How are devs using Spec Kit with Neovim

[RizaHKhan]

I have Neovim Codecompanion, but it doesn't support slash commands from Spec Kit.
Nor does Copilot CLI.

How are Neovim devs using Spec Kit in their workflow?

[thecontinium]

I haven't tried it in anger yet but sidekick looks like it will work.

https://github.com/folke/sidekick.nvim

[airtonix]

opencode

[RizaHKhan]

@airtonix , thank you, one of those 'where have you been my life' tools.

With regards to the slash commands, I'm using Github Copilot and I want to tell it to use the commands listed on .github/prompts folder.

In this video timestamp: 5:02, we actually tell the agent to remember these commands like so:

But is there a better way to add these commands to the opencode agent?

[RizaHKhan]

Perhaps:

{
  "$schema": "https://opencode.ai/config.json",
  "theme": "everforest",
  "model": "github-copilot/claude-sonnet-4.5",
  "autoupdate": true,
  "permission": {
    "edit": "allow",
    "bash": {
      "git": "allow",
      "npm": "allow",
      "pytest": "allow"
    },
    "webfetch": "ask"
  },
  "command": {
    "test": {
      "template": "Run the test suite and fix any failures",
      "description": "Run tests"
    },
    "review": {
      "template": "Review the code for bugs, performance issues, and best practices",
      "description": "Code review"
    },
    "speckit-specify": {
      "template": ".github/prompts/speckit.specify.prompt.md",
      "description": "Create or update the feature specification from a natural language feature description"
    },
    "speckit-plan": {
      "template": ".github/prompts/speckit.plan.prompt.md",
      "description": "Execute the implementation planning workflow using the plan template to generate design artifacts"
    },
    "speckit-tasks": {
      "template": ".github/prompts/speckit.tasks.prompt.md",
      "description": "Generate an actionable, dependency-ordered tasks.md for the feature based on available design artifacts"
    },
    "speckit-analyze": {
      "template": ".github/prompts/speckit.analyze.prompt.md",
      "description": "Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md"
    },
    "speckit-implement": {
      "template": ".github/prompts/speckit.implement.prompt.md",
      "description": "Execute the implementation plan by processing and executing all tasks defined in tasks.md"
    },
    "speckit-clarify": {
      "template": ".github/prompts/speckit.clarify.prompt.md",
      "description": "Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions"
    },
    "speckit-checklist": {
      "template": ".github/prompts/speckit.checklist.prompt.md",
      "description": "Generate a custom checklist for the current feature based on user requirements"
    },
    "speckit-constitution": {
      "template": ".github/prompts/speckit.constitution.prompt.md",
      "description": "Create or update the project constitution from interactive or provided principle inputs"
    }
  },
  "watcher": {
    "ignore": [
      "node_modules/**",
      "vendor/**",
      ".git/**",
      "*.log",
      "dist/**",
      "build/**"
    ]
  },
  "share": "manual",
  "username": "khanr"
}

[RizaHKhan]

Actually, the above doesn't work; it simply ends up reading and regurgitating what's in those files.

Better to do what the video suggested, encode the commands in an AGENTS.md file like this:

## SpecKit Slash Commands

The agent has access to the following slash commands for specification-driven development:

### `/speckit.specify <feature description>`
Create or update feature specification from natural language.
- Generates spec.md with user stories, requirements, success criteria
- Creates branch and checklist for validation
- Use for: Starting new features, documenting requirements

### `/speckit.clarify [context]`
Identify and resolve spec ambiguities (up to 5 targeted questions).
- Detects missing decisions, vague requirements, edge cases
- Updates spec.md with clarifications
- Use after: `/speckit.specify`, before `/speckit.plan`

### `/speckit.plan [context]`
Generate technical implementation plan from spec.
- Creates plan.md, data-model.md, contracts/, research.md
- Resolves technical unknowns through research
- Use after: `/speckit.clarify` (or `/speckit.specify` if no ambiguities)

### `/speckit.tasks [context]`
Generate dependency-ordered implementation tasks from plan.
- Creates tasks.md organized by user story
- Identifies parallel execution opportunities
- Use after: `/speckit.plan`

### `/speckit.checklist <type> [options]`
Generate custom checklist for requirements validation.
- Types: ux, api, security, performance, etc.
- Tests requirement quality (completeness, clarity, consistency)
- Use: Any time to validate spec quality before implementation

### `/speckit.analyze [context]`
Cross-artifact consistency analysis (spec, plan, tasks).
- Detects duplications, conflicts, coverage gaps
- Validates constitution alignment
- Use after: `/speckit.tasks`, before `/speckit.implement`

### `/speckit.implement [context]`
Execute implementation plan by processing tasks.md.
- Checks checklist completion status
- Follows TDD if tests specified
- Creates ignore files, implements features phase-by-phase
- Use after: `/speckit.tasks` (and optionally `/speckit.analyze`)

### `/speckit.constitution [principles]`
Create or update project constitution (governance and principles).
- Defines non-negotiable development rules
- Syncs with spec/plan/task templates
- Use: Project setup or when changing core principles

### Command Flow

/speckit.specify → /speckit.clarify → /speckit.plan → /speckit.tasks → /speckit.analyze → /speckit.implement
(optional) (optional) (optional)

### When to Use SpecKit Commands
Use for:
- Feature development requiring clear requirements
- Complex features needing structured planning
- Team projects requiring shared understanding
- Features with multiple user stories or phases

Skip for:
- Small bug fixes
- Simple refactoring
- Configuration changes
- Trivial updates

[airtonix]

the ai isn't suppose to use slash commands. only you are.