From python-engineering
Orchestrates multi-step Python engineering workflows from task descriptions by routing to SAM or direct tracks for planning, phased execution, quality gates, and commits.
npx claudepluginhub jamie-bitflight/claude_skills --plugin python-engineeringThis skill uses the workspace's default tool permissions.
Multi-step engineering workflow command.
Generates design tokens/docs from CSS/Tailwind/styled-components codebases, audits visual consistency across 10 dimensions, detects AI slop in UI.
Records polished WebM UI demo videos of web apps using Playwright with cursor overlay, natural pacing, and three-phase scripting. Activates for demo, walkthrough, screen recording, or tutorial requests.
Delivers idiomatic Kotlin patterns for null safety, immutability, sealed classes, coroutines, Flows, extensions, DSL builders, and Gradle DSL. Use when writing, reviewing, refactoring, or designing Kotlin code.
Multi-step engineering workflow command.
Task: $ARGUMENTS
If no argument is supplied, derive the task from the active conversation.
Read ../python3-core/references/python-development-orchestration.md.
Do not proceed to Step 2 until this file has been read. It contains agent selection criteria, workflow patterns, quality gates, and multi-agent chaining patterns needed for Step 2.
flowchart TD
Q{"Does the task meet ANY of:<br>- user said 'add a feature', 'plan', or 'track'<br>- requires 2+ agents in sequence<br>- spans multiple files or modules<br>- needs durable progress tracking across turns"}
Q -->|"Yes"| SAM["SAM Track — Step 3A"]
Q -->|"No — single focused task:<br>fix a bug, write tests for one file,<br>review code, one-shot refactor"| Direct["Direct Track — Step 3B"]
Then state aloud before the first Agent tool call:
Task: <one sentence>
Track: SAM | Direct
Workflow pattern: <TDD | Feature Addition | Refactoring | Debugging | Code Review>
Agent chain: <AGENT1> → <AGENT2> → ...
If you cannot fill in workflow pattern and agent chain from the guide read in Step 1, go back and read it.
flowchart TD
P1["Phase 1 — Plan<br>Skill: /dh:add-new-feature<br>Args: task description<br>Produces: ~/.dh/projects/{slug}/plan/P{NNN}-{slug}.yaml"]
P1 --> P1Q{"add-new-feature result?"}
P1Q -->|"BLOCKED — plan-validator gate failed"| P1Blocked["Surface blocker to user<br>Await clarification<br>STOP"]
P1Q -->|"PASS — task file produced"| P2
P2["Phase 2 — Execute<br>Skill: /dh:implement-feature<br>Args: path to task file<br>Loop: sam ready → start-task → SubagentStop hook marks COMPLETE<br>Repeat until no tasks remain"]
P2 --> P3["Phase 3 — Quality gates<br>Auto-invoked by implement-feature<br>Skill: /dh:complete-implementation<br>Runs: code review → feature verification → integration check<br>→ doc drift → doc update → context refinement → commit"]
P3 --> Done(["DONE — changes committed"])
When mcp__plugin_dh_sam__sam_create is called directly (e.g., from create-feature-task):
title: "<short imperative title>"
description: |
<what must be true when this task is done>
acceptance_criteria:
- Given <context>, when <action>, then <observable result>
phases:
- name: <phase name>
tasks:
- <concrete subtask>
Update task status with mcp__plugin_dh_sam__sam_update as phases complete.
Classify the task then delegate:
Agent routing — delegate rather than implement:
Each delegation must include:
Do NOT read source files before delegating. Agents search and read files themselves — pass file paths, not file contents.
python3-coreBefore reporting done:
uv run prek run --files <modified_files> — runs linting, formatting, and type checking
Fallback: uv run ruff format and uv run ruff check --fix only when no .pre-commit-config.yamluv run pytest — all pass, coverage ≥80%