 6–8, 🔴 1–5. Scores are a single overall judgment about the whole PR — not per file.

✨ Overall vibe: Three files (create-an-agent.mdx, create-a-tool.mdx, create-a-workforce.mdx) are clean, well-structured, and ready to merge. The PR is dragged down by create-knowledge.mdx — it reads like an AI-generated first draft with every heading in title case, four abstract sections that reference no UI, and absolute internal links — and by introduction.mdx, which needs its Quick Links and How It Works sections removed per CLAUDE.md. Fix those two files and the rest is solid.

🔧 Issues (30)

Banned words

• get-started/introduction.mdx:19 — "Integrate seamlessly" → "seamlessly" is a banned word. Rewrite: "Integrate directly with your existing tools and workflows"

Heading sentence case — get-started/introduction.mdx

• get-started/introduction.mdx:28 — Card title "Create Your First Agent" → "Create your first Agent"
• get-started/introduction.mdx:37 — Card title "Watch Video Tutorials" → "Watch video tutorials"
• get-started/introduction.mdx:41 — Card title "Get Help & Support" → "Get help & support"
• get-started/introduction.mdx:103 — Heading "## Build your first agent" → "## Build your first Agent" (Agent is a product name)
• get-started/introduction.mdx:111 — Heading "## How It Works" → "## How it works"
• get-started/introduction.mdx:124 — Accordion title "Customer Support" → "Customer support"
• get-started/introduction.mdx:128 — Accordion title "Sales & Lead Qualification" → "Sales & lead qualification"
• get-started/introduction.mdx:131 — Accordion title "Content Creation" → "Content creation"
• get-started/introduction.mdx:134 — Accordion title "Data Analysis & Reporting" → "Data analysis & reporting"
• get-started/introduction.mdx:137 — Accordion title "Scheduling & Coordination" → "Scheduling & coordination"

Heading sentence case — build/knowledge/create-knowledge.mdx (pervasive)
Every H2 and H3 in this file uses title case. Full list:

• :10 "## Creating Knowledge Sources" → "## Creating knowledge sources"
• :15 "### Manual Input" → "### Manual input"
• :29 "### Upload Documents" → "### Upload documents"
• :43 "### Extract Website Content" → "### Extract website content"
• :57 "### Connect Third-Party Integrations" → "### Connect third-party integrations"
• :68 "## Organizing Your Knowledge" → "## Organizing your knowledge"
• :73 "### Structured Tables" → "### Structured tables"
• :82 "### Document Chunking" → "### Document chunking"
• :91 "### Knowledge Collections" → "### Knowledge collections"
• :99 "## Enhancing Knowledge Quality" → "## Enhancing knowledge quality"
• :103 "### Metadata Enrichment" → "### Metadata enrichment"
• :112 "### Knowledge Verification" → "### Knowledge verification"
• :121 "### Semantic Organization" → "### Semantic organization"
• :148 "## Advanced Knowledge Configuration" → "## Advanced knowledge configuration"
• :152 "### Retrieval Settings" → "### Retrieval settings"
• :160 "### Knowledge Prioritization" → "### Knowledge prioritization"
• :169 "### Query Transformation" → "### Query transformation"
• :178 "## Best Practices for Knowledge Management" → "## Best practices for knowledge management"
• :189 "## Related Features" → "## Related features"

Internal links — build/knowledge/create-knowledge.mdx

• build/knowledge/create-knowledge.mdx:191-193 — All three "Related Features" links use absolute URLs (https://relevanceai.com/docs/build/knowledge/...). CLAUDE.md requires root-relative paths. Change to /build/knowledge/access-knowledge, /build/knowledge/enrich-with-tool, /build/knowledge/advanced-knowledge-retrieval.

UI element formatting — build/knowledge/create-knowledge.mdx

• build/knowledge/create-knowledge.mdx:20,22,32,34,50 — Button/option names are in double quotes: "Create Knowledge", "Empty Table", "Upload Files", "Website". CLAUDE.md standard is bold: **Create Knowledge**, **Empty Table**, etc.

