docs(TSP-1447): Add Confluent Cloud integration documentation

[claude]

Summary

• Adds a new integration page for Confluent Cloud at integrations/popular-integrations/confluent-cloud.mdx
• Updates docs.json to include the page under the Developer integrations group
• Covers authentication setup (cluster-scoped API keys, REST endpoint, cluster ID), all six available tool steps, use cases, and a setup guide with steps for retrieving credentials from Confluent Cloud
• Includes a prominent warning callout that API keys must be cluster-scoped (org-level keys return 401 errors)

Closes https://linear.app/relevance/issue/TSP-1447/

🤖 Generated with Claude Code

[mintlify]

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

Project	Status	Preview	Updated
relevanceai	🟢 Ready	View Preview	Sep 15, 2026, 3:04 AM

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

[linear]

TSP-1447

[github-actions]

🎯 Vibe check

Reviewed: 2 files (1 with issues, 1 clean)

Scores

	Dimension	Score	What's holding it back
🟡	Consistency	8/10	Two product term capitalization misses: "agents" and "tools" (lowercase) when referring to Relevance AI Agents and Tools.
🟢	Technical clarity	9/10	Precise and actionable throughout. The cluster-scoped key warning is well-placed, the lkc-xxxxxx format example is helpful, and the "secret is only shown once" gotcha is called out.
🟡	Non-technical clarity	8/10	Opens with "fully managed Apache Kafka service" without explaining what Kafka is — readers who don't already know won't have context for why this integration matters.
🟡	Structure	7/10	The credential-fetching section comes after the connection steps, but readers need credentials in hand before they can connect. The "Steps to connect" numbered list uses plain markdown while the very next section uses <Steps> — inconsistent on the same page.

✨ Overall vibe: Solid integration doc — accurate, specific, and well-warned. The main friction is organizational: get-credentials should come before connect-the-integration so the reader isn't sent back up the page, and a couple of capitalization fixes will keep terminology consistent with the rest of the docs site.

🔧 Issues (2)

• integrations/popular-integrations/confluent-cloud.mdx:7 — "lets your agents connect" → "lets your Agents connect" (Relevance AI product term)
• integrations/popular-integrations/confluent-cloud.mdx:58 — "for use in your agents and tools" → "for use in your Agents and Tools" (Relevance AI product terms)

🧩 Component suggestions (2)

• integrations/popular-integrations/confluent-cloud.mdx:33–40 — "Steps to connect" is a sequential procedure written as a plain numbered markdown list, while the very next section (lines 44–54) and sibling pages (github.mdx) use <Steps>. Wrap these six steps in a <Steps> component with a <Step> for each item so the page is visually consistent.

• integrations/popular-integrations/confluent-cloud.mdx:86–92 — "Example use cases" uses bold labels (**Event streaming automation**:) as titles within a bullet list. CLAUDE.md prohibits mixing bold text with heading function. Since the descriptions are short (one line each), either drop the bold and use plain bullets, or promote each to a #### heading. Accordions would be overkill for one-liners.

🏗️ Page structure (1)

• integrations/popular-integrations/confluent-cloud.mdx — The page opens with "Connect the integration" (which requires credentials) before "Getting your credentials from Confluent Cloud". Readers will hit step 4 ("Enter your API Key…") before they know where to find those values. Swap the two sections so credential retrieval comes first, then the connection steps.

✅ Clean files (1)

docs.json — navigation addition is correct; confluent-cloud is appropriately placed in the "Developer" group.

🔋 Credit usage

Item	Count
Files reviewed	2
Context pages read	1
Total lines processed	~572

Files read: integrations/popular-integrations/confluent-cloud.mdx (109 lines), docs.json (navigation section, ~20 lines), integrations/popular-integrations/github.mdx (443 lines)

[github-actions]

🎯 Vibe check

Reviewed: 2 files (1 with issues, 1 clean)

Scores

	Dimension	Score	What's holding it back
🟢	Consistency	8/10	"agents and tools" (line 56) and "agents" (line 83) need capitalization — both refer to Relevance AI product features.
🟢	Technical clarity	9/10	Excellent gotcha coverage (scoped API key warning, 401 FAQ, secret shown only once). Minor: credential retrieval steps appear after connection steps, so a reader encounters "Enter your API Key" before they've been told where to find it.
🟡	Non-technical clarity	7/10	"Apache Kafka", "cluster", and "consumer groups" are introduced without any definition. Even a one-sentence aside would help readers who are new to Confluent but using Relevance AI.
🟡	Structure	7/10	"Steps to connect" section comes before "Getting your credentials" — inverted ordering. No closing CTA for a how-to page that leaves readers without a clear next step.

Score key: 🟢 9–10, 🟡 6–8, 🔴 1–5.

✨ Overall vibe: Technically sharp — the Warning callout about cluster-scoped keys and the 401 FAQ are exactly the kind of gotcha content that makes integration docs useful. The main structural problem is that "Steps to connect" appears before "Getting your credentials", which means readers hit "Enter your API Key and API Secret" before the page has told them how to find it; flip those two sections.

🔧 Issues (3)

• integrations/popular-integrations/confluent-cloud.mdx:56 — "agents and tools" → "Agents and Tools" (Relevance AI product features are proper nouns)
• integrations/popular-integrations/confluent-cloud.mdx:81 — "Trigger agent actions" → "Trigger Agent actions"
• integrations/popular-integrations/confluent-cloud.mdx:83 — "Use agents to produce" → "Use Agents to produce"

🏗️ Page structure (2)

• integrations/popular-integrations/confluent-cloud.mdx — "Steps to connect" (lines 31–38) tells the reader to enter credentials before "Getting your credentials from Confluent Cloud" (lines 40–52) explains how to find them. Reorder: move the credential retrieval <Steps> block before the connection steps so the page reads: what you need → how to get it → how to use it.
• integrations/popular-integrations/confluent-cloud.mdx — No closing CTA. This is a how-to page and the reader has no obvious next step after connecting. A short ## What's next? pointing to /build/agents/build-your-agent/tools (to wire up the tool steps just listed) and /integrations/popular-integrations/slack or another integration as a parallel example would round it off.

✅ Clean files (1)

docs.json — Confluent Cloud is correctly placed in the "Developer" group under popular integrations. Valid JSON, no structural issues.

🔋 Credit usage

Item	Count
Files reviewed	2
Context pages read	1 (slack.mdx for pattern reference)
Total lines processed	~420

Files read: integrations/popular-integrations/confluent-cloud.mdx (101 lines), docs.json (140 lines sampled), integrations/popular-integrations/slack.mdx (320 lines)