withamplifier.com

Source-Driven
Content Generation

Building a system where website content is always accurate, always traceable, and always validated

Two Amplifier Sessions | January 2026 | ACTIVE
The Problem

Content Drift

Website content diverges from source truth over time

Manual Updates
Website content is copied from docs, edited, and quickly becomes stale as source files evolve
No Traceability
When content is wrong, there's no way to know which source file should be the authority
Silent Failures
Broken links, outdated commands, and wrong module names go undetected until users complain
The Discovery
12/12

Community Bundle Links Return 404

Every single bundle link on the live withamplifier.com website is broken. This wasn't a theoretical problem—it was already happening.

Session 1

The Design Phase

Analyzed withamplifier.com HTML structure
Mapped all 8 data-sections and their content patterns
Discovered the 404 crisis
All 12 community bundle links broken—dramatic proof the system is needed
Mapped content sources
13 sections → README.md, MODULES.md, USER_ONBOARDING.md
Created the outline pattern
withamplifier-content.json mapping sources → content
Built the recipe architecture
4-stage validation pipeline: Initialize → Generate → Validate → Output
The Pattern

Sources → Content

Every website element traces back to an authoritative source file

// withamplifier-content.json { "section": "quick-install", "web_element": "data-section='quick-install'", "source_files": ["README.md#installation"], "accuracy": "verbatim", "validation": { "priority": "critical", "checks": ["command_syntax", "url_validity"] } }
Architecture

4-Stage Validation Pipeline

STAGE 1
Initialize
→
STAGE 2
Generate
→
STAGE 3
Validate
→
STAGE 4
Output
Critical
Commands, URLs, installation steps—must be exact
High
Module names, bundle identifiers—must match source
Medium/Low
Descriptions and marketing—can be synthesized
Session 2

The Implementation Phase

Turning design into a working system—through iterative debugging

First run: 6 schema issues
Recipe template system had strict constraints we hadn't anticipated
Iterative debugging
Fixed each issue: description fields, output_key→output, Handlebars syntax
Redesigned output format
From multi-file cross-reference to single unified JSON
3 complete iterations
Each run improved coverage and accuracy
Debugging

6 Schema Fixes

Recipe template constraints required careful iteration

Issues Found
• description field on wrong elements
• output_key not recognized
• Handlebars {{#each}} incompatible
• Missing required fields
• Invalid template references
• Context variable scoping
Fixes Applied
• Moved descriptions to step level
• Changed to output: key format
• Used prose instructions instead
• Added all required schema fields
• Fixed file path references
• Restructured context flow
Evolution

Output Design Matters

Before: Multi-File
• outline.json (structure)
• content.json (text)
• validation.json (rules)
• mapping.json (web elements)

Required cross-referencing 4 files to understand any single section
After: Unified JSON
• Single website-content.json
• Web element mapping embedded
• Source files inline
• Validation rules included

Everything needed in one self-documenting file
Coverage
13
→
33

Sections Mapped

Expanded from initial mapping to cover all 8 data-sections on withamplifier.com

quick-install
main-features
ai-integration
terminal-experience
modules
enterprise
bundles
community
Classification

Accuracy Types

Different content requires different validation strategies

14
Verbatim
Commands, URLs, code—extracted exactly as written in source files
install commands
module names
10
Synthesized
Descriptions derived from source content but reworded for web context
feature descriptions
benefit summaries
9
Marketing
Creative content like taglines that can be generated fresh
hero headlines
CTAs
The Output
83KB

website-content.json

Complete, validated, source-traced content for every section of withamplifier.com

The Value

Content That Can't Drift

Always Accurate
Generated directly from source files—no manual copying
Always Traceable
Every item links to its authoritative source
Always Validated
URLs checked, commands verified, names matched
Always Fresh
Regenerate when sources change—automated updates
Artifacts

What We Built

📋
outlines/withamplifier-content.json
Content mapping with 33 sections, sources, and validation rules
48 KB
⚙️
recipes/website-content-generation.yaml
4-stage validation pipeline recipe
22 KB
✅
generated/website-content.json
Complete output with everything needed
83 KB
Velocity

Development Stats

2
Amplifier Sessions
3
Recipe Iterations
6
Schema Fixes
33
Sections Mapped
Insights

What We Learned

The 404 Discovery
Real broken links on production proved the system wasn't theoretical—it was urgent
Recipe Constraints Are Features
The 6 schema issues forced us to think more carefully about output structure
Output Design Is UX
Unified JSON beats multi-file—consumers shouldn't need to cross-reference
Classification Enables Automation
Verbatim vs synthesized vs marketing determines what validation is possible
Sources

Research Methodology

Data as of: February 20, 2026

Feature status: Active

Research performed:

Gaps: withamplifier.com repo not available locally for independent verification; 404 count from session analysis, not re-tested today

Primary contributors: MADE Team (session-based development)

Next Steps

Ready for Integration

The content generation system is complete and ready to power withamplifier.com updates

Run the Recipe
Execute website-content-generation.yaml to regenerate all content from latest sources
Integrate with CI
Automate content regeneration when source files change
Sessions 5e8b6e41 & 43feb37a | MADE Team
More Amplifier Stories