Skip to content

[Bug]: CommandRegistrar.render_skill_command drops argument-hint and post_process_skill_content on native skills agents #5448

Backlog

Description

@adityajai119

Bug Description

For native /SKILL.md integrations (claude, alquimia, droid, vibe, codex, kimi, agy, hermes), specify extension add, specify init --extension, and preset command alias registration write SKILL.md files through CommandRegistrar.register_commands() -> CommandRegistrar.render_skill_command() (src/specify_cli/agents.py).

Unlike ExtensionManager._register_extension_skills(), _PresetSkillMethods, and SkillsIntegration.setup(), CommandRegistrar.render_skill_command() neither calls self.apply_argument_hint(frontmatter, skill_frontmatter, integration) nor runs integration.post_process_skill_content(...). Because _register_commands_for_active_agent() runs before _register_extension_skills() during ExtensionManager.install_from_directory(..., register_commands=True), _register_extension_skills() sees the file already on disk and skips it.

As a result:

Steps to Reproduce

specify init demo-proj --integration claude --extension github --ignore-agent-tools --non-interactive
Get-Content demo-proj\.claude\skills\speckit-github-taskstoissues\SKILL.md -TotalCount 15

Or install any extension command carrying argument-hint: "<target>" in its YAML frontmatter into a Claude Code project via specify extension add <ext-dir> --dev and inspect .claude/skills/<skill>/SKILL.md.

Expected Behavior

CommandRegistrar.render_skill_command() applies apply_argument_hint (gated to integrations defining inject_argument_hint) before YAML serialization and runs integration.post_process_skill_content afterward, matching the other SKILL.md generators so argument-hint, user-invocable, disable-model-invocation, and skills-mode hook-command notes are preserved.

Actual Behavior

speckit-github-taskstoissues/SKILL.md is written with only the four base keys (name, description, compatibility, metadata), without user-invocable: true, disable-model-invocation: false, or the skills-mode hook-command note, and extension commands or preset command aliases with argument-hint lose argument-hint.

Specify CLI Version

1.4.1.dev0 (main at 3050be0; also present in 1.4.0)

AI Agent

Claude Code

Operating System

Windows 11 (platform-independent Python path)

Python Version

Python 3.14.7

Error Logs

---
name: speckit-github-taskstoissues
description: Convert existing tasks into actionable, dependency-ordered GitHub issues for the feature based on available design artifacts.
compatibility: Requires spec-kit project structure with .specify/ directory
metadata:
  author: spec-kit-core
  source: github:commands/speckit.github.taskstoissues.md
---

Additional Context

PR #2978 noted in its scope section that agents.py::render_skill_command still omitted argument-hint and left it for a follow-up. I have a fix with regression tests and will open a PR linking this issue right away.

AI Disclosure

Found, reproduced, and written up by OpenCode using Gemini 4 Carbon (high reasoning effort, autonomous agent mode) on behalf of @adityajai119; issue discovery, local CLI reproduction, and drafting.

Activity

  1. added
    triage-must-haveVerdict: high-value, important work for Spec Kit — do first
    on Nov 16, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    triage-must-haveVerdict: high-value, important work for Spec Kit — do first

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions