sediment v0.1.0 — Turn AI developer workflow traces into RL-ready training data.

USAGE:
  sediment <command> [options]

  Sediment CLI: login/logout, remote facts + commit, quarantine, exports,
  reports, mirror GC.

COMMANDS:
  report                read-only reports (sediment report --help)
  delivery              prepared capture delivery (sediment delivery --help)
  mirror-gc             remove mirrors past push retention (dry-run; --apply
                        deletes)
  facts                 fact counts per table (remote; --database-url for
                        direct)
  demo                  plant one synthetic session so facts is non-zero
  db                    provision database roles or inspect and upgrade the
                        schema
  quarantine            quarantine one fact by id
  release               release one quarantined fact by id
  quarantine-log        the org's quarantine history
  quarantine-inference-calls
                        bulk-quarantine inference calls by filter (dry-run by
                        default)
  install               wire a repo + this machine: git hooks, agent hooks,
                        telemetry env (sediment install --help)
  uninstall             remove the per-repo git hooks (--agents: user-level too)
  doctor                check attribution + server health on this machine
  server                run a local API server with managed PostgreSQL
  login                 store credentials for a server
  logout                remove stored credentials
  commit                pretty-print attributions for a commit
  derive                materialize a reviewed canonical derivation bundle
  export                run an export projection

OPTIONS:
  -h, --help            show this help message and exit
  --version             show program's version number and exit
