CLI Reference

The db-cluster CLI is the operator surface. Every command is documented via --help; this page is the high-level map.

Command groups

# Cluster lifecycle
db-cluster init                          # initialize .db-cluster/
db-cluster doctor [--json]               # health assessment
db-cluster verify [--json] [--sample N]  # data integrity proofs

# Artifact / entity primitives
db-cluster ingest <path>                                     # ingest an artifact
db-cluster entity create --kind <K> --name <N> [--attr <J>]  # create an entity
db-cluster link --entity-id <E> --artifact-id <A>            # link evidence

# Discovery / retrieval
db-cluster find <query>                  # candidate index records
db-cluster inspect <id>                  # entity / artifact / receipt details
db-cluster retrieve <query>              # evidence bundle with confidence
db-cluster explain-retrieval <query>     # what the bundle can / cannot claim
db-cluster resolve <uri>                 # owner truth from any cluster:// URI

# Provenance
db-cluster trace <uri> [--direction] [--depth] [--graph]
db-cluster why <uri>
db-cluster lineage <uri>                 # bidirectional full trace
db-cluster trace-bundle <query>          # retrieve + trace
db-cluster explain-index <record-id>     # why this index record exists

# Mutation lifecycle
db-cluster propose <command-json>
db-cluster validate <command-id>
db-cluster approve <command-id> [--note]
db-cluster reject <command-id> --reason
db-cluster commit <command-id>
db-cluster compensate <command-id> --reason
db-cluster inspect-command <command-id>
db-cluster receipts [--limit N]

# Index maintenance
db-cluster index rebuild [--dry-run] [--yes]
db-cluster index status [--json]
db-cluster index stale
db-cluster index explain <record-id>

# Backup / restore
db-cluster backup [-o <file>] [--force-overwrite] [--yes]
db-cluster restore <file> [--yes]

# Policy
db-cluster policy explain --principal <json>
db-cluster policy test --principal <json> --verb <V> --resource <uri>

# Postgres canonical store
db-cluster stores verify
db-cluster stores migrate
db-cluster stores list
db-cluster migration-status [--json]
db-cluster verify-schema [--json]

# Help meta
db-cluster --help                        # top-level
db-cluster <cmd> --help                  # per-command
db-cluster --help-exit-codes             # the exit-code table

Global flags

Flag	Effect
`--actor <id>`	Operator identity for this invocation (overrides `DB_CLUSTER_OPERATOR` / OS user).
`--quiet`	Suppress STDOUT non-error output entirely. Errors still emit to STDERR.
`--log-level <level>`	`debug` / `info` / `warn` / `error`. Default: `info`.
`--no-color`	Disable ANSI color codes. `NO_COLOR` env var also honoured (https://no-color.org).
`--debug`	(env: `DEBUG=1`) — surface raw stack traces on STDERR.

Exit codes

db-cluster uses the sysexits.h convention. Codes are stable across versions — CI scripts can branch on them safely.

Exit	Sysexits	Typed-error codes mapped here
0	EX_OK	success
1	(general)	`NOT_FOUND`, `PROVENANCE_MISSING`, `COMMAND_NOT_VALIDATED`, `COMMAND_REJECTED`, usage error
65	EX_DATAERR	`CONTENT_HASH_MISMATCH`, `INVALID_CONTENT_HASH`, `STAGED_CONTENT_TAMPERED`, `IMPORT_CONFLICT`, `INVALID_CONTENT_SHAPE`
70	EX_SOFTWARE	`CORRUPT_STORE`, `COMMAND_QUEUE_CORRUPT`, `COMMAND_QUEUE_PERSISTENCE_LOST`, `LEDGER_CYCLE_DETECTED`, `RECEIPT_FAILED`, `BUFFER_SIDE_CHANNEL_NOT_SUPPORTED`
77	EX_NOPERM	`POLICY_DENIED`
78	EX_CONFIG	`INVALID_POLICY_CONFIG`, `INVALID_REDACTION_RULE`, `INVALID_ROTATE_TIMESTAMP`, `ROTATE_BOUNDARY_IN_FUTURE`

Run db-cluster --help-exit-codes to print the current table from the live CLI.

Branching example

if ! db-cluster verify --json > /tmp/verify.json; then
    case $? in
        65) echo "Data integrity issue — see /tmp/verify.json" ;;
        70) echo "Cluster corruption — run doctor + rebuild index" ;;
        77) echo "Policy denied — check DB_CLUSTER_PRINCIPAL" ;;
        78) echo "Config error — check policies file" ;;
        *)  echo "General failure — exit code $?" ;;
    esac
    exit 1
fi

Output discipline

CLI output is uniformly colored when running on a TTY without --no-color:

Errors (red) — Error: <message> prefix on every cliCommand catch arm.
Hints (dim italic) — the → try: <remediation> line that follows every typed error.
Success (green) — Cluster initialized, Ingested, Created entity, Linked, Committed.
Headers (bold cyan) — entity inspection headlines, doctor / verify section markers.

When piped (db-cluster doctor --json | jq), or when --no-color is set, or when NO_COLOR is in env, no ANSI bytes are emitted.

Environment variables

Variable	Effect
`DB_CLUSTER_DIR`	Override the default `.db-cluster/` location.
`DB_CLUSTER_PRINCIPAL`	JSON principal identity (schema-validated, fail-closed).
`DB_CLUSTER_POLICIES_FILE`	Path to a policies JSON file (path-sandboxed against cwd).
`DB_CLUSTER_OPERATOR`	Operator name for the actor field (used when `--actor` not passed).
`DB_CLUSTER_POSTGRES_URL`	Postgres connection string for the canonical store backend. For TLS, include `sslmode=require` in the string (the `pg` driver honours it); db-cluster does not configure SSL/TLS itself in v1.0.0.
`NO_COLOR`	Any non-empty value disables ANSI color (https://no-color.org).
`DEBUG`	`1` enables raw stack traces on STDERR.