| dataset: slack-bench | |
| tags: | |
| - agent-evals | |
| - slack | |
| - api | |
| - messaging | |
| license: mit | |
| # Slack Bench | |
| Slack Bench contains 37 tasks for evaluating AI agents on Slack API operations. | |
| ## Task Categories | |
| - **Basic messaging**: Send messages, DMs, group conversations | |
| - **Channel operations**: Create channels, invite users, archive | |
| - **Rich text formatting**: Block Kit (bold, italic, code blocks, lists, tables) | |
| - **Thread replies**: Reply to specific messages in threads | |
| - **Cross-channel operations**: Copy/summarize content between channels | |
| - **User management**: Add/remove users from channels | |
| - **Reactions**: Add emoji reactions to messages | |
| ## Files | |
| - `data/train.jsonl` | |
| - `dataset_infos.json` | |
| - `seeds/slack_bench_default.json` | |
| ## How to run? | |
| **Colab:** [Run Linear-Bench](https://colab.research.google.com/drive/1Hext-WWDsm9BxsOrASYoMjgu1N_lN0Fz?usp=sharing) | **Docs:** [Agent Diff Evaluation Docs](https://agentdiff.mintlify.app/core-concepts/evaluations) | |
| ## Read more | |
| **Website:** [agentdiff.dev](https://agentdiff.dev) | **Repo:** [Github](https://github.com/hubert-marek/agent-diff) | |