MCP tools reference
Every tool the prompt2love MCP server exposes, with its inputs and outputs.
All tools are scoped to your own library (per-user OAuth) and are read-only. Nothing here can edit, publish, or delete.
Generic tools (ChatGPT convention)
These two follow the ChatGPT connector convention and are what most clients call automatically.
| Tool | Input | Output |
|---|---|---|
search | query (string) | results[] — each { id, title, url } |
fetch | id (string) | { id, title, text, url, metadata } |
Typical flow: the client calls search(query) to find candidates, then fetch(id) on the chosen result to load its full text.
prompt2love tools
| Tool | Input | Returns |
|---|---|---|
list_my_prompts | — | Your prompts |
get_prompt | id or slug | One prompt |
search_my_prompts | query (string) | Matching prompts |
list_my_skills | — | Your skills |
get_skill | id or slug | One skill |
get_prompt and get_skill accept either an ID or a slug, so you can reference items the way you find most convenient.
The /prompt primitive
/prompt is a slash primitive (not a tool call) that inserts a prompt straight into the conversation. In Claude Code it is namespaced:
/mcp__prompt2love__prompt <term>
Use it when you want to drop a known prompt in directly instead of searching and fetching.
Field notes
id— stable identifier for an item.slug— human-readable URL fragment; an alternative key forget_*.url— link back to the item on prompt2love.text— the full content returned byfetch.metadata— extra structured fields attached to the item.