agentsclimarketplace

Orchestrate sw dev en

Skill roebi/sw-dev-agent-framework/skills/orchestrate-sw-dev-en

An agentic framework that does exactly one thing: **software development**. Built with TDD. Every line of production code was preceded by a failing test.

Install
npx -y skills add roebi/sw-dev-agent-framework --skill orchestrate-sw-dev-en

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Orchestrates the full software development lifecycle as an agentic pipeline. Delegates each phase to a focused sub-skill in order: gather requirements, design architecture, implement with TDD, review code, and release. Use when a user or agent wants to build a software project from a goal statement to a released artifact using a structured, test-driven process. Activate for trigger phrases like: "build a software project", "develop this from scratch", "orchestrate sw development", "run the full dev pipeline", or "implement this end to end".

The file declares its own license as CC BY-NC-SA 4.0. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

2.9 KB, as published. Nobody here has run it

Orchestrate SW Dev

Runs the full software development lifecycle as a sequential agentic loop. Each phase delegates to a dedicated sub-skill. State is tracked in Handover-State.md so the pipeline can be resumed across sessions.

Lifecycle Phases

REQUIREMENTS -> DESIGN -> IMPLEMENT -> REVIEW -> RELEASE
PhaseSkillOutput
1. Requirementsgather-requirements-enrequirements.md
2. Designdesign-architecture-enarchitecture.md
3. Implementapply-tdd-loop-enpassing tests + code
4. Reviewcode-review-enreview checklist + fixes
5. Releaserelease-sw-project-entagged release + changelog

Handover State

Track progress across sessions in Handover-State.md:

# Handover-State.md
skill: orchestrate-sw-dev-en
goal: <one-sentence goal>
current_phase: requirements | design | implement | review | release
results:
  - phase: requirements
    status: done | pending | in_progress | blocked
    output: <brief summary or file reference>

Orchestration Loop

LOOP:
  1. Read current_phase from Handover-State.md (or start at requirements)
  2. Load and run the sub-skill for that phase
  3. When the sub-skill completes, record its output in results[]
  4. Advance current_phase to the next phase
  5. Write updated Handover-State.md
  6. If current_phase has no next -> EXIT, project is released

Entry Point

Input: a single goal statement, e.g.:

"Build a Python CLI tool that converts CSV files to Markdown tables."

Pass this goal to gather-requirements-en as the first phase input.

Phase Gate Rules

  • A phase must be status: done before the next phase starts.
  • If a phase returns status: blocked, stop and report the blocker to the user.
  • Never skip a phase. IMPLEMENT always uses TDD via apply-tdd-loop-en.
  • REVIEW must run even if the developer wrote the code - no self-merge rule.

Session Resume

If a session is interrupted, resume by:

1. Load Handover-State.md
2. Find the first phase where status != done
3. Re-run that phase's sub-skill with the existing context

References

  • references/phase-contracts.md - input/output contracts per phase

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.