How do I connect Cursor to Dailybot?

Install the Dailybot CLI in Cursor’s integrated terminal, authenticate with dailybot login or DAILYBOT_API_KEY, add agent_scripts/dailybot-report.sh to your repo, and document when to report in .cursorrules. Send a test report and confirm it in your Dailybot feed.

What .cursorrules configuration is needed?

Include clear instructions for when the agent should run bash agent_scripts/dailybot-report.sh, with reporting thresholds (for example: report after shipped features and bug fixes, not after typos or lockfile-only changes) and that --metadata should include the model identifier.

How do I verify the connection works?

Run the reporter script manually with a test message. If the entry appears in your team feed within a few seconds and shows expected metadata in the dashboard, the integration is working.

Connecting Cursor to Dailybot (step-by-step) - Academy

This is the hands-on tutorial. The Cursor integration overview explains what the connection does; here you get the exact steps. By the end, agents working in Cursor can post standup-style progress to your Dailybot feed when they finish meaningful work.

Note: Claude Code projects often use CLAUDE.md or AGENTS.md for agent instructions. In Cursor, the usual place for those rules is .cursorrules. The CLI and agent_scripts/ flow are the same; only where you document behavior changes.

What you need before starting

A Dailybot account (free tier is fine).
Cursor installed, with a project open.
Terminal access—use Cursor’s integrated terminal so paths and environment match where your agent runs commands.

If you develop inside a devcontainer, install and authenticate inside that environment so sessions persist the way your team expects.

Step 1: Install the Dailybot CLI in Cursor’s terminal

Open the integrated terminal (Terminal → New Terminal) and install the CLI globally:

npm install -g @dailybot/cli

Confirm it is on your PATH:

dailybot --version

For devcontainers, add a postCreateCommand or Dockerfile step so npm install -g @dailybot/cli runs on rebuild—same pattern as other global tools.

Step 2: Authenticate

Interactive login (works well on your machine):

dailybot login [email protected]

Follow the email verification flow and pick the right workspace if prompted.

For CI-like or headless setups, use an API key instead:

export DAILYBOT_API_KEY="your-api-key-here"

Create the key from the Dailybot dashboard under Settings. Never commit secrets to git.

Step 3: Add the reporter script to the project

Most repos keep the wrapper under agent_scripts/ at the repository root (for example agent_scripts/dailybot-report.sh). It formats messages, injects repo and branch metadata, and calls the Dailybot API.

If your project does not have it yet, create agent_scripts/ and add the script from Dailybot’s documentation or from a template repo your team uses. The agent should invoke that script, not hand-roll curl calls.

Step 4: Configure `.cursorrules` for reporting

Edit .cursorrules (or merge into your existing rules) so the agent knows when and how to report. Be explicit about the threshold so you do not flood the feed.

Examples—do report after: shipping a feature, fixing a user-facing bug, finishing a multi-step refactor, completing a milestone plan, or several related commits that tell a coherent story.

Examples—do not report after: typos, formatting-only edits, dependency bumps with no functional change, exploratory reading, or unfinished work.

Include a concrete command pattern, for example:

bash agent_scripts/dailybot-report.sh "Short standup-style summary of what changed and why it matters." \
  --metadata '{"model":"your-model-id"}'

Adjust flags (--milestone, --json-data) when your team’s workflow needs richer updates.

Step 5: Test the connection

From the project root in Cursor’s terminal:

bash agent_scripts/dailybot-report.sh "Test report — verifying Cursor integration with Dailybot." \
  --metadata '{"model":"cursor-test"}'

Open your team’s Dailybot feed. The message should appear within a few seconds, with context such as repository and branch when metadata detection succeeds.

Step 6: Verify in the dashboard

In the Dailybot web app, open the feed or reporting view for your workspace. Confirm the test entry, the source labels, and that metadata looks correct. If project name warnings appear, set DAILYBOT_PROJECT_NAME in .env or your environment per the reporter script’s comments.

Troubleshooting

“CLI not authenticated”: Run dailybot login again or export a valid DAILYBOT_API_KEY. In containers, mount env files or use compose environment—do not bake keys into images.

Script not found: Run commands from the repo root, or use the path relative to root (bash agent_scripts/dailybot-report.sh).

Reports missing: Check workspace settings, network egress from your environment, and that you are logged into the intended Dailybot organization.

Rules ignored: Ensure .cursorrules is at the project root and mentions reporting explicitly; Cursor reads that file for agent behavior, unlike Claude Code’s CLAUDE.md convention.

Once this works, your team gets the same visibility for Cursor-driven work as for other agents—without extra manual copy-paste in chat.