workflow

📁 bntvllnt/agent-skills 📅 Jan 29, 2026

总安装量

周安装量

#13407

全站排名

安装命令

npx skills add https://github.com/bntvllnt/agent-skills --skill workflow

Agent 安装分布

claude-code 10

opencode 9

codex 5

amp 4

github-copilot 4

Skill 文档

Workflow

High-velocity solo development. Idea to production same-day.

Agent Capabilities

Capability	Used For	Required	Fallback
File read/write	Specs, config, history	Yes	â
Code search (grep/glob)	Discovery, context	Yes	â
Shell/command execution	Quality gates (lint, build, test)	Yes	List commands for user to run
Task/todo tracking	Phase management	Recommended	Track in spec Progress section
User interaction	Stuck escalation, risk flags	Recommended	Log decisions in spec Notes
Web/doc search	Pattern lookup	No	Use embedded patterns

Fallback rule: If your agent lacks a capability, use the fallback. Never skip the workflow step â adapt the method.

Commands

Command	Action	Reference
`plan {idea}`	Create spec	plan.md
`spike {question}`	Time-boxed exploration	spike.md
`ship` / `ship {idea}`	Implement + validate	ship.md
`review`	Multi-perspective code review	review.md
`spec-review`	Adversarial spec analysis	spec-review.md
`focus`	Priority analysis + task proposals	focus.md
`done`	Validate + retro + archive	done.md
`drop`	Abandon, preserve learnings	drop.md
`workflow`	Show state + suggest next	Status (below)

No flags needed. The agent auto-detects intent from context:

“review the spec” â manual review pause
“skip tests” â skip test gate (documented)
“emergency fix” â bypass spec ceremony
“production ready” â production validation

Flow

focus â plan {idea} â ship â [implement/review/fix loop] â done

Quick mode (<2h): ship {idea} â done Don’t know what to work on: focus

Philosophy

Spec-first: All work needs a spec (creates one if missing)
Ship loop: Build â review â fix until clean
Quality gates: lint â typecheck â build â test (auto-detected per project)
Human controls deployment: Agent codes, you push/deploy
Done same-day: Scope to what ships today
Own planning: Never use the host agent’s built-in plan mode (EnterPlanMode, etc.). This skill writes real spec files to specs/active/.

Spec Tiers

Tier	Size	Spec	Task Tracking
trivial	<5 LOC	None â just do it	No
micro	<30 LOC	Inline comment in code	No
mini	<100 LOC	Spec file, minimal	Yes (if available)
standard	100+ LOC	Full spec with checklist	Yes (if available)

Action Router

User input
  â
  ââ "plan", "spec", "design"           â Load references/actions/plan.md
  ââ "spike", "explore", "investigate"   â Load references/actions/spike.md
  ââ "ship", "implement", "fix", "build" â Load references/actions/ship.md
  ââ "review", "check code"              â Load references/actions/review.md
  ââ "review spec", "analyze spec",
  â  "challenge spec"                    â Load references/actions/spec-review.md
  ââ "focus", "what should i do",
  â  "prioritize", "overwhelmed"         â Load references/actions/focus.md
  ââ "done", "finish", "complete"        â Load references/actions/done.md
  ââ "drop", "abandon"                   â Load references/actions/drop.md
  ââ "workflow", "what's next", "what now",
     "what's up", "whats up", "status"  â Status Action (below)

Loading rule: Read the action file BEFORE executing. The action file contains all logic, task templates, and references needed.

Status Action

No separate action file â logic is inline here. Detect current state, suggest next action:

1. Check specs/active/ for active spec
2. Check git status for uncommitted work
3. Check task list for in-progress items

State â Suggestion:
  No spec, no changes    â "Ready. Run: plan {idea}"
  Active spec, no code   â "Spec ready. Run: ship"
  Active spec, code WIP  â "In progress. Run: ship (resumes)"
  Active spec, code done â "Ready to close. Run: done"
  No spec, dirty tree    â "Uncommitted work. Run: ship (creates spec) or done"

Output: Follow status-output.md.

Project Structure

specs/
  active/       â Current work (0-1 specs)
  backlog/      â Queued work from focus
  shipped/      â Completed features
  dropped/      â Abandoned with learnings
  history.log   â One-line per feature shipped/dropped

Configuration

All behavior is configurable by editing the skill files directly.

What to change	Edit
Action logic, gates, limits	`references/actions/{action}.md`
Output format	`references/templates/{action}-output.md`
Spec structure	`references/spec-template.md`
Quality gate commands/levels	`references/quality-gates.md`
Session resume, stuck detection	`references/session-management.md`

References

Actions:

Plan | Ship | Review | Spec Review | Focus | Done | Drop | Spike

Output templates:

Plan + Spec Review | Ship | Review | Focus | Done | Drop | Spike | Status

Review standards:

Production Standards

Specs & gates:

Spec template | Quality gates | Session management | Memory update | Testing automation

Patterns:

Implementation | Planning | Debugging | Decisions | Decomposition | Regression testing

GitHub 仓库 ↗ ← 返回陌讯 Skills 聚合平台