benjaminshoemakerCommunity·31 Agent Skills Included

ai_coding_project_base

Structured spec, plan, and verified autonomous code execution

Turns rough software ideas into verified, working code through a three-phase workflow: guided spec writing, execution planning, and autonomous task-by-task implementation. Eliminates ad-hoc prompting, untested code, and agents spinning on failures by enforcing TDD, security scans, and automatic verification after every task. Adds cross-model review, stuck detection, and auto-advancing phases so projects finish faster with human oversight only where needed.

39|5

↗

npx skills add benjaminshoemaker/ai_coding_project_base --all -g -y

Available:

Gives the agent durable workflow rules for this toolkit: TDD guardrails, verification-first escalation order, plan-status conventions, and how to route work through the spec, plan, and execution skills.

All Skills in This Repository (31)

Pure Emerald Level Indicators

📦 In Repo

benjaminshoemaker

auto-verify

Automate task criteria verification across HTTP, browser, and file checks.

verify-task

Verify task acceptance criteria from EXECUTION_PLAN.md with per-criterion checks.

codex-consult

Review documents with OpenAI Codex and compare model critiques.

list-todos

Analyze TODOS.md items and output a ranked list with implementation guidance.

feature-technical-spec

Generate a FEATURE_TECHNICAL_SPEC.md from a feature SPEC and project context.

analyze-sessions

Analyze Claude Code session logs to identify automation opportunities.

code-verification

Coordinate multi-agent code verification against requirements with retry and audit logs.

configure-verification

Automate configuration of verification commands into .claude/verification-config.json.

update-docs

Analyze commit ranges or working tree changes to update documentation files.

phase-checkpoint

Execute and validate Phase Checkpoint criteria from EXECUTION_PLAN.md.

progress

Analyze repository plans and compute completion metrics from Markdown checkboxes.

setup

Initialize projects with AI Coding Toolkit components and metadata.

Community

Advanced

Frequently Asked Questions

FAQPage Schema

How to install ai_coding_project_base?▼

Run `npx skills add benjaminshoemaker/ai_coding_project_base --all -g -y` in your terminal to install all skills globally.

How does this toolkit build software autonomously?▼

It guides you through a product spec and technical spec, generates a phased execution plan, then works task-by-task with automatic test, lint, and security verification after each task.

Does it work with Claude Code and Codex?▼

Yes. Skills follow the standard SKILL.md format and run in Claude Code, with optional cross-model review and task delegation through the OpenAI Codex CLI.

What happens when the agent gets stuck?▼

Built-in stuck detection tracks repeated failures and escalates to you with a clear report and options instead of retrying the same failing approach endlessly.

Can I use it for existing projects, not just new ones?▼

Yes. A dedicated feature workflow (`/feature-spec`, `/feature-technical-spec`, `/feature-plan`) adds new features to existing codebases with the same verification gates.