STOP - Read before ANY tool call
Read _shared/skill-boilerplate.md first. It holds the guardrails read, the session gate rule, the user sim rule, the one-step rule, the gate check, and the user chat language rule. The steps below are skill-specific only.
Skill: python-developer
Trigger
Negative signal: applies as defined in _shared/skill-boilerplate.md; the skill-specific signals are listed below.
Invoke when the user asks for: /python-developer, python fix, implement FastAPI/Flask feature, or for small Python work that does not need a full PRD/PLAN cycle.
Outcome
Working CPython host code and tests in the open workspace: the repository's configured test runner (pytest only as a suitable fallback without a baseline) and project lint/type checks when configured, on a valid feature branch, with optional commit handoff. Supports FastAPI and Flask — not a FastAPI-only skill. Does not replace SDD for multi-step or cross-repo features.
Host boundary: use this skill for CPython applications, host tools, and host tests, including tools that communicate with a board. Do not use it to recommend MicroPython device APIs, device-only modules, REPL procedures, or flashing/deployment steps.
Lazy-load (only when needed)
| When | Path (after syncing the active adapter) |
|---|---|
| Repo context | {{TOOLKIT_ROOT}}/skills/_shared/developer-common/step-0-context.md |
| Before coding | {{TOOLKIT_ROOT}}/skills/_shared/developer-common/step-0.5-review-guidelines.md |
| Structure / quality (when coding) | {{TOOLKIT_ROOT}}/skills/_shared/code-guidelines/principles/structure-and-quality.md |
| Branching | {{TOOLKIT_ROOT}}/rules/branch-validation.mdc, {{TOOLKIT_ROOT}}/skills/_shared/developer-common/step-3-branching.md |
| Pre-commit | {{TOOLKIT_ROOT}}/skills/_shared/developer-common/step-3.5-precommit-validation.md |
| Commit / PR | {{TOOLKIT_ROOT}}/skills/_shared/developer-common/step-4-commits-pr.md, {{TOOLKIT_ROOT}}/rules/conventional-commits.mdc |
| Pre-PR gate | {{TOOLKIT_ROOT}}/skills/_shared/developer-common/step-7-checklist.md |
| FastAPI | {{TOOLKIT_ROOT}}/skills/_shared/python-guidelines/fastapi.md |
| Flask | {{TOOLKIT_ROOT}}/skills/_shared/python-guidelines/flask.md |
| Architecture (hub / structure) | {{TOOLKIT_ROOT}}/skills/_shared/python-guidelines/architecture.md |
| Architecture (greenfield / style unset) | {{TOOLKIT_ROOT}}/skills/_shared/code-guidelines/principles/architecture-selection.md — then one approved style only |
| Architecture style (when ARCH names a style) | one of principles/architecture/concentric-dependency.md | vertical-slice.md | ddd-tactical.md | event-driven.md (never glob) |
| pytest | {{TOOLKIT_ROOT}}/skills/_shared/python-guidelines/pytest.md |
| Packaging / pyproject | {{TOOLKIT_ROOT}}/skills/_shared/python-guidelines/packaging-pyproject.md |
| Typing | {{TOOLKIT_ROOT}}/skills/_shared/python-guidelines/typing.md |
| Async pitfalls | {{TOOLKIT_ROOT}}/skills/_shared/python-guidelines/async-pitfalls.md |
| Style / principles | {{TOOLKIT_ROOT}}/skills/_shared/python-guidelines/google-style.md, {{TOOLKIT_ROOT}}/skills/_shared/python-guidelines/principles.md |
| Subagent-first / SPAWN.md | {{TOOLKIT_ROOT}}/skills/_shared/developer-common/subagent-first.md, {{TOOLKIT_ROOT}}/skills/_shared/agents/SPAWN.md |
| Caveman Mode (if active) | {{TOOLKIT_ROOT}}/skills/_shared/caveman/CAVEMAN.md - Full cap |
| Reference index (routing only) | {{TOOLKIT_ROOT}}/skills/python-developer/reference.md |
| Process step detail (lazy) | {{TOOLKIT_ROOT}}/skills/python-developer/references/<section>.md |
Do not preload other stack guideline packs or corporate pipeline docs. Load only the python-guidelines rows needed for the current task (FastAPI and/or Flask and pytest as applicable). MUST NOT glob architecture/** — load hub + one principles style from ARCH/CONTINUITY (brownfield: discover-first if style omitted).
Never by default: do not preload all references/*.md. do not preload other stack guideline packs. Load only Python rows needed for the current task. Do not dump full stack guideline packs or memory-bank. Load one section per Process step (SKILL-REFERENCE-RETRIEVAL.md).
Reference routing
| Situation | Path |
|---|---|
| Scope / SDD escalation / design brief | references/scope.md |
| Implement flow | references/execute-flow.md |
| Must not (full) | references/must-not.md |
Process
Read references/<section>.md for procedural detail — not full reference.md.
Step -1b - Caveman Mode (Full cap)
- Read
{{SDD_ROOT}}/preferences.json(create{ "caveman_mode": false, "caveman_level": "full" }if missing). - If
caveman_modeis false: continue without compression. - If true: load
{{TOOLKIT_ROOT}}/skills/_shared/caveman/CAVEMAN.md; apply Full participation cap + prefscaveman_level(Lite skills never escalate); show once:[Caveman] Modo ativo (respostas compactas, level={effective}). Digite caveman off para desativar. - Honor
caveman on|off|status|lite|full|ultra(andstop caveman/normal mode) during the session. - Auto-Clarity + never-compress gates/drafts/paths per
CAVEMAN.md.
0. Scope
Follow references/scope.md (SDD escalation, design brief / host detection when present).
1. Execute
Follow references/execute-flow.md (subagent-first, workspace → guidelines → branch → implement → tests → handoff).
Must not
Enforce the full list in references/must-not.md. Critical: no guideline dumps into children; no auto-commit; lazy-load stack guidelines only — never dump full packs or memory-bank. If runtime, target, or port evidence is missing and the answer could change between CPython and MicroPython, ask for that context before recommending an API or deployment operation; never infer device runtime from .py alone.
Handoff
| Situation | Next |
|---|---|
| Commit | /commit |
| Review | /code-review |
| Large scope | /sdd-spec -> sdd-plan -> sdd-develop |
| Next PLAN step | New chat -> /sdd-develop - PLAN/... - Step N |