spec-pipeline

Discovers unimplemented feature specs and orchestrates review, correction, and implementation workflow.

Updated Jul 9, 2026
One-click install
npx skills add https://github.com/DimonikRV/ghost-pilot-project --skill spec-pipeline-dimonikrv
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: spec-pipeline
Source: https://github.com/DimonikRV/ghost-pilot-project/tree/main/.agents/skills/spec-pipeline
Command: npx skills add https://github.com/DimonikRV/ghost-pilot-project --skill spec-pipeline-dimonikrv

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Feature specs often drift out of sync with the actual codebase, leading to implementations based on stale or incorrect assumptions. This Skill enforces a disciplined pipeline that finds the next unimplemented spec, reviews it against codebase reality, fixes it, and only then guides implementation. ## Core Features & Use Cases - Spec Discovery: Scans context/feature-specs/ for numbered spec files and cross-references context/progress-tracker.md to identify the next unimplemented feature. - Pre-Implementation Review: Reviews only the target spec against the codebase and relevant skill best practices, requiring all Critical and High findings to be resolved in the spec file before coding begins. - Guided Implementation: After explicit user approval, follows a structured workflow of reading context files, marking progress, implementing exactly as specified, and verifying against the spec's checklist. - Use Case: A developer asks "what's next" in a spec-driven project; the pipeline identifies spec 04, finds its proposed schema conflicts with the actual Prisma models, updates the spec, gets approval, and implements it correctly. ## Quick Start Ask the assistant to pick up the next feature spec and run it through the review and implementation pipeline.

Frequently Asked Questions about spec-pipeline

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I implement the next feature spec in a spec-driven project?▼

Read the progress tracker to find completed specs, then select the lowest-numbered unimplemented spec from the feature-specs directory. Review it against the codebase, fix any findings in the spec file, get user approval, then implement exactly as specified.

How to review a feature spec against the actual codebase before coding?▼

Read the spec file and only the codebase files it touches, then check the proposed approach against relevant best-practice documentation. Resolve Critical and High severity findings by updating the spec file before any implementation begins.

Can I review multiple feature specs at once?▼

No. The pipeline reviews only the single spec assigned as the immediate next implementation target. Batch-reviewing multiple specs or reviewing specs out of order is explicitly prohibited to keep reviews focused and actionable.

What happens if a spec review finds critical issues?▼

Critical and High severity findings are blocking and must be resolved by updating the spec file before implementation. If a Critical issue cannot be resolved by correcting the spec, implementation stops and the problem is reported to the user.

What if the progress tracker file is missing or malformed?▼

The pipeline warns the user and attempts recovery by scanning the progress tracker for completed specs. If no unimplemented specs exist at all, it prompts the user to create a new spec instead of proceeding.