moonlight-lupin/agent-skills / agent-ops/claude-plugin-converter
claude-plugin-converter
VerifiedUse when converting Claude Code plugins into self-contained Hermes plugins — discovery analysis then full conversion.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 13.9 KB |
| references/mapping-tables.md | markdown | 5.6 KB |
| scripts/analyze.py | script | 22.3 KB |
| scripts/convert.py | script | 66.7 KB |
| scripts/schemas/mcp.schema.json | config | 3.3 KB |
| scripts/schemas/plugin.schema.json | config | 1.8 KB |
| tests/test_converter.py | script | 59.3 KB |
references/mapping-tables.md
View on GitHub# Full Mapping Tables
## Contents
- Hook Event Mapping (Complete)
- Hook Type Mapping
- Agent Frontmatter Mapping
- Manifest Field Mapping
- MCP Config Format Mapping
## Hook Event Mapping (Complete)
| Claude Event | Hermes Hook | Convertibility | Notes |
|---|---|---|---|
| `SessionStart` | `on_session_start` | ✅ | Direct match. New session created. |
| `SessionEnd` | `on_session_end` | ✅ | Direct match. Session ends. |
| `Setup` | ❌ | ⏭️ | No Hermes equivalent (CI/one-time prep). |
| `UserPromptSubmit` | `pre_llm_call` | ✅ | Both fire before LLM processes. Hermes can inject context via `{"context": str}`. |
| `UserPromptExpansion` | ❌ | ⏭️ | No command expansion equivalent. |
| `PreToolUse` | `pre_tool_call` | ✅ | Both can block. Claude uses `matcher` field; Hermes checks `tool_name` param. Return `{"action": "block", "message": str}` to veto. |
| `PermissionRequest` | `pre_approval_request` | ✅ | Similar concept — approval prompt about to fire. |
| `PermissionDenied` | `post_approval_response` | ✅ | Similar — user responded to approval. |
| `PostToolUse` | `post_tool_call` | ✅ | Direct match. After tool returns. |
| `PostToolUseFailure` | `post_tool_call` | ⚠️ | Same hook, check `result` for error content. No separate failure hook in Hermes. |
| `PostToolBatch` | ❌ | ⏭️ | No batch-level hook in Hermes. |
| `Notification` | ❌ | ⏭️ | No notification hook in Hermes. |
| `MessageDisplay` | `transform_llm_output` | ⚠️ | Closest match — can transform output before delivery. Different semantics. |
| `SubagentStart` | `subagent_start` | ✅ | Direct match. delegate_task child about to run. |
| `SubagentStop` | `subagent_stop` | ✅ | Direct match. delegate_task child finished. |
| `TaskCreated` | ❌ | ⏭️ | No task-creation hook. Hermes has `todo` tool but no hooks. |
| `TaskCompleted` | ❌ | ⏭️ | No task-completion hook. |
| `Stop` | `post_llm_call` | ⚠️ | Closest match. Fires after tool-calling loop, before final response delivered. |
| `StopFailure` | ❌ | ⏭️ | No API-error hook. |
| `TeammateIdle` | ❌ | ⏭️ | No agent-teams equivalent in Hermes. |
| `InstructionsLoaded` | ❌ | ⏭️ | No context-file-loaded hook. |
| `ConfigChange` | ❌ | ⏭️ | No config-change hook. |
| `CwdChanged` | ❌ | ⏭️ | No cwd-change hook. |
| `FileChanged` | ❌ | ⏭️ | No file-watcher hook. |
| `WorktreeCreate` | ❌ | ⏭️ | No worktree hook. |
| `WorktreeRemove` | ❌ | ⏭️ | No worktree hook. |
| `PreCompact` | ❌ | ⏭️ | No pre-compaction hook. |
| `PostCompact` | ❌ | ⏭️ | No post-compaction hook. |
| `Elicitation` | ❌ | ⏭️ | No MCP elicitation hook. |
| `ElicitationResult` | ❌ | ⏭️ | No MCP elicitation hook. |
**Summary: 10 direct matches, 4 partial, 16 no equivalent.**
## Hook Type Mapping
| Claude Type | Implementation | Notes |
|---|---|---|
| `command` | `subprocess.run(command, shell=True, ...)` in Python callback | Translate `${CLAUDE_PLUGIN_ROOT}` to plugin dir. |
| `http` | `httpx.post(url, json=event_data, timeout=...)` in Python callback | Event data = the context dict. |
| `mcp_tool` | `ctx.dispatch_tool(tool_name, args)` in Python callback | Strip plugin namespace from tool names. |
| `prompt` | ⏭️ No equivalent | Note in MANUAL_STEPS. Could spawn one-shot AIAgent. |
| `agent` | ⏭️ No equivalent | Note in MANUAL_STEPS. Could use delegate_task. |
## Agent Frontmatter Mapping
| Claude Field | Hermes Handling | Notes |
|---|---|---|
| `name` | Skill name | Used in `ctx.register_skill()` |
| `description` | Skill description | Copied to SKILL.md frontmatter |
| `model` | Noted in delegation skill | Hermes uses `delegate_task` which inherits parent model. Note as config hint. |
| `effort` | Noted in delegation skill | No direct equivalent. |
| `maxTurns` | Noted in delegation skill | Map to `delegation.max_iterations` concept. |
| `tools` | Noted in delegation skill | If restricted, note which tools to pass. |
| `disallowedTools` | Noted in delegation skill | If specified, note which tools to block. |
| `skills` | Bundled alongside | If agent references skills, copy them too. |
| `memory` | ❌ | No per-agent memory in Hermes. Note in report. |
| `background` | ❌ | No background-agent flag. Note in report. |
| `isolation` | Noted | `worktree` isolation maps to Hermes `-w` flag. |
## Manifest Field Mapping
| Claude `plugin.json` | Hermes `plugin.yaml` | Notes |
|---|---|---|
| `name` | `name` | Direct copy. Used as plugin identifier. |
| `description` | `description` | Direct copy. |
| `version` | `version` | Direct copy. |
| `author.name` | `author` | Flatten to string. |
| `author` | `author` | If string, direct copy. |
| `homepage` | `homepage` | Direct copy (optional). |
| `repository` | `repository` | Direct copy (optional). |
| `license` | `license` | Direct copy (optional). |
| (missing) | `provides_tools` | Generated from MCP servers if present. |
| (missing) | `provides_hooks` | Generated from hooks.json if present. |
| (missing) | `requires_env` | Inferred from MCP server env vars. |
## MCP Config Format Mapping
Claude `.mcp.json`:
```json
{
"mcpServers": {
"server-name": {
"command": "${CLAUDE_PLUGIN_ROOT}/servers/db-server",
"args": ["--config", "config.json"],
"env": { "DB_PATH": "/data" }
}
}
}
```
Hermes `mcp_config.yaml`:
```yaml
mcp_servers:
server-name:
command: /path/to/servers/db-server
args: ["--config", "config.json"]
env:
DB_PATH: /data
```
Key differences:
- `${CLAUDE_PLUGIN_ROOT}` → absolute path to plugin directory
- `mcpServers` (camelCase) → `mcp_servers` (snake_case)
- JSON → YAMLRelated skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |