Your agent can search your email.
It can’t understand it.
Deepmail gives AI agents persistent email understanding. 10 MCP tools to investigate, cross-reference, and remember—across sessions.
Every agent session starts from zero.
Gmail MCP servers let agents read individual messages. But understanding—knowing who matters, what relationships exist, what’s happening—requires investigation, cross-referencing, and memory.
Agents can’t do that in a single search.
What your agent gets
- search_emails Search Gmail by any query
- read_email Read the full body of a message
- build_profile Deep investigation—extract people and context from Gmail
- build_status Check progress of a running investigation
- who_is Look up a person from cached knowledge
- about_me What the agent knows about you
- get_candidates Review structured results from the latest build
- profile_health Check freshness and coverage of cached profiles
- reset_profile Wipe generated data and start fresh
- check_auth Verify Gmail authentication status
How it works
1
Install
pip install deep-email
2
Authenticate
deep-email auth
3
Your agent knows you
The skill tells your agent what tools are available and when to use them. No configuration needed.
Works with Claude Code, Cursor, Codex, GitHub Copilot, Windsurf, Gemini CLI, and 40+ other MCP-compatible agents.
Example
“Find all my investors.”
- Agent searches for term sheets, fund communications, board emails
- Follows leads—reads specific messages, cross-references names
- Builds a complete investor table across 3 funding rounds
Next session: already knows. No re-searching.
Get started
$ pip install deep-email
$ deep-email auth
$ npx skills add dennisonbertram/deepmail
View on GitHub →