AWS Blog Co-Author¶

Capability Slug: aws_blog_co_author Plan Tier: Command Credit Cost: 35 credits per draft Category: Content Creation Default blog type: technical_howto Default target audience: technical_decision_makers

Overview¶

AWS Blog Co-Author generates AWS Marketplace blog drafts that are structured, formatted, and compliance-checked the way an AWS Bar Raiser would expect. It encodes the editorial rubric an AWS reviewer applies during co-authored blog review, runs your topic + author info through Claude with that rubric in the prompt, then validates the output against codified compliance rules so you know what's blocking publication before you submit.

The capability is opinionated on purpose. Co-authored AWS blogs have a specific shape — seven sections, defined ownership per section, AWS-service hyperlinks on first mention, WCAG 2.1 alt-text on figures, long-tail keywords woven in. Trying to remember all of that while writing leads to revision cycles. AWS Blog Co-Author bakes the rubric into the generation step.

What gets generated¶

For every run, the capability returns:

Output	Description
`draft_id`	The persisted `AwsBlogDraft` record. Reference it later to revise, re-validate, or hand off.
`blog_draft`	The structured schema — metadata, authors, sections, SEO, action items. Useful for downstream automation.
`blog_markdown`	The rendered Markdown ready for the AWS editorial pipeline.
`compliance_report`	Score 0-100, overall status, four category scores, per-rule pass/fail, blocking issues.
`action_items`	Per-owner task list — what AWS needs to do, what the partner needs to do, what's automatic.

The draft is saved to the database as part of the run, so you can iterate without re-paying the 35 credits each time you want to look at the same draft.

The seven-section structure¶

Every draft follows the AWS Marketplace co-authored blog structure exactly, with explicit section ownership:

Order	Section	Owner	Purpose
1	Introduction	AWS	Set context, state the problem, preview the solution
2	Prerequisites	Partner	Specific technical requirements (accounts, tools, versions)
3	Solution Overview	Co-authored	High-level architecture and diagram description
4	Solution Walkthrough	Partner	Step-by-step numbered instructions with code blocks
5	Cleanup	Partner	Commands to tear down the demo environment
6	Conclusion	Co-authored	Summarize outcomes, link to product in Marketplace
7	About the Authors	Auto	Generated from author bios

What "owner" means in the output. The capability returns each section with an explicit owner. Partner-owned sections (Prerequisites, Walkthrough, Cleanup) come back with status: "needs_input" to flag that the partner has to review them for technical accuracy before submission. AWS-owned and co-authored sections come back as status: "draft" — substantive, but expected to go through an editorial pass.

The Bar Raiser rules (encoded in the prompt)¶

The prompt embeds the editorial rules so the model produces a compliant draft on the first attempt:

Title ≤ 75 characters, must contain an AWS service name or "AWS Marketplace", must use an action verb. Example: Deploy [product] using [AWS service] on AWS Marketplace.
Section structure — exactly seven sections in the order above, with the listed owners.
AWS service linking — first mention of any AWS service in the body is hyperlinked. Example: [Amazon EKS](https://aws.amazon.com/eks/).
WCAG 2.1 — every image gets a descriptive alt-text suggestion; every figure gets a descriptive caption. Example: "Figure 3: Architecture diagram showing primary and DR EKS clusters with Trident CSI replication."
Long-tail keywords — specific technical phrases woven naturally into the body, not just the primary keyword.
LLM discoverability — clear numbered steps, labeled figures, explicit outcomes per step, summary lists. Structured for AI parsing.
Author bios — 50-75 words each, covering name, title, company, years of experience, area of expertise.

The compliance validator¶

After generation, AwsBlogComplianceService runs the draft through a checks suite and returns a structured report. Categories and the rules that score them:

Category	Rules
Editorial	`title_length`, `title_aws_reference`, `structure_quality`
Accessibility	`image_alt_text`, `author_bio_length`
SEO	`long_tail_keywords`, `llm_discoverability`, `aws_service_links`
Structure	`section_structure`, `section_balance`

The report includes:

overall_score — weighted percentage across all rules
overall_status — passed, needs_revision, or blocked
category_scores — per-category percentages so you know which dimension to focus on
passed_rules and failed_rules — counts
error_count and warning_count — by severity
blocking_issues — explicit list of failures that would prevent publication
per-rule items: rule_name, status, severity, recommendation

Rules live in the compliance_rules table keyed by category = 'aws_blog', so the rubric can be tuned without redeploying code.

Required and optional parameters¶

Parameter	Required	Description
`topic`	Yes	The blog topic — e.g. "Multi-region disaster recovery for Trident on Amazon EKS"
`partner_company`	Yes	The partner company name — drives ownership labeling and intro framing
`aws_services`	Yes	Array of AWS services covered (e.g. `["Amazon EKS", "Amazon S3", "AWS Backup"]`) — drives the AWS service hyperlinks rule
`authors`	Yes	Array of author objects: `{name, title, company, role, bio}`. Role is `primary`, `co_author`, or `contributor`. Bio is rewritten to 50-75 words automatically.
`blog_type`	No	`technical_howto` (default), `product_overview`, or `customer_success_story`
`target_audience`	No	Defaults to `technical_decision_makers`
`custom_instructions`	No	Free-form additional instructions appended to the prompt
`use_knowledge_base`	No	Defaults to `true`; set to `false` to skip the partner-context KB query

Knowledge base integration¶

If your agent has the query_knowledge_base capability enabled, AWS Blog Co-Author will run a similarity search over your KB before generation:

query: "{partner_company} {topic} AWS blog technical walkthrough"
top_results: 3
min_similarity: 0.7

The top three matches get woven into the prompt as reference material so the draft uses your real product details (feature names, pricing, integration points) instead of generic placeholders. If no results meet the similarity threshold, the run continues without KB context. Pass use_knowledge_base: false to opt out entirely.

When to use it¶

Scenario	Why this capability fits
Co-authored launch blog with AWS	The seven-section structure with explicit AWS / partner ownership is exactly what the editorial team expects
Partnership announcement post	`product_overview` blog type with paired authors
Customer reference with hard numbers	`customer_success_story` blog type, KB-grounded with real customer outcomes
Anchor content for a Marketplace launch	Pair with Generate Content Series (`generate_content_series`) for the supporting social/email assets

It is not:

A general-purpose blog generator. For non-AWS blogs, use Generate Text Content (generate_text) — same pricing structure, no AWS rubric overhead.
An editor for existing drafts. Use Refine Content (refine_content) on an existing blog.
A publication pipeline. The output is editorial-ready Markdown; the AWS submission and review flow is owned by AWS.

Example flow¶

Pick a topic and authors. Decide which AWS services are central, which authors will be on the byline, and which one is the primary (vs. co_author or contributor).
Run the capability. Navigate to Content Creation → AWS Blog Co-Author, fill in topic / partner company / AWS services / authors, leave technical_howto as the default, and click Generate.
Review the compliance report. Look at blocking_issues first. If empty, you're publish-ready. If not, the recommendations tell you what to change.
Hand off action items. The action_items array is keyed by assignee (partner or aws). Send each side their list.
Iterate from the draft record. The draft_id lets you re-run validation after edits without paying the 35 credits again.

API access¶

AWS Blog Co-Author is exposed through the agent runtime as aws_blog_co_author. Partner API keys scoped with manage_content plus the agent capability slug can generate drafts programmatically through the agent execution endpoint. The response includes the draft ID so you can fetch the AwsBlogDraft record later through your own integration. See Partner API Capabilities and Authentication & API Keys.