Frontmatter conflict — build/knowledge/create-knowledge.mdx

• build/knowledge/create-knowledge.mdx:7 — Page has both a frontmatter title: "Knowledge" and an inline H1 # Create and Organize Knowledge. In Mintlify the frontmatter title renders as the page H1, so the inline one creates a duplicate. Remove the inline H1.

Cross-page capitalization inconsistency

• get-started/quick-start-guide.mdx:49,50,62,65 — These lines use "agent" and "tools" (lowercase) in steps that are word-for-word identical to steps in build/agents/create-an-agent.mdx, which capitalizes them as "Agent" and "Tools". Align to match create-an-agent.mdx (capitalize, since they refer to the product features).

🧩 Component suggestions (1)

• build/agents/build-your-agent/triggers.mdx:85-90 — <Columns cols={2}> is not a standard Mintlify component and its rendering is not verified in this repo. The four <Card> elements inside should use <CardGroup cols={2}> instead, which is the established Mintlify pattern used throughout the rest of the PR.

🏗️ Page structure (4)

• get-started/introduction.mdx:22-54 — The "## Quick Links" section explicitly violates CLAUDE.md: "No redundant 'Quick Links' sections that duplicate sidebar navigation." All four cards link to pages that are already in the nav. Remove the section; the Core Features cards below serve the same purpose better.

• get-started/introduction.mdx:111-119 — The "## How It Works" section violates CLAUDE.md: "No 'How It Works' sections that repeat information already explained elsewhere on the page." Steps 1–5 (build agents, give tools, add knowledge, connect workforces, deploy) duplicate both the intro paragraph and the Core Features cards. Remove it.

• build/agents/evals/introduction.mdx:92 — "Next: Learn how to create Tests" is a floating inline sentence between the cost tables and the FAQ — it breaks reading flow. Move it to a ## What's next? section after the FAQ accordions, or integrate it as a proper closing CTA with an <Info> or <CardGroup>.

• build/knowledge/create-knowledge.mdx:68-188 — The "Organizing your knowledge", "Enhancing knowledge quality", "Advanced knowledge configuration", and "Best practices" sections are generic knowledge management advice with no Relevance AI UI references. None of these sections tell the reader where to click, what settings to change, or what the product shows. Either replace with product-specific guidance (e.g., "Navigate to your knowledge base → click Settings → adjust chunk size") or cut them and link to the dedicated pages that already cover this (e.g., /build/knowledge/advanced-knowledge-retrieval).

⚠️ Contradictions (1)

• build/agents/build-your-agent/triggers.mdx:41-43 describes the pause UI as a "pause button" / "resume button" next to each trigger. The sibling page build/agents/build-your-agent/agent-triggers/scheduled-triggers.mdx:151 describes the same action as "simply toggle off the schedule." These are different UI metaphors for what appears to be the same control. Verify which term matches the actual UI label and align both pages.

✅ Clean files (3)

build/agents/create-an-agent.mdx, build/tools/create-a-tool.mdx, build/workforces/create-a-workforce.mdx

🔋 Credit usage

Item	Count
Files reviewed	8
Context pages read	3
Total lines processed	~1,492

Files read: build/agents/build-your-agent/triggers.mdx (126 lines), build/agents/create-an-agent.mdx (96 lines), build/agents/evals/introduction.mdx (109 lines), build/knowledge/create-knowledge.mdx (197 lines), build/tools/create-a-tool.mdx (107 lines), build/workforces/create-a-workforce.mdx (50 lines), get-started/introduction.mdx (169 lines), get-started/quick-start-guide.mdx (141 lines), get-started/core-concepts/agents.mdx (230 lines), get-started/core-concepts/knowledge.mdx (110 lines), build/agents/build-your-agent/agent-triggers/scheduled-triggers.mdx (159 lines)

[jordanc-relevanceai]

You changed all callouts to that. It should only be academy ones. The other ones on the page should stay the same. Please also make sure the info one is the blue info bubble, not the grey one like it is now.

[claude]

Fixed the callout types per review feedback.

What changed:

• build/agents/create-an-agent.mdx: Reverted the two non-Academy <Info> blocks (Option 1 Inventor tip and Option 2 Marketplace tip) back to <Tip>. The Academy CTA at the bottom of the page stays as <Info>.
• build/workforces/create-a-workforce.mdx: Reverted the "Map out your process before building" callout back to <Tip>. The Academy CTA stays as <Info>.

