agentleFS
rcdelfinrcdelfin/agentkit/skills/planning/openspec/SKILL.md

openspec

Skill2 starsChanged 3 months agoMITView on GitHub

OpenSpec artifact-driven workflow. Covers OPSX commands, schemas, project config. Use when applying the artifact-driven workflow (OPSX), planning or reviewing changes based on artifact dependencies, or working with OPSX commands and schema/template resolution. Keywords: OPSX, artifact graph, /opsx:.

7 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

12 rules. 0 are common, 10 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 10

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill to guide or reason about the OpenSpec artifact-driven workflow system (OPSX), including artifact graphs, schema/template …

    line 12
  2. Very specific:

    Create openspec/config.yaml for per-project settings:

    line 65
  3. Distinctive:

    Prefer update when intent stays the same and you are refining scope or approach.

    line 93
  4. Distinctive:

    Prefer new change when intent or scope fundamentally shifts, or the original can be completed independently.

    line 94
  5. Very specific:

    Always preserve a clear history of why artifacts changed (proposal/specs/design/tasks).

    line 95
  6. Distinctive:

    Create change directory and README.

    line 116
  7. Distinctive:

    Prefer change metadata if available.

    line 127
  8. Distinctive:

    Do not treat the system as a linear workflow engine.

    line 133
  9. Distinctive:

    Do not assume a change is active without explicit selection.

    line 134
  10. Distinctive:

    Do not silently fall back between schemas or templates without reporting.

    line 135
  11. Distinctive:

    Do not copy long vendor docs verbatim.

    line 136
  12. Distinctive:

    Use explicit change identifiers in examples.

    line 141

“Common” is measured: kept by at least 6 in 10 of the 4 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

OpenSpec (OPSX) Skill

Use this skill to guide or reason about the OpenSpec artifact-driven workflow system (OPSX), including artifact graphs, schema/template resolution, change lifecycle, and experimental commands/skills.

Quick Navigation

  • Artifact graph core concepts: references/artifact-core.md
  • OPSX workflow behavior and usage: references/opsx-workflow.md
  • Setup + profiles (init, update, config profile): references/setup-profiles.md
  • Schema customization workflow and gaps: references/schema-customization.md
  • End-to-end schema workflow gaps and proposed solution: references/schema-workflow-gaps.md
  • Experimental release plan and rollout checklist: references/experimental-release-plan.md

Release Highlights

1.6.0

  • New command /opsx:onboard: guided walkthrough of a complete change cycle (explore → new → proposal → specs → design → tasks → apply → archive) using a real codebase task. Teaching overlay, not a new workflow phase.

1.2.0 → 1.3.0

  • More tool integrations: adds support for Junie, Lingma, ForgeCode, and IBM Bob.
  • Safer setup: shell completion installation is now opt-in, and Copilot auto-detection no longer triggers from a bare .github/ directory alone.
  • Adapter fixes: Pi command generation was corrected, and OpenCode now uses the canonical .opencode/commands/ path.
  • Safer status checks: openspec status now exits cleanly when a project has no changes yet.

OPSX Commands

CommandPurpose
/opsx:onboardGuided first-cycle walkthrough (teaching overlay)
/opsx:exploreThink through ideas, investigate problems (no structure)
/opsx:newStart a new change
/opsx:continueCreate next artifact based on dependencies
/opsx:ffFast-forward — create all planning artifacts at once
/opsx:applyImplement tasks, updating artifacts as needed
/opsx:verifyValidate implementation matches spec
/opsx:syncSync delta specs to main specs
/opsx:archiveArchive single completed change
/opsx:bulk-archiveArchive multiple completed changes at once

Legacy (non-OPSX) command: /openspec:proposal creates all planning artifacts at once. Prefer OPSX, but this can be useful for small/straightforward changes or older setups.

Schema Management

openspec schemas                    # List available schemas
openspec schema which --all         # Show resolution sources
openspec schema init my-workflow    # Create new schema interactively
openspec schema fork spec-driven my-workflow  # Fork existing schema
openspec schema validate my-workflow  # Validate schema structure

Project Configuration

Create openspec/config.yaml for per-project settings:

schema: spec-driven

context: |
  Tech stack: TypeScript, React, Node.js
  Testing: Vitest, Playwright

rules:
  proposal:
    - Include rollback plan
  specs:
    - Use Given/When/Then format

Schema precedence: CLI flag → Change metadata → Project config → Default (spec-driven)

Core Concepts

  • Artifact graph, not a workflow engine: Dependencies enable actions; they do not force linear phases.
  • Filesystem-as-database: Completion is derived from file existence, not stored state.
  • Deterministic CLI: Commands require explicit change context (agent infers, CLI remains strict).
  • XDG schema resolution: User overrides take precedence over built-ins.
  • Templates are schema-scoped: Templates live next to schema and resolve with a strict 2-level fallback.

Decision Rules

  • Prefer update when intent stays the same and you are refining scope or approach.
  • Prefer new change when intent or scope fundamentally shifts, or the original can be completed independently.
  • Always preserve a clear history of why artifacts changed (proposal/specs/design/tasks).

Recipes

1) Show what is ready to create next

  1. Determine the active change id.
  2. Query status and ready artifacts with the change explicitly set.
  3. Present ready artifacts and their dependencies.

Expected behavior: show ready artifacts, not required steps.

2) Generate instructions for a specific artifact

  1. Resolve schema and template with XDG fallback.
  2. Build context: change metadata, dependency status, and target paths.
  3. Return enriched instructions in plain Markdown.

3) Start a new change

  1. Validate change name (kebab-case).
  2. Create change directory and README.
  3. Show initial status and first ready artifact.

4) Schema customization guidance

  1. Explain XDG override paths.
  2. Describe copying built-in schema + templates.
  3. Provide verification steps or recommended CLI commands for listing and resolving.

5) Explain schema binding for a change

  1. Prefer change metadata if available.
  2. Fallback to project default schema if configured.
  3. Otherwise default to spec-driven.

Prohibitions

  • Do not treat the system as a linear workflow engine.
  • Do not assume a change is active without explicit selection.
  • Do not silently fall back between schemas or templates without reporting.
  • Do not copy long vendor docs verbatim; summarize and provide actionable guidance.

Output Expectations

  • Give clear artifact readiness and dependency explanations.
  • Use explicit change identifiers in examples.
  • Provide concise, actionable steps and indicate whether they are informational or required.

Links

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.