#cfp #conference
CFP Tracker is your personal CFP sleuth. It collects conference Calls for Papers (CFPs) from public sources, saves them as Markdown notes, and organizes them into Obsidian Bases views.
This is a personal tracker, not another CFP index. Thank you to all the CFP site maintainers whose public feeds, datasets, calendars, and CFP pages make this project possible.
Created by sujee.dev · Project page · Updates @sujee_dev
Made with ❤️ as open source software, using opencode + Kimi 3.1 Pro and DeepSeek 4.1 Flash.
Click a thumbnail to open the video or a full-size screenshot.
| | | | | Upcoming<sub>Open CFPs, nearest deadline first</sub> | Short listed<sub>Your picks, in one view</sub> | Refresh<sub>One command, per-source counts</sub> |
- Python 3.10 or later, to fetch the CFPs.
- Git, to clone the repository.
- Obsidian , to view and manage the notes. Enable the Bases core plugin.
You do not need Obsidian to refresh the CFPs. The pipeline writes plain Markdown.
Optional:
- uv , for the quickest Python setup.
- The Dataview plugin, for the dashboard.
git clone git@github.com:sujee/cfp-tracker.git
cd cfp-tracker
ln -s /path/to/your-vault/cfps/data data
ln -s /path/to/your-vault/cfps/archived archived
With uv (recommended)
uv sync
Without uv
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
src/config.py contains the defaults.
You can customize as below; only customize what you need to overwrite
cp src/config_local.example.py src/config_local.py # <- gitignored
You can configure:
- Sources to fetch with
INPUT_SOURCES. - Event types to retain with
EVENT_TYPES. - Event names and topics to reject with
IGNORE_NAMESandIGNORE_TOPICS.
See configuration
Option-1 (easy): Using an agent
Open an agent in this folder. Then enter:
update cfps
That's it!
Option 2: To run the pipeline yourself:
uv run python src/process-all.py --out /tmp/cfp-scratch --no-archive
uv run python src/process-all.py
The run prints a summary for each source. It also prints an event-type matrix.
Use uv run python src/process-all.py --help to see all options.
Fetched CFPs are saved in plain md format like below.
---
type: cfp
event_name: Example Conf 2026
event_date: 2026-12-01
event_type: conference
location: Berlin, Germany
country: Germany
cfp_close: 2026-10-31
topics: [ai, cloud]
event_url: https://example.com
cfp_url: https://example.com/cfp
source: developers.events
shortlist: false
dismissed: false
submissions: []
created: 2026-11-02T00:00:00
updated: 2026-11-12T00:00:00
---
## Notes
_Add notes here._
See the note format reference for every field, and for the rules that decide which values survive a refresh.
Copy cfps.base into your vault. Copy
cfps-dashboard.md too if you want summary tables.
The base requires Obsidian Bases (included with Obsidian core). The dashboard requires the Dataview plugin.
The base reads notes from data/. The archive views read notes from archived/. If your notes use
different folders, update the paths in both files.
Personalize the location views in the formulas section of cfps.base:
- Set
my_countryto the canonicalcountryvalue for your notes. - Edit
is_local_areato list your local cities and area names.
- Refresh the CFPs (run python fetcher)
- Open Upcoming view in Obsididian base (
cfps.base) to find open CFPs. - Set
shortlist: truefor interesting events. - Set
dismissed: trueto hide an event from live views. - Add wiki links to
submissionsafter you submit a talk.
Most views let you include or exclude dismissed notes.
Since the CFPs are saved as local markdown files, you can use an Agent to analyze them
cd cfps
start your agent and ask questions like
find me all AI conference CFPsfind me CFPs that are due within a month
prestige.md assigns optional reputation tiers. Tier 1 means flagship, tier 2 means major, and tier 3 means notable. The refresh adds these tiers to matching notes.
A tier describes the conference. Your shortlist describes your interest.
uv run python src/apply-prestige.py --check
The pipeline reads from various sources.
See docs/cfp-sources.md for source details and attribution.
Run the tests:
uv run python -m unittest discover -s tests -v
Project details:
docs/README.md: pipeline, note format, configuration, and commands.INSTRUCTIONS.md: refresh procedure and operating rules.AGENTS.md: data shapes, configuration, and preservation rules.docs/spec.md: historical specification from 2026-11-09. UseINSTRUCTIONS.mdandAGENTS.mdfor current behavior.
MIT. See LICENSE.