跳到正文
cccskills
免费在 GitHub 公开

bdd

Behavior-first feature development — use when building new capabilities, continuing feature work, or when work introduces new state or multiple user flows. Discovers desired behavior through examples and scenarios before implementation. Do NOT use for bug fixes, typos, or small isolated changes.

中文简介正在准备中。当前显示原文说明。

查看安装方法

包含的文件(7)

  • SKILL.md5.2 KB
  • DISCOVERY.md13.5 KB
  • DONE.md555 B
  • SCENARIOS.md7.4 KB
  • SPLITTING.md1.8 KB
  • TDD.md3.5 KB
  • VERIFY.md1.2 KB

SKILL.md(原文)

安装前可以查看交给 AI 代理的指令内容。

BDD Orchestrator

Behavior-first development for features. Discovery → Scenarios → Implementation.

Iron Law: DEFINE BEHAVIOR BEFORE IMPLEMENTATION

Phase Tracking

Features progress through phases. Track in ticket frontmatter:

---
type: feature
phase: implement # intake | define-behavior | scenario-gate | implement | verify | done
---

Phase meanings:

PhaseWhat happensDetails
intakeContext check, discoveryDISCOVERY.md
define-behaviorWriting Given/When/ThenSCENARIOS.md
scenario-gateValidating scenariosSCENARIOS.md
implementOutside-in TDDTDD.md
verifyEvidence gate: /verify + /auditVERIFY.md
doneClose ticketDONE.md

Update phase when:

  • Completing a BDD phase → set next phase
  • Scenario-gate complete → set implement (test-layer + sequencing happens at the scenario-gate exit)
  • Handing off to TDD → set implement
  • All scenarios pass → set verify
  • /verify + /audit complete (verify.md exists) → set done

Phase-exit review (Tier 2)

Leaving a phase is gated on an independent review of that phase's work — not your own pass. (Your own inline pass is Tier 1: /self-review, per asset, as you author.) The phase exit is reviewed by a fresh reviewer with no conversation history so the author can't grade their own work: run it in a forked subagent — a skill with context: fork, or an explicit subagent — hand it only the phase's artifacts and the ticket's scope, and let its verdict decide. On a pass, record the stamp that unblocks the advance:

bun .safeword/hooks/write-review-stamp.ts --phase <phase you are leaving>

If the reviewer finds blocking issues, fix them and re-review — don't stamp. To skip a trivial or docs-only phase, append a reason (… --phase <phase> "<why no independent review is needed>"). The phase-advance gate enforces this when the review gate is enabled, and is inert otherwise.


Resume Logic

When user references a ticket, resume work:

  1. Read ticket → get current phase:
  2. Find progress → first unchecked [ ] in test-definitions
  3. Check context → read last work log entry
  4. Announce resume → "Resuming at [phase]. Last: [log entry]."

Resume by phase:

PhaseResume action
intakeStart understanding (propose-and-converge)
define-behaviorContinue drafting scenarios
scenario-gateContinue validating scenarios
implementFind first unchecked scenario, run TDD
verifyRun /verify and /audit, write verify.md
doneClose ticket (verify.md must exist)

Current Behavior

  1. Understand first (see SAFEWORD.md "Understanding") — propose-and-converge until user accepts proposal with structured scope
  2. Size internally (see SAFEWORD.md "Sizing") — state scope assessment in proposal, not as a separate announcement
  3. If user references iteration/story/phase from a spec:
    • Check if child ticket exists for that iteration
    • If not → create ticket, run full BDD
    • If yes → resume at current phase
  4. If ticket exists: Read phase, resume at appropriate point
  5. Artifact-first rule: Before doing work, create/verify the phase artifact:
    • intake → ticket at .safeword-project/tickets/{id}-{slug}/ticket.md
    • define-behavior → test-definitions at .safeword-project/tickets/{id}-{slug}/test-definitions.md
  6. Execute phase using the appropriate phase file
  7. Update phase in ticket when transitioning

Phase Files

Load the appropriate file based on current phase:

PhaseFile
intakeDISCOVERY.md
define-behaviorSCENARIOS.md
scenario-gateSCENARIOS.md
implementTDD.md
verifyVERIFY.md
doneDONE.md

For splitting large features, see SPLITTING.md.


Key Takeaways

  • patch/task → TDD directly (RED → GREEN → REFACTOR)
  • feature → full BDD flow, track in ticket phase: field
  • Resume → read ticket, find first unchecked scenario, continue
  • Split → check thresholds at Entry, define-behavior, scenario-gate; user decides (see SPLITTING.md)
  • Verify gate → run /verify + /audit, writes verify.md. Stop hook blocks done without it.
  • Done → close ticket (trivial — verify.md must already exist)
  • When unsure → default to task, user can /bdd to override

评价

还没有评价。欢迎分享你的使用感受。

评论

无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。

还没有评论。来写第一条吧。

提交后会立即公开。请勿填写邮箱、电话等个人信息。URL 最多 2 个。

0 / 1000 字

此表单受 reCAPTCHA 保护,适用 Google 的隐私权政策和服务条款。

同一仓库的技能

audit

免费

Run comprehensive code audit for architecture, dead code, and test quality. Use when reviewing overall codebase health, checking for architectural violations, or before marking a feature complete.

中文简介正在准备中。当前显示原文说明。

TheMostlyGreat/mythos192026年6月14日 更新

Use when the user wants to explore options, weigh approaches, or think through uncertainty before committing to a direction. Collaborative brainstorming and rubber ducking — divergence-first thinking partner.

中文简介正在准备中。当前显示原文说明。

TheMostlyGreat/mythos192026年6月14日 更新

Kill zombie dev servers and test processes. Use when ports are blocked, processes are hanging, or test runners won't start.

中文简介正在准备中。当前显示原文说明。

TheMostlyGreat/mythos192026年6月14日 更新

debug

免费

Root cause debugging before fixes. Use when investigating bugs, diagnosing test failures, troubleshooting unexpected behavior, or when previous fix attempts failed. Enforces investigate-first discipline.

中文简介正在准备中。当前显示原文说明。

TheMostlyGreat/mythos192026年6月14日 更新

elicit

免费

Extract tacit knowledge through non-obvious microquestions — things only the user knows that can't be found in code, docs, or research. Use when you're about to guess at intent, context, or constraints during SAFEWORD's understanding flow. Also use when user says 'ask me', 'what do you need to know', or when another skill (bdd, brainstorm, debug) needs user context before proceeding. Do NOT use for questions answerable by reading the codebase or searching the web.

中文简介正在准备中。当前显示原文说明。

TheMostlyGreat/mythos192026年6月14日 更新

Explore and debate options with fresh documentation and research before committing. Use when facing a real decision with multiple plausible approaches — library/framework choice, architecture call, API or schema design, algorithm selection, or any communication / strategy call where being wrong has cost. Enumerates relevant research domains, looks up current docs and evidence-based methods, weighs options on correctness and elegance, resists bloat. Do NOT use for divergent ideation (brainstorm), extracting user intent (elicit), or reviewing already-written code (quality-review).

中文简介正在准备中。当前显示原文说明。

TheMostlyGreat/mythos192026年6月14日 更新

查看 TheMostlyGreat 的全部技能

报告此技能的问题