All other files were already correct — only Academy callouts (those linking to relevanceai.com/academy) are <Info> blocks; all other callouts remain <Tip>.

[github-actions]

🎯 Vibe check

Reviewed: 8 files (6 with issues, 2 clean)

Scores

	Dimension	Score	What's holding it back
🔴	Consistency	4/10	create-knowledge.mdx uses Title Case for every heading — 18+ violations. introduction.mdx has 5 heading + 3 card title case violations and one banned word ("seamlessly"). triggers.mdx and quick-start-guide.mdx drop "Agent", "Tool", and "Trigger" to lowercase throughout.
🟡	Technical clarity	7/10	create-knowledge.mdx links to itself using absolute https://relevanceai.com/docs/... URLs (three links) instead of root-relative paths. triggers.mdx uses <Columns cols={2}> instead of <CardGroup cols={2}> — unverified whether it renders card tiles correctly.
🟡	Non-technical clarity	7/10	create-knowledge.mdx's later sections (Retrieval Settings, Query Transformation, Knowledge Prioritization) drop jargon without context, but the core creation steps are readable. Other pages are accessible.
🟡	Structure	6/10	introduction.mdx has two explicitly prohibited section types ("Quick Links" and "How It Works"). create-knowledge.mdx has an H1 inside the body when the frontmatter title already serves that role — risks rendering a double H1.

Score key: 🟢 9–10, 🟡 6–8, 🔴 1–5.

✨ Overall vibe: The how-to pages (create-an-agent, create-a-workforce) are clean and follow the template well. The main drag is create-knowledge.mdx, which reads like it came from a different style guide entirely — Title Case throughout, absolute URLs, and a nested H1. introduction.mdx and the trigger/quick-start pages have widespread product term capitalization issues that need a systematic pass.

🔧 Issues (22)

Banned word

• get-started/introduction.mdx:19 — "Integrate seamlessly with your existing tools" — "seamlessly" is a banned word; rewrite as "Integrate directly with your existing tools" or similar

Heading case — get-started/introduction.mdx

• :23 — ## Quick Links → ## Quick links
• :57 — ## Core Features → ## Core features
• :103 — ## Build your first agent → ## Build your first Agent (Agent is a product name)
• :111 — ## How It Works → ## How it works
• :141 — ## Learning Resources → ## Learning resources

Card title case — get-started/introduction.mdx

• :28 — title="Create Your First Agent" → "Create your first Agent"
• :38 — title="Watch Video Tutorials" → "Watch video tutorials"
• :44 — title="Get Help & Support" → "Get help & support"

Heading case — build/knowledge/create-knowledge.mdx (Title Case used throughout; all need sentence case)

• :8 — # Create and Organize Knowledge → # Create and organize Knowledge
• :10 — ## Creating Knowledge Sources → ## Creating Knowledge sources
• :15 — ### Manual Input → ### Manual input
• :29 — ### Upload Documents → ### Upload documents
• :45 — ### Extract Website Content → ### Extract website content
• :58 — ### Connect Third-Party Integrations → ### Connect third-party integrations
• :68 — ## Organizing Your Knowledge → ## Organizing your Knowledge
• :72 — ### Structured Tables → ### Structured tables
• :81 — ### Document Chunking → ### Document chunking
• :90 — ### Knowledge Collections → ### Knowledge collections
• :99 — ## Enhancing Knowledge Quality → ## Enhancing knowledge quality
• :103 — ### Metadata Enrichment → ### Metadata enrichment
• :112 — ### Knowledge Verification → ### Knowledge verification
• :121 — ### Semantic Organization → ### Semantic organization
• :147 — ## Advanced Knowledge Configuration → ## Advanced knowledge configuration
• :149 — ### Retrieval Settings → ### Retrieval settings
• :157 — ### Knowledge Prioritization → ### Knowledge prioritization
• :167 — ### Query Transformation → ### Query transformation
• :177 — ## Best Practices for Knowledge Management → ## Best practices for knowledge management

