claude

graphstack

Graph-engineering pipeline (Think -> Plan -> Build -> Review -> Test -> Ship -> Reflect) powered by logic ported from gstack, over a local task backlog and append-only review/evidence ledgers.

0 on GitHubby setov0.1.0updated 1 month agodevelopment
/

Plugin graphstack from marketplace ph-agent-plugins in p-hubbell/ph-agent-plugins. Commands read from the repo's manifest — 1 month ago.

What it installs

Skills (7)

  • land-and-deploy

    Optional stage after Ship — merges the open PR, waits for CI, waits for the deploy, runs a lightweight post-deploy check, and offers a revert if something looks broken. Human-checkpoint-heavy by design (merging and deploying are semi-irreversible). Invoke explicitly after /graphstack:ship when the u

  • pipeline

    Runs the full graphstack pipeline end to end — Think -> Plan -> Build -> Review -> Test -> Ship -> Reflect — from a single invocation. Use when the user hands over an idea, a spec file/folder, or just says "build this" and wants the whole loop driven for them, rather than running one stage at a time

  • plan

    Use when the user wants a graphstack brief (or any raw idea/spec, if Think hasn't run yet) turned into a scoped plan and a task backlog before implementation starts. Second stage of the graphstack pipeline — reads `_docs/graphstack/brief.md`, sequences a strategist pass then an architect pass, and w

  • reflect

    Seventh and final stage of graphstack, run after Ship (or at the end of a work session) — analyzes what happened this pipeline run and writes durable learnings that feed back into the next Think invocation, closing the loop.

  • review

    Use after Build to independently review an implemented task's diff for structural issues Test won't catch — SQL safety, race conditions, LLM trust boundaries, and other checklist/specialist findings. Fourth stage of the graphstack pipeline; run on a task with status `implemented`, before /graphstack

  • ship

    Use once one or more tasks reach status `tested` and are ready to leave the branch — gate-checks the review/test ledgers for each task, syncs the base branch, bumps the version, updates the CHANGELOG, writes a PR body, commits, pushes, and opens a PR via `gh`. Sixth stage of the graphstack pipeline;

  • think

    Use when the user has flexible input for something to build — a vague idea, a one-liner, a pasted spec, a markdown file, or a folder of docs — and wants it turned into a clear problem brief before planning starts. First stage of the graphstack pipeline; run before /graphstack:plan.

Subagents (7)

  • architect

    Use to lock in the architecture for a graphstack plan once scope is settled — data flow, edge cases, failure modes, and the test matrix. Runs the architecture pass of the Plan stage, after the strategist agent. Never writes code.

  • facilitator

    Use to run forcing-questions discovery on a vague idea from the graphstack Think stage, turning a one-liner into a shape clear enough to write a problem brief. Asks one question at a time, pushes for specificity over comfort. Never writes code.

  • groomer

    Use to groom a raw backlog task from the graphstack workflow into a precise, checkable specification (Goal, Acceptance Criteria, Out of Scope, Constraints) before an engineer implements it. Never writes code.

  • qa-engineer

    Use to validate a reviewed backlog task from the graphstack workflow against its Acceptance Criteria and produce a PASS/FAIL verdict, with findings classified by severity and category. Never fixes code, never writes to review-log.jsonl or evidence.jsonl.

  • retro-analyst

    Use to analyze git commit history plus the graphstack review-log.jsonl/evidence.jsonl ledgers for the Reflect stage — commit counts, shipping streak, test-health trend, review/test pass-fail patterns. Read-only analysis; reports findings back to the calling skill and never writes learnings.md, sessi

  • software-engineer

    Use to implement a groomed backlog task from the graphstack workflow against its Goal, Acceptance Criteria, and Constraints, including tests. Re-invoke with Review Log or QA Log feedback appended to the task file when a task previously failed Review or Test.

  • strategist

    Use to judge whether the scope of a graphstack brief or plan input is right — expand it, hold it, or cut it — before architecture gets locked in. Runs the scope-framing pass of the Plan stage, ahead of the architect agent. Never writes code.

View source on GitHub

More from p-hubbell/ph-agent-plugins

All 2 plugins →

the-heist

claude

The Heist: Crew of Thieves — Mastermind (think), Forger (UI design), Safecracker (implement), with Lookout audits and Cleaner code review.

3 skills · 5 agents

0

Similar plugins

superpowers

claudecodex

Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques

15 skills · hooks

obra/superpowers 316,836

context7

claudecodex

Up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your coding agent's context.

1 MCP server

upstash/context7 64,709

Build powerful, stateful CLI interfaces for any GUI application using the cli-anything harness methodology.

5 commands

HKUDS/CLI-Anything 58,665

Automates browser interactions for web testing, form filling, screenshots, and data extraction

12 skills

vercel-labs/agent-browser 50,373

Use Alibaba Open Code Review from local Codex.

2 skills

alibaba/open-code-review 50,068

slidev

claude

Claude Code skill for creating and presenting Slidev slidedecks

1 skill

slidevjs/slidev 49,452

Maintain graphstack?

This page updates itself from your repo. Add the badge to your README, or get it Featured to pin it above the organic results with a followed link to your site.

claude & codex plugins2 plugins