From tomer-skills
Spec-driven development (SDD) workflow for AI-assisted coding. Use this skill whenever the user wants to plan a feature before implementing it, create a spec for a change, generate a proposal + design + tasks structure, review requirements before coding, or use commands like /opsx-propose, /opsx-new, /opsx-ff, /opsx-apply, /opsx-archive, /opsx-explore, /opsx-verify, or /opsx-onboard. Also trigger when the user says "let's plan this properly", "write a spec for this", "create a proposal", "what should we build", or starts a feature discussion before any code exists. This skill installs OpenSpec slash commands into the project and teaches Claude how to execute them.
How this skill is triggered — by the user, by Claude, or both
Slash command
/tomer-skills:openspecThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
OpenSpec adds a lightweight spec layer so you and Claude agree on *what* to
README.mdcommands/opsx-apply.mdcommands/opsx-archive.mdcommands/opsx-bulk-archive.mdcommands/opsx-continue.mdcommands/opsx-explore.mdcommands/opsx-ff.mdcommands/opsx-init.mdcommands/opsx-new.mdcommands/opsx-onboard.mdcommands/opsx-propose.mdcommands/opsx-status.mdcommands/opsx-verify.mdreferences/artifact-templates.mdreferences/spec-format.mdOpenSpec adds a lightweight spec layer so you and Claude agree on what to build before any code is written. Plans live in the repo alongside your code — context that persists across sessions.
→ fluid not rigid — update any artifact anytime
→ iterative not waterfall — plan enough, then start building
→ brownfield-first — works on existing codebases
→ specs in the repo — context that survives session resets
openspec/
├── specs/ # Living capability specs
│ └── <capability>/
│ └── spec.md
└── changes/ # Active and archived changes
├── <change-id>/ # Active change
│ ├── proposal.md
│ ├── design.md
│ ├── tasks.md
│ └── specs/ # Delta specs for this change
│ └── <capability>/
│ └── spec.md
└── archive/ # Completed changes
└── <date>-<change-id>/
Read the corresponding file in commands/ for the full instructions of each
command. Summary:
| Command | Purpose |
|---|---|
/opsx-propose <id> | One-shot: create change + all planning artifacts |
/opsx-new <id> | Create a new change scaffold (no artifacts yet) |
/opsx-ff | Fast-forward: generate all remaining artifacts |
/opsx-continue | Generate the next artifact (incremental) |
/opsx-explore | Think through ideas before committing to a change |
/opsx-apply [id] | Implement tasks from tasks.md |
/opsx-verify [id] | Validate implementation against spec |
/opsx-archive [id] | Merge specs, archive the change |
/opsx-bulk-archive | Archive all completed changes |
/opsx-onboard | Interactive tutorial using your codebase |
/opsx-status [id] | Show artifact completion status |
Specs use SHALL/GIVEN/WHEN/THEN language. Read references/spec-format.md for
the full format guide and templates.
If a user invokes an /opsx-* command and the openspec/ directory doesn't
exist, run the setup procedure:
openspec/specs/ and openspec/changes/ directoriesopenspec/README.md explaining the structurecommands/ into .claude/commands//opsx-propose <feature-name>"commands/<command>.mdreferences/spec-format.mdreferences/artifact-templates.mdCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub thamam/a2x-marketplace