Absolute URLs — build/knowledge/create-knowledge.mdx

• :191 — https://relevanceai.com/docs/build/knowledge/access-knowledge → /build/knowledge/access-knowledge
• :192 — https://relevanceai.com/docs/build/knowledge/enrich-with-tool → /build/knowledge/enrich-with-tool
• :193 — https://relevanceai.com/docs/build/knowledge/advanced-knowledge-retrieval → /build/knowledge/advanced-knowledge-retrieval

Product term capitalization — build/agents/build-your-agent/triggers.mdx

• :12–14 — Bullets: "Set your agent to run", "Configure your agent to respond", "Allow users to directly engage with the agent" — "agent" → "Agent"
• :18 — "select the agent you want to configure" → "Agent"
• :22,24,31,35,39,40,41,46,50,52,54,56,57 — Widespread lowercase "trigger/triggers" throughout the managing section; product feature references should be "Trigger/Triggers" (e.g. "once you've set up triggers" → "once you've set up Triggers", "Find the trigger you want" → "Find the Trigger you want")

Product term capitalization — build/tools/create-a-tool.mdx

• :73 — heading ## Editing shared tools → ## Editing shared Tools
• :74–81 — "tool" and "agent" used 7 times in lowercase; each should be "Tool" and "Agent" (e.g. "When you open a tool that is used by two or more agents" → "When you open a Tool that is used by two or more Agents")

Product term capitalization — get-started/quick-start-guide.mdx

• :49–50 — "Review the generated agent" → "Agent"; link text "suggest tools" → "Tools"; "your tools need" → "Tools"
• :65 — "Review the prompt, tools, and settings" → "Tools"
• :79–80 — Link texts "tools" and "knowledge" → "Tools" and "Knowledge" (compare: create-an-agent.mdx:79–80 has these correctly capitalized — these two pages share the same steps)
• :88 — "give your agent a task" → "Agent"
• :121 — "@ mention any agent in your project" → "Agent"
• :128,130 — "clone an agent" → "Agent"; "the agent's prompt, tools" → "Agent's … Tools"

🧩 Component suggestions (1)

• build/agents/build-your-agent/triggers.mdx:85–90 — <Columns cols={2}> wraps four <Card> elements with href props. <CardGroup cols={2}> is the standard Mintlify component for linked card grids (used consistently elsewhere in the repo). <Columns> is not seen elsewhere for this pattern — verify it renders card titles and icons correctly, or swap to <CardGroup cols={2}>.

🏗️ Page structure (4)

• get-started/introduction.mdx:23–54 — "Quick Links" section (## Quick Links + <CardGroup cols={2}>) is explicitly prohibited by CLAUDE.md: "No redundant 'Quick Links' sections that duplicate sidebar navigation." Four of the four links (Quick Start, Chat, YouTube, Support) are already reachable from the sidebar. Remove the section or fold the content into the intro paragraph.

• get-started/introduction.mdx:111–119 — "How It Works" section is explicitly flagged in CLAUDE.md: "No 'How It Works' sections that repeat information already explained elsewhere on the page." The five-step numbered list covers the same ground as the intro paragraph at lines 7–9 and the Core Features section at lines 57–101. Remove or fold into the intro.

• build/knowledge/create-knowledge.mdx:7 — # Create and Organize Knowledge is an H1 inside the file body. Frontmatter title: "Knowledge" already renders as the page H1 — adding a second one risks a double H1 in the rendered output and breaks heading hierarchy. Remove the H1 and let the frontmatter title stand, or promote its text into the frontmatter title / description fields.

• build/agents/evals/introduction.mdx:92 — "Next: Learn how to create Tests…" sits between the main content and the FAQ block. Reader momentum flows better when navigational "next" cues come after the FAQ, not before it. Move line 92 to after the </AccordionGroup> closing tag (currently line 104).

✅ Clean files (2)

build/agents/create-an-agent.mdx, build/workforces/create-a-workforce.mdx

🔋 Credit usage

Item	Count
Files reviewed	8
Context pages read	2
Total lines processed	~1,340

Files read: triggers.mdx (126 lines), create-an-agent.mdx (96 lines), evals/introduction.mdx (109 lines), create-knowledge.mdx (197 lines), create-a-tool.mdx (107 lines), create-a-workforce.mdx (50 lines), introduction.mdx (170 lines), quick-start-guide.mdx (141 lines), core-concepts/agents.mdx (230 lines), core-concepts/knowledge.mdx (110 lines)

[jordanc-relevanceai]

Maybe just change the write up of the callout to not mention workforce but just generally mention building on the platform. I also feel that the callouts can be placed better: it shouldn't be at the end, it should be at the start or maybe at the start of build manually section or something.

