command-prompt-sync
Ensures synchronization between Claude commands and Copilot prompt files.
Install
mkdir -p .claude/skills/command-prompt-sync && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/14953" && unzip -o skill.zip -d .claude/skills/command-prompt-sync && rm skill.zipInstalls to .claude/skills/command-prompt-sync
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
Detects and resolves drift between .claude/commands/*.md (Claude commands) and .github/prompts/*.prompt.md (Copilot prompt files). Ensures every command has a matching prompt and vice versa. Invoked exclusively by the agent-manager agent.Key capabilities
- →Detect drift between Claude commands and Copilot prompt files
- →Flag orphan commands or prompts
- →Sync a single named command to its corresponding prompt file
- →Sync all commands from the Claude directory to prompt files
- →Show a structured diff between a Claude command and Copilot prompt
- →Extract `name` and `description` from frontmatter
How it works
The skill detects and resolves drift between Claude command files and Copilot prompt files by comparing their content and synchronizing them, using Claude commands as the authoritative source.
Inputs & outputs
When to use command-prompt-sync
- →Sync commands and prompts
- →Audit inventory
- →Resolve drift
About this skill
Command or Prompt Sync
Use this skill when command and prompt files have drifted apart, when a command was added to one platform but not the other, or when a full inventory audit is needed. Parse the args to determine the operation mode, then execute the corresponding procedure.
Phase 0 — Context Load (silent)
- Read
.claude/CLAUDE.mdandAGENTS.mdto internalize project conventions and portability rules. - Invoke
Skill("manage-memory", args: "agent-manager")to load persistent memory (prior sync decisions, known naming conventions). - Glob
.claude/commands/*.mdand.github/prompts/*.prompt.mdto build the current command/prompt inventory.
Mode: audit
Args: audit
Scan all commands and prompts, flag orphans and content drift.
- Glob
.claude/commands/*.md— for each file read frontmatter and extract:name(derived from filename),description. - Glob
.github/prompts/*.prompt.md— for each file read frontmatter and extract:name(derived from filename, strip.promptsuffix),description,agent,argument-hint. - Normalise names: strip
.md/.prompt.mdextensions to get the bare skill name. - Build a cross-reference map keyed on bare name.
- Classify each entry:
- present-both — command exists in
.claude/commands/and prompt exists in.github/prompts/ - command-only —
.claude/commands/<name>.mdexists but no.github/prompts/<name>.prompt.md - prompt-only —
.github/prompts/<name>.prompt.mdexists but no.claude/commands/<name>.md - drifted — both files exist but
descriptiondiffers significantly
- present-both — command exists in
- For drifted pairs, show the description from each side.
- Output a structured audit report:
## Command / Prompt Drift Report
| Name | Claude command | Copilot prompt | Status | Drift |
|------|---------------|----------------|--------|-------|
| <name> | ✅ | ✅ | in sync | — |
| <name> | ✅ | ❌ | — | missing prompt |
| <name> | ❌ | ✅ | — | missing command |
| <name> | ✅ | ✅ | drifted | description mismatch |
- Summarise counts: total pairs, fully in sync, drifted, orphans.
Mode: sync
Args: sync <name>
Sync a single named command→prompt pair, using the Claude command as the authoritative source.
- Read
.claude/commands/<name>.md; fail with a clear error if it does not exist. - Extract frontmatter
descriptionfrom the Claude command. - Read
.github/prompts/<name>.prompt.mdif it already exists (to detect what would change). - Derive the Copilot prompt content from the Claude command:
- Frontmatter: set
descriptionto match Claude command'sdescription; preserve or setagent: "agent"andargument-hint - Body: adapt the Claude command body for Copilot prompt format (see Phase 3 templates in command-management skill)
- Frontmatter: set
- Write
.github/prompts/<name>.prompt.md. - Confirm: "Command
<name>synced to.github/prompts/<name>.prompt.md."
Mode: sync-all
Args: sync-all
Run sync <name> for every command currently in the Claude commands directory.
- Glob
.claude/commands/*.md— collect all bare names. - For each name, execute Mode: sync in sequence.
- Produce a summary: pairs synced, any errors encountered.
Mode: diff
Args: diff <name>
Show a structured diff between the Claude command and Copilot prompt for a single name.
- Read
.claude/commands/<name>.md; fail if it does not exist. - Read
.github/prompts/<name>.prompt.md; fail if it does not exist. - Compare frontmatter:
description— show both; flag if differentagent— Copilot only; note presence/absenceargument-hint— Copilot only; note presence/absence
- Compare body sections: identify headings present in one file but absent from the other.
- Output:
## Command/Prompt Diff: <name>
### Frontmatter
| Field | Claude command | Copilot prompt |
|-------|---------------|----------------|
| description | <val> | <val> |
| agent | n/a | <val> |
| argument-hint | n/a | <val> |
### Body Sections
| Section heading | Claude command | Copilot prompt |
|-----------------|---------------|----------------|
| ## Usage | ✅ | ✅ |
| ## Modes | ❌ | ✅ |
- Conclude with: In sync / Drifted — run
sync <name>to resolve.
Validation Rules
- Both
.claude/commands/<name>.mdand.github/prompts/<name>.prompt.mdmust exist as a pair. descriptionfrontmatter is required in both files.- Copilot prompt files must have
agent: "agent"in frontmatter. - Never delete any command or prompt file — use deprecation patterns when retiring.
- Never read or modify
.envfiles or sensitive configuration.
When not to use it
- →When reading or modifying `.env` files
- →When reading or modifying sensitive configuration
- →When deleting any command or prompt file
Limitations
- →The skill must never delete any command or prompt file
- →It must never read or modify `.env` files
- →It must never read or modify sensitive configuration
How it compares
This skill automates the synchronization and auditing of command and prompt files across different platforms, preventing manual drift and ensuring consistency.
Compared to similar skills
command-prompt-sync side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| command-prompt-sync (this skill) | 0 | 3mo | Review | Intermediate |
| n8n-mcp-tools-expert | 7 | 2mo | No flags | Advanced |
| workflow-automation | 3 | 6mo | Review | Intermediate |
| agent-workflow-automation | 4 | 6mo | Review | Advanced |
Try saying
Example prompts that trigger this skill in your AI assistant.
More by rafsanulhasan
View all by rafsanulhasan →You might also like
n8n-mcp-tools-expert
czlonkowski
Expert guide for using n8n-mcp MCP tools effectively. Use when searching for nodes, validating configurations, accessing templates, managing workflows, or using any n8n-mcp tool. Provides tool selection guidance, parameter formats, and common patterns.
workflow-automation
ruvnet
Workflow creation, execution, and template management. Automates complex multi-step processes with agent coordination. Use when: automating processes, creating reusable workflows, orchestrating multi-step tasks. Skip when: simple single-step tasks, ad-hoc operations.
agent-workflow-automation
ruvnet
Agent skill for workflow-automation - invoke with $agent-workflow-automation
workflow-execute
catlog22
Coordinate agent execution for workflow tasks with automatic session discovery, parallel task processing, and status tracking. Triggers on "workflow execute".
project-os
Peiiii
AI project OS for autonomous loop, automated orchestration, and rule-driven execution.
Release
tobagin
Create a new Keymaker release — analyze changes, bump version, update changelog/README/AppStream metadata, build, commit, tag, push, and create GitHub release