Skip to main content
All of your session data stays on your machine. CostHQ writes everything to a local SQLite database that is created automatically the first time you run any cs command — nothing is sent to a remote server and no account is required.

Database location

PlatformPath
Mac / Linux~/.costhq/sessions.db
WindowsC:\Users\<username>\.costhq\sessions.db

Custom database path

You can point CostHQ at a different file using the COSTHQ_DATA_DIR environment variable. This is useful for per-project isolation or test environments:
export COSTHQ_DATA_DIR=/path/to/sessions.db

Privacy guarantees

  • No cloud sync — all session data stays on your local machine.
  • No telemetry — CostHQ never phones home or sends usage data externally.
  • No API key storage — the proxy forwards Authorization headers as-is and never writes them to disk.
  • Full control — delete or export your data at any time without contacting anyone.

Backing up and exporting your data

# Copy the database file
cp ~/.costhq/sessions.db ~/backups/costhq-$(date +%Y%m%d).db

# Export session list as plain text
cs list -l 1000 > sessions-export.txt

# Export as CSV
cs export --format csv > sessions.csv

# Export as JSON (last 50 sessions)
cs export --format json --limit 50 > sessions.json

Migration from older versions

If you used CostHQ when it stored data under ~/.devsession/, your data migrates automatically to ~/.costhq/ on first run. The migration copies files — it does not delete the old directory. Once you confirm that everything looks correct in the new location, you can safely remove ~/.devsession/ manually.

Resetting all data

Deleting the database is permanent and cannot be undone. Back up the file before you proceed.
# Back up first!
cp ~/.costhq/sessions.db ~/sessions-backup.db

# Then delete to reset
rm ~/.costhq/sessions.db
CostHQ recreates the database automatically the next time you run any command.

Concurrency

CostHQ opens the SQLite database in WAL (Write-Ahead Logging) mode, which allows multiple agents or terminal sessions on the same machine to write records simultaneously without blocking each other. You can run parallel agent sessions — for example in separate project directories — and all writes land safely in the same database.