sdd-init
GitHub初始化SDD上下文,检测项目技术栈、测试能力及持久化模式,生成技能注册表。支持Engram、OpenSpec及混合模式,处理严格TDD配置逻辑。
Trigger Scenarios
Install
npx skills add Gentleman-Programming/gentle-ai --skill sdd-init -g -y
SKILL.md
Frontmatter
{
"name": "sdd-init",
"license": "MIT",
"metadata": {
"author": "gentleman-programming",
"version": "3.0",
"delegate_only": true
},
"description": "Trigger: sdd init, iniciar sdd, openspec init. Initialize SDD context, testing capabilities, registry, and persistence.",
"user-invocable": false,
"disable-model-invocation": true
}
Execution Role
Confirm your role before acting. You are the dedicated sdd-init sub-agent unless you loaded this skill directly through the skill() tool.
- If you are the
sdd-initsub-agent, continue with the phase work below. Do not delegate. Do not call the Skill tool. - If you loaded this skill through the
skill()tool, you are the orchestrator. Stop here and delegate to the dedicatedsdd-initsub-agent using your platform's delegation primitive (for example,task(...)or a sub-agent invocation).
Language Domain Contract
Generated technical artifacts default to English. Do not inherit the user's conversational language or the active persona's regional voice for SDD artifacts unless the user explicitly requests that artifact language or the project convention requires it.
If technical artifacts are explicitly requested in another language, use a neutral/professional register unless the user explicitly requests a different tone or regional variant.
Public/contextual comments follow the target context language by default. Explicit user language or tone overrides win; otherwise use a neutral/professional register unless the target context clearly calls for another tone or regional variant.
Activation Contract
Run this phase when the orchestrator/user asks to initialize SDD in a project. You are the phase executor: do the work yourself, do not delegate, and do not behave like the orchestrator.
Hard Rules
- Detect the real stack, conventions, architecture, testing tools, and persistence mode; never guess.
- In
engrammode, do not createopenspec/. - In
openspecmode, follow../_shared/openspec-convention.mdand write file artifacts. - In
hybridmode, write both openspec files and Engram observations. - Always persist testing capabilities separately as
sdd/{project}/testing-capabilitiesoropenspec/config.yamltesting:. - Always build
.atl/skill-registry.md; also saveskill-registryto Engram when available. - Use
capture_prompt: falsefor automated SDD/config saves when supported; omit it if the tool schema lacks it. - If
openspec/already exists, report what exists and ask before updating it.
Decision Gates
| Input | Action |
|---|---|
mode=engram |
Save context and capabilities to Engram only. |
mode=openspec |
Create/update openspec bootstrap files only. |
mode=hybrid |
Do both Engram and openspec persistence. |
mode=none |
Return detected context only; write no SDD artifacts except registry if required. |
explicit strict_tdd: false marker/config |
Preserve strict_tdd: false. |
explicit strict_tdd: true marker/config and an explicit workspace-level test command covers every in-scope project |
Use strict_tdd: true. |
explicit strict_tdd: true marker/config without that workspace-level command |
Fail closed to strict_tdd: false and explain that downstream execution requires a workspace-wide command. |
| no marker/config, non-empty discovered project set, and an explicit workspace-level test command covers every in-scope project | Default strict_tdd: true. |
| zero projects are discovered or no explicit workspace-level test command covers every in-scope project | Set strict_tdd: false; preserve and report every project-local command, including missing or independent commands; those local facts do not override a workspace-level command that covers every in-scope project. Explain the no-runner or workspace-wide-command fallback. |
Execution Steps
- Identify the authoritative workspace root. Before classifying a stack or applying any no-runner fallback, discover every in-scope project root from that root using the bounded rules in
references/init-details.md. - Inspect each discovered project for
package.json,go.mod,pyproject.toml,Cargo.toml, CI, and lint/test config; preserve its relative path and summarize its stack/conventions. - Detect each project's test runner and command, test layers, coverage, linter, type checker, and formatter. Aggregate those project-to-tool associations in the one workspace-level result; never select one project runner for the workspace.
- Resolve Strict TDD from an agent marker or
openspec/config.yamlonly after every discovered project has been evaluated. Set it to true only for a non-empty discovered project set when one explicit workspace-level test command covers every in-scope project. Preserve and report every project-local command, including missing or independent commands; those local facts do not override a workspace-level command that covers every in-scope project. Use the false fallback only when zero projects are discovered or no explicit workspace-level test command covers every in-scope project. - Initialize persistence for the resolved mode.
- Build
.atl/skill-registry.mdusing the skill-registry scan rules. - Persist testing capabilities and project context.
- Return the structured initialization envelope.
Output Contract
Return status, executive_summary, artifacts, next_recommended, and risks. Include project, stack, persistence mode, Strict TDD status, testing capability table, saved observation IDs/paths, registry path, and next /sdd-explore or /sdd-new step.
References
- references/init-details.md — detection checklist, Engram payloads, config skeleton, and output templates.
../_shared/engram-convention.md— Engram artifact naming.../_shared/openspec-convention.md— openspec layout and rules.
Version History
-
cc4ed69
Current 2026-08-28 11:03
修复严格TDD回退优先级逻辑,对齐工作区严格TDD交接机制,改进嵌套工作区项目的发现能力。
-
35deba3
2026-08-20 00:48
重构角色声明逻辑,消除编排器与执行器的指令冲突,增加自动化测试验证委托技能状态。
- e01b114 2026-07-25 06:59


