Building a Granular Technical Blog Pipeline for Four Fleet Sites

```html

This session involved architecting and implementing an automated technical blog generation system across four separate domain properties: queenofsandiego.com, sailjada.com, dangerouscentaur.com, and burialsatseasandiego.com. The goal was to create a mechanism that captures development work at the granular level—tracking specific file modifications, infrastructure changes, and command execution—and automatically generates detailed technical posts for engineering visibility.

Problem Statement

Without a structured technical documentation system, development work across multiple properties becomes invisible to stakeholders. The need was for near-real-time technical blog posts that record granular details: exact file paths modified, infrastructure resources changed, command execution, and architectural decisions. This required automation that would trigger at the end of development sessions without manual intervention.

Architecture Overview

The solution consists of four primary components:

Session Capture: Claude Code's session transcripts recording all file operations and commands
Blog Generator: Python tool parsing session data and creating HTML posts
Infrastructure Layer: S3 buckets, CloudFront distributions, and DNS records for each tech blog domain
Automation Hook: Stop hook script that triggers on session end, generating and deploying posts

Infrastructure Setup

Each tech blog required its own S3 bucket and CloudFront distribution:

tech-queenofsandiego-com S3 bucket with CloudFront distribution (using existing *.queenofsandiego.com wildcard ACM certificate)
tech-sailjada-com S3 bucket with CloudFront distribution (using existing *.sailjada.com wildcard ACM certificate)
dc-sites S3 bucket with existing CloudFront wildcard distribution at dangerouscentaur.com (added tech.dangerouscentaur.com as additional domain via Route53 CNAME through Namecheap)
tech-burialsatseasandiego-com S3 bucket with new CloudFront distribution (ACM cert validation via GoDaddy DNS API, as burialsatseasandiego.com uses GoDaddy nameservers)

DNS configuration varied by registrar:

Route53 domains: Direct alias records pointing CloudFront distributions
Namecheap: CNAME records for dangerouscentaur.com tech blog
GoDaddy: CNAME records for burialsatseasandiego.com, with ACM validation also via GoDaddy DNS API

Core Implementation Files

tech_blog_generator.py — Main parsing and HTML generation engine:

Parses Claude Code JSONL session transcripts from ~/.claude/sessions/
Extracts file modifications (paths, operations: create/write/edit)
Extracts command execution history with parameters (no credential values)
Organizes findings into HTML post structure with sections for files, commands, decisions, and next steps
Sanitizes sensitive data: removes credentials, API keys, database passwords, personal identifiers
Generates timestamped posts with appropriate site slug (qos, jada, dc, bats)

tech_blog_init.py — Infrastructure initialization script:

Creates S3 buckets with private ACL and versioning enabled
Provisions CloudFront distributions with appropriate origin configurations
Handles certificate validation: Route53 for AWS-managed domains, DNS API calls for GoDaddy/Namecheap
Creates DNS records (alias or CNAME) pointing to CloudFront endpoints
Stores infrastructure configuration in ~/.claude/projects/memory/infrastructure_config.json for retrieval
Supports dry-run mode for validation before actual resource creation

tech_blog_stop.sh — Claude Code stop hook:

Triggered automatically when Claude Code session ends
Invokes blog generator on current session transcript
Determines target blog(s) based on files modified (matching file paths to site domains)
Uploads generated HTML to appropriate S3 bucket
Invalidates CloudFront cache for the tech blog domain
Logs execution details to ~/.claude/logs/tech_blog_hook.log

Site Integration

Updated queenofsandiego.com/index.html to add "Tech Blog" link in the Ship's Papers navigation dropdown menu, linking to https://tech.queenofsandiego.com/. Similar nav updates would apply to other fleet sites, making technical documentation discoverable alongside operational resources.

Session Data Flow

Claude Code sessions automatically capture all interactions in JSONL format. Each session transcript includes:

tool_use entries for file operations with exact paths and operation type
command entries with full command text (excluding values for sensitive parameters)
Timestamps for each operation
User request context and reasoning notes

The blog generator parses these transcripts, groups related operations, and synthesizes them into a narrative technical post with specific infrastructure details, architectural decisions, and technical reasoning.

Credential and Sensitive Data Handling

Critical security measure: all generated posts are sanitized to remove:

AWS access keys, secret keys, session tokens
API credentials (GoDaddy, Namecheap, etc.)
Database passwords and connection strings
Personal information or email addresses beyond public context
Specific values for authentication parameters

Commands and file paths are included for technical clarity; only the sensitive values are stripped.

Tested Deployment

All four tech blog sites are now live and serving HTTP traffic:

https://tech.queenofsandiego.com/ — CloudFront distribution, Route53 alias
https://tech.sailjada.com/ — CloudFront distribution, Route53 alias
https://tech.dangerouscentaur.com/ — Wildcard CF distribution, Namecheap CNAME
https://tech.burialsatseasandiego.com/ — CloudFront distribution, GoDaddy CNAME with ACM validation

Test post was generated from this session transcript and deployed to all four sites.

Key Architectural Decisions

Session-based automation: Avoids need for manual blog post creation; developers focus on work, not documentation
Granular capture: Post includes specific file