Skip to main content
Organize your transcription library with broad topic tags like “AI”, “Health”, “Gaming”, “Startups”. Tags appear in the Web UI Memory Explorer as clickable filters. New transcriptions are automatically tagged using semantic search — content is compared against existing tagged transcriptions and matching tags are assigned. Claude also adds new categories when needed via the MCP tagging hint.

​
Example: Add tags

Request: 
{
  "cache_key": "abc123:tiny",
  "action": "add",
  "tags": ["AI", "Startups"]
}
Response: 
{
  "action": "add",
  "cache_key": "abc123:tiny",
  "added": [
    {"name": "AI", "category": "topic", "source": "auto"},
    {"name": "Startups", "category": "topic", "source": "auto"}
  ]
}

​
Example: List tags

Request: 
{
  "cache_key": "abc123:tiny",
  "action": "list"
}
Response: 
{
  "action": "list",
  "cache_key": "abc123:tiny",
  "tags": [
    {"name": "AI", "category": "topic", "source": "auto"},
    {"name": "Startups", "category": "topic", "source": "auto"}
  ]
}

​
Example: Remove tags

Request: 
{
  "cache_key": "abc123:tiny",
  "action": "remove",
  "tags": ["Startups"]
}
Response: 
{
  "action": "remove",
  "cache_key": "abc123:tiny",
  "removed": 1
}

​
Parameters

ParameterRequiredDefaultDescription
cache_keyYes—The cache key of the transcription (returned by transcribe_audio and take_notes)
actionYes—add, remove, or list
tagsNo—Array of tag names. Required for add and remove actions

​
Notes

Use 2-4 broad, high-level categories per transcription. Good: “AI”, “Health”, “Gaming”. Bad: people’s names, specific tools, granular subtopics.
Tags are case-insensitive. “ai” and “AI” are treated as the same tag.
The cache_key is returned in the response of transcribe_audio and take_notes. You don’t need to construct it manually.
Tags appear as clickable filters in the Web UI Memory Explorer. Click a tag to show only transcriptions with that topic.