[claude]

Actioned all review feedback. Here's what changed:

New callout copy

All 6 pages now use the same platform-general copy (no more workforce-specific mention):

New to Relevance AI? Relevance AI Academy has structured courses that walk you through building on the platform step by step.

Repositioned callouts

Each callout moved from the bottom of the page to earlier in the content flow:

Page	Previous position	New position
build/agents/create-an-agent.mdx	Bottom	Start of Option 3: Build from scratch section
build/tools/create-a-tool.mdx	Bottom	Start of Option 2: Build from scratch section
build/workforces/create-a-workforce.mdx	Bottom	Start of Create your first Workforce section
build/knowledge/create-knowledge.mdx	Bottom	Start of Creating Knowledge Sources section
build/agents/evals/introduction.mdx	Bottom	After the opening paragraph, before the video embed
build/agents/build-your-agent/triggers.mdx	Bottom	Start of Setting up a trigger section

All 6 pages updated in commit afb2c14.

[github-actions]

🎯 Vibe check

Reviewed: 8 files (4 with issues, 4 clean)

Scores

	Dimension	Score	What's holding it back
🔴	Consistency	3/10	create-knowledge.mdx has 20 heading capitalization violations — nearly every heading is title-cased instead of sentence-cased. introduction.mdx adds 13 more (headings, card titles, accordion titles) and one banned word.
🟡	Technical clarity	7/10	create-knowledge.mdx uses absolute internal URLs (lines 195–197) instead of root-relative paths. Everything else is specific and accurate.
🟡	Non-technical clarity	7/10	create-knowledge.mdx reads like a reference dump: 8 subsections with numbered bullet lists under every heading rather than concise prose. The other files are clear and accessible.
🟡	Structure	6/10	introduction.mdx contains two sections explicitly prohibited by CLAUDE.md — a "Quick Links" block that duplicates sidebar navigation, and a "How It Works" block that repeats content already on the page. evals/introduction.mdx has a minor sequencing issue.

Score key: 🟢 9–10, 🟡 6–8, 🔴 1–5.

✨ Overall vibe: Four of the eight files (create-an-agent.mdx, create-a-tool.mdx, create-a-workforce.mdx, quick-start-guide.mdx) are clean and follow the style guide well — tight prose, correct sentence case, proper components. The PR falls apart on create-knowledge.mdx, which looks like it was written against a much older style guide and brings in ~20 heading capitalization errors plus absolute internal links. introduction.mdx needs the Quick Links and How It Works sections removed per CLAUDE.md rules.

🔧 Issues (35)

get-started/introduction.mdx

• :19 — **Integrate seamlessly** — "seamlessly" is a banned word. Rewrite: **Integrate with existing tools** or similar.
• :27 — Card title "Create Your First Agent" → sentence case: "Create your first Agent"
• :39 — Card title "Watch Video Tutorials" → "Watch video tutorials"
• :45 — Card title "Get Help & Support" → "Get help & support"
• :57 — heading ## Core Features → ## Core features
• :111 — heading ## How It Works → ## How it works
• :121 — heading ## Popular Use Cases → ## Popular use cases
• :124 — Accordion title "Customer Support" → "Customer support"
• :127 — Accordion title "Sales & Lead Qualification" → "Sales & lead qualification"
• :130 — Accordion title "Content Creation" → "Content creation"
• :133 — Accordion title "Data Analysis & Reporting" → "Data analysis & reporting"
• :136 — Accordion title "Scheduling & Coordination" → "Scheduling & coordination"
• :141 — heading ## Learning Resources → ## Learning resources
• :153 — Card title "Blog & Updates" → "Blog & updates"

