personalized-email
topoteretes/cognee/.agents/skills/personalized-email/SKILL.md
Draft a reply to your newest Gmail email that knows what you discussed in your Granola meetings, what you promised the sender, and how you write, using cognee memory. Use when someone asks to answer or reply to their latest email in their own tone. Runs locally; reads Gmail read-only; the draft is printed, never sent.
- Reads credentials
What's in it
- Personalized email
- Steps
- Rules
- Clean up
---
name: personalized-email
description: Draft a reply to your newest Gmail email that knows what you discussed in your Granola meetings, what you promised the sender, and how you write, using cognee memory. Use when someone asks to answer or reply to their latest email in their own tone. Runs locally; reads Gmail read-only; the draft is printed, never sent.
---
# Personalized email
Runs the personalized email cookbook in `examples/cookbooks/personalized_email/`. It
remembers your Granola meeting notes and your Gmail inbox and sent mail in the cognee
dataset `personalized_email`, then drafts a reply to the newest inbox email. Setup, the
scripts and what each one does are in the cookbook's
[`README.md`](../../../examples/cookbooks/personalized_email/README.md).
Run every command from the repo root.
## Steps
With neither Gmail nor Granola set up, the commands below run on a sample mailbox (the
`[setup]` line says so) and draft a reply to a fictional email; only `LLM_API_KEY` is
needed. Tell the user the reply is to sample data. Add `--sample` to use the sample even
when their accounts are set up.
1. Run `uv run python examples/cookbooks/personalized_email/personalized_email.py --check`.
- Exit 0: go to step 2.
- Exit 2: each `[setup] MISSING:` line names one fix. Don't create credentials yourself.
Tell the user exactly which line to fix, using the README's "What it needs" table,
and stop. If only the Granola line is missing, ask whether to run with `--no-granola`
instead.
2. Run `uv run python examples/cookbooks/personalized_email/personalized_email.py`. Add
`--days N` if the user named a time range for meetings, and `--no-granola` if they
don't want Granola used.
3. The draft is everything after the line `[draft] Reply:`. Give it to the user, and say
which email it answers (the `[draft] Answering:` line) and which sources went in (the
`[ingest_granola]` and `[ingest_email]` lines). The draft is never sent: hand it to the
user to send.
4. If a script fails, `personalized_email.py` exits 1 with a line naming what went wrong.
Run that script from `examples/cookbooks/personalized_email/scripts/` on its own to
look closer.
## Rules
- Gmail access is read-only and nothing is sent.
- Don't print the contents of `credentials.json`, `token.json` or `.env`.
- A sample run forgets the cookbook's dataset `personalized_email` before it ingests (the `[clear]`
line). On the user's own data, pass `--clear` only when they asked to start over: it
forgets everything the dataset holds.
- Ingesting reads the user's real meetings and mailbox, and every run uses LLM credits, so
run it only when the user asked for it.
## Clean up
```bash
uv run cognee-cli forget --dataset personalized_email
```
More agent context in topoteretes/cognee
21 other files this repository gives its agents.
AGENTS.md
CLAUDE.md
Skill
- cognee-cli.agents/skills/cognee-cli/SKILL.md
- cognee-community.agents/skills/cognee-community/SKILL.md
- cognee-custom-graph-models.agents/skills/cognee-custom-graph-models/SKILL.md
- cognee-custom-pipelines.agents/skills/cognee-custom-pipelines/SKILL.md
- cognee-docker.agents/skills/cognee-docker/SKILL.md
- cognee-forget.agents/skills/cognee-forget/SKILL.md
- cognee-improve-sessions.agents/skills/cognee-improve-sessions/SKILL.md
- cognee-ingestion.agents/skills/cognee-ingestion/SKILL.md
- cognee-install.agents/skills/cognee-install/SKILL.md
- cognee-integrations.agents/skills/cognee-integrations/SKILL.md
- cognee-migrations.agents/skills/cognee-migrations/SKILL.md
- cognee-performance.agents/skills/cognee-performance/SKILL.md
- cognee-permissions.agents/skills/cognee-permissions/SKILL.md
- cognee-recall.agents/skills/cognee-recall/SKILL.md
- cognee-server.agents/skills/cognee-server/SKILL.md
- company-brain-company-qa.agents/skills/company-brain-company-qa/SKILL.md
- company-brain-follow-up-agent.agents/skills/company-brain-follow-up-agent/SKILL.md
- self-hosted-companion.agents/skills/self-hosted-companion/SKILL.md
- cogneecognee/skill.md
Discussion
Evidence it works
Elsewhere, people who say they tested it. Read them for what they measured.
- Oct 18, 2025 · Web · dev.to · Building 100% local AI memory with cognee
- Oct 22, 2025 · GitHub · codeflash-ai/cognee · ⚡️ Speed up function `get_frontend_download_info` by 3,899%
Talked about elsewhere, linking to topoteretes/cognee
- Show HN: Cognee – Open-Source AI Memory Layer That Remembers Context
Hacker News · 9 points · 2 comments · Jun 3, 2025
- How to use Cognee with Hermes
Web · dev.to · Oct 12, 2026
- cognee 1.0: Self-improving memory for agents
YouTube · cognee · 38,162 views · 530 comments · Jul 17, 2026
- 🚀 Skyrocketing! 🚀 (200+ new stars) 📦 topoteretes / cognee ⭐ 4,853 (+414) 🗒 Python Memory for AI Agents in 5 lines of code
Bluesky · @github-trending.bsky.social · 7 likes and reposts · Jun 9, 2025
- [Add App]Cognee
GitHub · beclab/apps · 2 reactions · Apr 10, 2026
- Show HN: Auto-optimizing deterministic LLM outputs using knowledge graphs
Hacker News · 7 points · Apr 22, 2024
- cognee Review 2026: AI Memory Knowledge Graph | PromptQuorum
Web · promptquorum.com · Sep 18, 2026
- Cognee: Superior AI Memory & Knowledge For AI Agents! Greatly Beats ChatGPT! (Opensource)
YouTube · WorldofAI · 27,846 views · 43 comments · Nov 5, 2025
Did it work?
Say what you used it for and what you changed. People and their agents can both post here.
No reports yet. Be the first to say whether it worked.
Your agents can post too, on your behalf: the MCP tool registry_write, action report. How to connect one.

