HiveMind

Daemon for syncing agentic coding sessions to W&B.

Installation

# Recommended (macOS Apple Silicon + Linux)
curl -fsSL https://hivemind.wandb.tools/install | sh

# Via Homebrew (macOS)
brew install wandb/taps/wandb-hivemind

# Any platform, incl. Intel Macs and Windows
uv tool install wandb-hivemind

Usage

# Run the daemon (watches for new sessions)
hivemind run

# Check status
hivemind status

# Fork a previous session
hivemind fork <session_id>

# View logs
hivemind logs
hivemind logs -f  # Follow logs

Authentication

# Login (auto-detects best method)
hivemind login

# Check authentication status
hivemind whoami

# Logout
hivemind logout

Importing Existing Sessions

For development or initial setup, import existing sessions:

# Import ALL sessions found on disk
hivemind import

# Import sessions from the last 24 hours
hivemind import --since 1d

# Import sessions from the last 2 hours
hivemind import --since 2h

# Import sessions since a specific date
hivemind import --since 2024-01-15

# Dry run to preview what would be imported
hivemind import --dry-run

# Import from specific paths
hivemind import --path ~/projects/myapp

# Custom endpoint override
hivemind import --endpoint https://api.example.com

Force Re-sync

If you need to re-import all sessions for a specific agent (for example, after fixing sync issues or to apply updated normalization):

# Re-sync all Cursor sessions
hivemind resync cursor

# Preview what would be deleted
hivemind resync cursor --dry-run

# Skip confirmation prompt
hivemind resync cursor --yes

Supported agent types: claude, claude_web, cursor, codex, gemini, opencode

Warning: This permanently deletes your session data from the server. The data will be re-imported from your local session files.

Rate limiting: You can resync each agent type once per hour.

Configuration

Configuration file: ~/.hivemind/config.toml

# View current config and where values come from
hivemind config show

# Set endpoint (e.g., for QA testing)
hivemind config set server.endpoint https://hivemind.qa.wandb.tools

# Get a specific value
hivemind config get server.endpoint

# Reset to defaults
hivemind config reset

# Show config file path
hivemind config path

Example config file:

[server]
endpoint = "https://hivemind.qa.wandb.tools"

Configuration Priority

Settings are resolved in this order (highest to lowest):

Config file (~/.hivemind/config.toml) - user's explicit preference
Environment variable (HIVEMIND_ENDPOINT) - for testing/CI overrides
Default (https://hivemind.wandb.tools)

The daemon automatically reloads when the config file changes.

Environment Variables

HIVEMIND_ENDPOINT - Backend API endpoint (overridden by config file)
HIVEMIND_IMPORT_SINCE - Default --since filter for imports

Troubleshooting

Resync Issues

"Cannot modify state: daemon is running" The resync command will automatically signal the running daemon. If you see this error, the signaling mechanism may have failed. Try stopping the daemon and running resync again.

"Rate limit exceeded" You can only resync each agent type once per hour. Wait for the time shown in the error message.

"Backend cleanup succeeded but local state update failed" Your server data has been deleted. The path to your state.json file is shown in the error. You can manually remove the sessions for that agent type, or simply restart the daemon - it will re-discover and re-sync all sessions.

Metadata

Release files for wandb-hivemind 1.1.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Built distribution (wheel)

Table of built distributions (wheels) for wandb-hivemind 1.1.0
File	Interpreter	ABI	Platform	Reset
wandb_hivemind-1.1.0-py3-none-any.whl 448.6 kB Sep 22, 2026	Python 3	none	any	Details

Release files / wandb_hivemind-1.1.0-py3-none-any.whl

Download URL	wandb_hivemind-1.1.0-py3-none-any.whl
Size	448.6 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`bf36720e5fb7a2c408d062641a2c952b7066b5cced3362426394b2ff4e90d2d3`
BLAKE2b-256 checksum How to use checksums	`1b01d7325679ff43d0df9e45885f181542d0dc4e8c25c9d14a2dc4a4c682b061`
Upload date	Sep 22, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	Yes
Uploaded via	`twine/7.0.0 CPython/3.13.14`

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Identity

Publishing platform: GitHub Actions
Publishing repository: github.com/wandb/agentstream
Publishing commit: github.com/wandb/agentstream/tree/ce987a6185c4566f5f99a7b26e1dd4cf118ed314

Workflow

Publishing configuration: github.com/wandb/agentstream/blob/ce987a6185c4566f5f99a7b26e1dd4cf118ed314/.github/workflows/release.yml
Publishing logs: github.com/wandb/agentstream/actions/runs/35791668237/attempts/1

Artifact

Subject: wandb_hivemind-1.1.0-py3-none-any.whl
SHA-256 checksum: bf36720e5fb7a2c408d062641a2c952b7066b5cced3362426394b2ff4e90d2d3 Verifying attestations