build/knowledge/create-knowledge.mdx

• :7 — H1 # Create and Organize Knowledge → # Create and organize Knowledge ("Organize" is not a proper noun)
• :12 — heading ## Creating Knowledge Sources → ## Creating Knowledge sources
• :19 — heading ### Manual Input → ### Manual input
• :35 — heading ### Upload Documents → ### Upload documents
• :49 — heading ### Extract Website Content → ### Extract website content
• :61 — heading ### Connect Third-Party Integrations → ### Connect third-party integrations
• :72 — heading ## Organizing Your Knowledge → ## Organizing your Knowledge
• :77 — heading ### Structured Tables → ### Structured tables
• :85 — heading ### Document Chunking → ### Document chunking
• :94 — heading ### Knowledge Collections → ### Knowledge collections
• :103 — heading ## Enhancing Knowledge Quality → ## Enhancing Knowledge quality
• :109 — heading ### Metadata Enrichment → ### Metadata enrichment
• :119 — heading ### Knowledge Verification → ### Knowledge verification
• :125 — heading ### Semantic Organization → ### Semantic organization
• :151 — heading ## Advanced Knowledge Configuration → ## Advanced Knowledge configuration
• :154 — heading ### Retrieval Settings → ### Retrieval settings
• :163 — heading ### Knowledge Prioritization → ### Knowledge prioritization
• :171 — heading ### Query Transformation → ### Query transformation
• :181 — heading ## Best Practices for Knowledge Management → ## Best practices for Knowledge management
• :193 — heading ## Related Features → ## Related features
• :195–197 — three "Related Features" links use absolute URLs (https://relevanceai.com/docs/build/...). CLAUDE.md requires root-relative paths. Change to /build/knowledge/access-knowledge, /build/knowledge/enrich-with-tool, and /build/knowledge/advanced-knowledge-retrieval.

🧩 Component suggestions (2)

• build/agents/build-your-agent/triggers.mdx:89 — <Columns cols={2}> is not a standard Mintlify component and isn't used anywhere else visible in this repo. The standard equivalent for a grid of linked Cards is <CardGroup cols={2}>. Swap the wrapper; the inner <Card> elements are fine as-is.

• get-started/introduction.mdx:23–54 — The ## Quick Links <CardGroup> links to the quick-start guide, chat intro, YouTube, and support. All four destinations already appear in the sidebar. Per CLAUDE.md, redundant Quick Links sections should be removed. The cards under ## Core Features (lines 58–101) serve the same navigational purpose and are more meaningful — remove Quick Links entirely.

🏗️ Page structure (3)

• get-started/introduction.mdx:23 — The ## Quick Links section is explicitly prohibited by CLAUDE.md: "No redundant 'Quick Links' sections that duplicate sidebar navigation." Remove the section; the Core Features CardGroup below it already provides navigation.

• get-started/introduction.mdx:111 — The ## How It Works numbered list (build agents → give them tools → add knowledge → connect workforces → deploy) restates exactly what the Core Features cards and the "Build your first agent" Supademo already cover. CLAUDE.md prohibits "How It Works sections that repeat information already explained elsewhere on the page." Remove or collapse into the intro paragraph.

• build/agents/evals/introduction.mdx:96 — The Next: navigation sentence appears before the FAQ <AccordionGroup>, so readers who scroll to the bottom of the FAQ never see it. Move the Next: line to after the closing </AccordionGroup> tag (after line 108), or replace it with a ## What's next? section at the end. This matches the pattern used in test-sets.mdx where the same construct also sits above the FAQ — worth fixing in both files.

✅ Clean files (4)

build/agents/create-an-agent.mdx, build/tools/create-a-tool.mdx, build/workforces/create-a-workforce.mdx, get-started/quick-start-guide.mdx

🔋 Credit usage

Item	Count
Files reviewed	8
Context pages read	4
Total lines processed	~1,570

Files read: triggers.mdx (126 lines), create-an-agent.mdx (96 lines), evals/introduction.mdx (109 lines), create-knowledge.mdx (198 lines), create-a-tool.mdx (107 lines), create-a-workforce.mdx (50 lines), get-started/introduction.mdx (170 lines), quick-start-guide.mdx (141 lines), core-concepts/agents.mdx (230 lines), evals/test-sets.mdx (132 lines), core-concepts/knowledge.mdx (110 lines), core-concepts/tools.mdx (101 lines)