go-skeleton

2026-02-17 ~ 2026-03-14 · 25 days

README Dreamer

Suffocated under the weight of its own 'production-ready' ambition and AI dreams

“Born for production, died in documentation”

Death Type

README Dreamer

This project was a 'README Dreamer', having dedicated +683 lines to `README.md` and 870 lines to `implementation-playbook.md` detailing AI agent skills. It envisioned 'production-ready' Go projects with 'hexagonal architecture' and 'AI coding agent support' across three distinct types, all before sustaining more than a single day of development.

Cause of Death

1. Flash-in-the-pan development

The entire project lifecycle, from 'initial scaffold' to 'resolve templates', spanned less than 24 hours. All 5 recorded commits were made by 'jose.castillo' on a single day, 2026-03-14, followed by complete silence.

2. Documentation over deployment

`README.md` saw +683 lines added and -313 lines removed across commits. This significant documentation churn suggests a fervent focus on explaining what the project *could* do, rather than shipping what it *did*.

3. AI dreams, human labor

`implementation-playbook.md` boasted 870 lines detailing AI agent skills. Yet, the project's 'Vibe Score: 19/100' explicitly confirmed 'Hand-coded. Respect.', with 'AI tools detected: none' in the actual codebase.

Vibe Score

19/ 100

Hand-coded. Respect.

What They Did

This 'go-skeleton' project aimed to be the ultimate Go scaffolding CLI tool, generating 'production-ready' boilerplate with 'hexagonal architecture'. It ambitiously supported three project types – HTTP API (Gin), CLI (Cobra/Viper), and MCP server (mcp-go) – all while promising 'built-in AI coding agent support' via `AGENTS.md` and extensive skill documentation.

github.com/spf13/cobragithub.com/inconshreveable/mousetrapgithub.com/spf13/pflagGinVipermcp-goGo 1.26.0

Burnout Analysis

Burnout was not the culprit here. All 5 commits landed on 2026-03-14 by 'jose.castillo', indicating a single, intense burst of activity rather than a slow decline. The project simply ceased development after this initial, comprehensive effort, a sprint to the finish line that was also the starting gun.

Dependency Archaeology

The scaffolding tool itself, `go-skeleton`, used a lean 2 dependencies, `github.com/spf13/cobra v1.10.1` being primary for Go 1.26.0. However, it promised to generate projects laden with Gin, Cobra, Viper, and mcp-go, including 3 separate `openai_repository.go.tpl` files (each 378 lines) for each project type, suggesting a generated project dependency count far exceeding its own.

Autopsy: File Structure

├──README.mdReceived +683 lines of pure ambition, becoming a magnificent tome of unfulfilled promises.

├──AGENTS.mdThe manifesto for AI agents that were extensively documented but never truly materialized.

├──template/skills/tdd-workflows-tdd-green/resources/implementation-playbook.mdAn 870-line testament to what AI could do, if only a human had built it first.

├──template/mcp/platform/openai/openai_repository.go.tpl378 lines of OpenAI integration for a Microservice Control Plane that never flew.

├──template/api/platform/openai/openai_repository.go.tplAnother 378 lines, ensuring the API template was AI-ready, even if the API itself wasn't.

├──template/cli/platform/openai/openai_repository.go.tplYet another 378 lines, demonstrating a commitment to AI integration across all project types, even in death.

├──tools/scaffold/main.goThe +504 lines of core scaffolding logic, which itself became a fixed, unmoving structure.

└──tools/scaffold/go.modDeclares Go 1.26.0 and Cobra v1.10.1, the minimal dependencies for maximum ambition.

Eulogy Stats

Total Commits: 5
Ambitious Adjectives: 9
Deploy Config: No
Estimated Users: 0 (excluding jose.castillo's local machine)

Last Words

“The final words, 'fix(scaffold): resolve templates from executable when run outside repo', suggest a last-ditch effort to make the skeleton runnable, perhaps outside its intended grave.”