dev-pipeline_skill

This skill automates standardized dev pipelines by integrating version-manager and project-manager, guiding users from init to seal with actionable steps.

Python

2.6k

GitHub Stars

2

Bundled Files

2 months ago

Catalog Refreshed

3 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill openclaw/skills --skill dev-pipeline

_meta.json286 B
SKILL.md14.5 KB

Overview

This skill implements a versioned development pipeline tightly integrated with version-manager and project-manager to provide a standardized, auditable development flow. It orchestrates analysis, code generation, review, deploy and archival steps, with optional sub-agent execution for heavy tasks. The goal is predictable releases, clear task outputs, and safe archiveable versions.

How this skill works

The pipeline coordinates commands (init, analyze, confirm, write, review, fix, deploy, seal) and delegates actions to version-manager and project-manager for checks, preparation, updates and archiving. For heavy work (analysis, code generation, review, fix) it can spawn sub-agents that call an oracle model (Claude Opus) and return structured JSON results. State is tracked in .state.json and project board is synchronized via project-update and project-changelog.

When to use it

Start a new versioned development cycle (init)
Generate a machine-readable development plan from requirements (analyze)
Automate large code generation or review tasks using isolated sub-agents
Enforce safety checks before production deployment (deploy)
Create immutable archives of releases for compliance (seal)

Best practices

Use init to enforce version-prepare and a known base version
Run analyze and write inside sub-agents to avoid main-session token bloat
Always confirm DEV_PLAN.md before entering write phase
Require review_passed state before deploy; run version-validate prior to production
Keep .dev-pipeline/config.json correct and ensure oracle credentials are available to sub-agents

Example use cases

Initialize v1.3.5, run architecture analysis in a sub-agent, confirm plan, and generate code artifacts
Assign a long running write task to a skill-runner sub-agent to produce multiple files and return a files_created list
Run automated code review via a sub-agent that outputs a structured review report and review_score
Deploy only after version-validate and archive the release with version-archive and project-changelog updates
Run iterative fix cycles (fix → review) until review_passed then seal and archive the release

FAQ

sessions_spawn runs heavy tasks (analyze, write, review, fix) in ephemeral or persistent sub-agents so the main session stays lightweight. Use it for operations that produce large outputs or call the oracle model.

How is state tracked across the pipeline?

Pipeline state is stored in .state.json (version, status, current_task, tasks, etc.). Each command updates status and project-manager is called to keep the project board in sync.