Content Studio¶

Umbrella for 4 capabilities — see the cluster table below for slugs, costs, and links Plan Tier: Accelerate (full set available) Backend: Claude (all four capabilities)

Overview¶

Content Studio is the umbrella for the four agent capabilities that handle the create-and-refine motion: generate single pieces, generate coordinated series, refine drafts against feedback, and auto-extract brand voice from a website to ground all of the above. Each capability is independently scoped — pick only the slugs you need on a Partner API key.

This page is the consolidated reference for the cluster. For the higher-tier specialized content capabilities, see AWS Blog Co-Author (aws_blog_co_author, 35 credits) and AI Tool Listing Card Generator (agent_tool_card, 20-40 credits) — those have purpose-built rubrics and persistence the Content Studio cluster doesn't.

The four capabilities at a glance¶

#	Slug	Name	Credit cost
1	`generate_text`	Generate Text Content	5 / 10 / 20 (by length)
2	`generate_content_series`	Generate Content Series	`series_count × (5 / 10 / 20)`
3	`refine_content`	Refine Content	10 (flat)
4	`enrich_brand_voice`	Enrich Brand Voice from Website	8 (flat)

Per-capability detail follows.

1. `generate_text` — single-piece content generation¶

The workhorse capability. Generates a single piece of content of a given type, length, tone, and topic, grounded in your brand-voice context.

Required: content_type, topic. Cost: scales by length:

`length`	Credits	Word count target
`short`	5	~150-300 words
`medium` (default)	10	~500-800 words
`long`	20	~1000-1500 words

Supported content types (each gets type-specific prompt instructions):

`content_type`	Output shape
`blog_post`	Title, intro, body sections, conclusion with CTA
`article`	Long-form article with structured sections
`social_post`	Platform-conscious post with hook + CTA
`email`	Subject line, body with greeting/content/CTA, signature
`product_description`	Marketplace-aligned product description
anything else	Falls through to a generic "Generate high-quality {type} content" instruction so you can extend without code changes

Optional parameters: tone (defaults to professional), length (defaults to medium).

When to use: any standalone content piece. For coordinated series across multiple pieces, see generate_content_series instead.

2. `generate_content_series` — coordinated multi-piece series¶

Generates a series of related content pieces around one theme — useful for launches, campaigns, and email sequences where the pieces need to flow together.

Required: content_type, topic, series_count. Optional: series_type (defaults to general), length, tone.

Cost: series_count × per-piece cost from the same length matrix above.

Series shape	Cost
3 pieces × medium length	3 × 10 = 30 credits (default)
5 pieces × short length	5 × 5 = 25 credits
5 pieces × long length	5 × 20 = 100 credits

The series is generated as a single Bedrock call so the pieces stay coherent — themes, references, and progression flow across the set rather than being independently regenerated.

When to use: launch sequences, drip campaigns, multi-part blog series, coordinated social-media pushes around a single moment.

3. `refine_content` — iterate on a draft¶

The iteration step. Takes a draft and structured analysis feedback, returns a revised draft that addresses the feedback.

Required: original_content, analysis_results. Optional: refinement_instructions (defaults to Improve the content based on analysis feedback).

Cost: flat 10 credits.

The analysis_results parameter is intentionally generic — any structured feedback works:

SEO score breakdowns from seo_content_analysis
Brand-alignment scores from analyze_content (documentation in progress)
Readability or compliance findings
Manual review comments structured as JSON

The capability extracts content from dependency outputs as well — if original_content isn't directly provided but the agent run includes a content-producing step upstream, it pulls the latest content or refined_content from the dependency chain.

When to use: between an analysis pass and a publish step. Run an analyzer, hand the results to refine, get a polished draft.

4. `enrich_brand_voice` — auto-extract from your website¶

Fetches the company's website and uses Claude to extract brand voice signals — tone, value propositions, target audience, competitive differentiators, product highlights — then writes them back to the Company record.

