JSON Output

Every Atlas command supports --json for machine-readable output, making it straightforward to integrate Atlas with scripts, pipelines, and other tools.

Using —json

Add --json to any command:

atlas status --json
atlas history ~/README.md --json
atlas find --duplicates --json
atlas log --today --json
atlas inputs ~/project/index.html --json
atlas compare ~/project-v1/ ~/project-v2/ --json

Output is written to stdout as JSON. Errors and warnings go to stderr, so they won’t interfere with JSON parsing.

Export Format

The atlas export command outputs JSON Lines (JSONL) — one JSON object per line:

atlas export

The first line is always a version header:

{"atlas_export_version":1,"schema_version":10}

Subsequent lines include a _table field indicating their source table:

{"id":"a1b2c3d4-...","created_at":"2025-03-10T14:30:00+00:00","created_by":"user","forked_from_entity_id":null,"_table":"entities"}
{"entity_id":"a1b2c3d4-...","hash":"e1a2b3c4...","previous_hash":null,"recorded_at":"2025-03-10T14:30:01+00:00","_table":"hash_history"}

Export a specific table:

atlas export --table entities
atlas export --table hash_history
atlas export --table path_history
atlas export --table name_history
atlas export --table current_state
atlas export --table edges
atlas export --table file_traits

Scripting Examples

Count tracked files

atlas stats --json | jq '.active_entities'

List all current file paths

atlas export --table current_state | jq -r '.current_path'

Find files modified today

atlas log --today --json | jq '.entries[].name'

Check if daemon is running

atlas status --json | jq '.daemon.running'

Pipe duplicates to another tool

atlas find --duplicates --json | jq -r '.groups[].files[].path'

AI Agent Integration

Atlas can generate ready-to-use instruction blocks for AI agents:

atlas agent-setup --format claude    # CLAUDE.md instructions
atlas agent-setup --format mcp       # MCP server config (JSON)

For MCP-native integration, Atlas includes a full MCP server with 14 tools including aggregate queries and edge search. See Atlas Connect for details.

Integration Ideas

  • AI agents — Use atlas agent-setup to configure Claude, Cursor, Copilot, or any MCP client
  • CI/CD — Use atlas verify --json to check database integrity as part of a pipeline
  • Monitoring — Pipe atlas log --json to a log aggregator
  • Backup validation — Use atlas compare ~/project/ /backup/project/ --json to verify backups, or compare atlas export snapshots to detect drift
  • Custom dashboards — Query atlas stats --json and atlas log --summary --json for metrics
  • Duplicate cleanup scripts — Use atlas find --duplicates --json to feed a cleanup tool