Required: company_id. Cost: flat 8 credits (covers the website fetch plus the AI extraction).

What gets populated on the Company record:

tone_of_voice
target_audience
value_propositions (array)
competitive_differentiators (array)
target_keywords (array)
Product highlights for each product on the company

After this runs, every other capability that pulls brand-voice context produces output that sounds like the partner instead of generic marketing speak.

When to use: onboarding a new partner, refreshing brand voice after a website redesign, or filling gaps surfaced by Content Readiness Analyzer.

A typical end-to-end content motion¶

The four capabilities chain naturally with the SEO and analysis clusters:

enrich_brand_voice              → "Populate brand voice from the website"
   ↓
content_readiness_analyzer      → "Confirm the brand profile is complete enough"
   ↓
keyword_research                → "Pick a target keyword for the piece"
(SEO Intelligence)
   ↓
generate_text (or generate_     → "Draft the piece(s)"
content_series)
   ↓
seo_content_analysis            → "Score the draft against SEO factors"
(SEO Intelligence)
   ↓
refine_content                  → "Iterate on the draft against the analysis"
   ↓
generate_meta_tags              → "Title and description for publishing"
(SEO Intelligence)

Total credits for one full pass: roughly 8 + 10 + 5 + 10 + 8 + 10 + 5 = 56 credits for an end-to-end "from brand-voice setup to publish-ready content" run on a single piece.

Credit costs reference¶

Sourced from getEstimatedCredits() on each capability handler.

Capability	Cost formula	Default
`generate_text`	`length: short=5, medium=10, long=20`	10
`generate_content_series`	`series_count × (length: 5/10/20)`	30 (3 medium pieces)
`refine_content`	flat	10
`enrich_brand_voice`	flat	8

When to use the Content Studio cluster vs. the flagship capabilities¶

Need	Use
Standalone blog post, email, social post, or product description	`generate_text`
Coordinated launch sequence across 3-5 pieces	`generate_content_series`
Iterate on a draft using structured analyzer feedback	`refine_content`
Auto-populate brand voice from a website	`enrich_brand_voice`
Bar-raiser-compliant AWS Marketplace blog with co-author handoff	AWS Blog Co-Author (`aws_blog_co_author`) — flagship 35 cr
Listing card for AWS AI Agents & Tools with MCP/OpenAPI schemas	AI Tool Card Generator (`agent_tool_card`) — flagship 20-40 cr

The flagship capabilities have purpose-built rubrics, compliance validation, and persistence (AwsBlogDraft and AgentToolCard records). The Content Studio cluster is the general-purpose lower-tier — same Claude backend, simpler scope, lower cost.

What this is not¶

A CMS or publishing pipeline. Output is content; publishing happens elsewhere.
An image or video generator. See Image Generation (generate_image) and Media Studio for visuals.
A content analysis or quality scoring tool. See SEO Intelligence (seo_content_analysis) and Analysis Hub (analyze_content).
Ad-hoc website scraping. enrich_brand_voice works on a company's own website (the one stored on the Company record) — it's not a general-purpose URL fetcher. For arbitrary URL fetching see fetch_external_url in the capability registry.

Plan availability¶

Capability	Starter	Accelerate	Command
`generate_text`	—	✅	✅
`generate_content_series`	—	✅	✅
`refine_content`	—	✅	✅
`enrich_brand_voice`	—	✅	✅
`aws_blog_co_author` (flagship)	—	—	✅
`agent_tool_card` (flagship)	—	—	✅

API access (api_access scope) starts on Accelerate. See Partner API Capabilities.

API access¶

All four capabilities are exposed through the agent runtime under their snake_case slugs. Sample scoping for a content team:

"scoped_capabilities": [
  "api_access",
  "manage_content",
  "manage_companies",
  "generate_text",
  "generate_content_series",
  "refine_content",
  "enrich_brand_voice"
]

Add manage_seo and the SEO capability slugs if the team also runs the SEO Intelligence cluster. See Partner API Capabilities and Authentication & API Keys.