octopus-platform-admin-cli

Platform administrator CLI for BotWorks/Octopus.

Auth

octopus-platform-admin-cli auth login
octopus-platform-admin-cli whoami

Authentication uses syngy-account through @syngy/account-cli-auth with admin scope. Credentials are stored under .config/octopus-platform-admin-cli.

Profile Targets

Each profile can point to production, prerelease, or custom hosts.

octopus-platform-admin-cli profile target production
octopus-platform-admin-cli profile target prerelease
octopus-platform-admin-cli profile show --json

Built-in targets:

• production: https://api.autostaff.cn
• prerelease: https://api.pr.autostaff.cn

Platform Admin Commands

Platform admin commands are generated from host/types/admin/admin_swagger.json. They execute through the generated @syngy/api-client-host-admin TypeScript SDK.

octopus-platform-admin-cli help-json
octopus-platform-admin-cli teams-list --json
octopus-platform-admin-cli skills-scan-create --body-json '{"dryRun":true}' --json

Commands with path parameters expose matching required flags, for example --tenant-id.

Team Points

Use team points get/list to read the business available points from Host. The response includes both ledger balance and pending Host usage, so it matches the Portal billing page and the metered LLM outbound gate.

Use team points topup for platform manual topups. Topups still write through the BotWorks Account admin API; Host remains the only read path for business availability.

octopus-platform-admin-cli team points get --team-id <team-id> --json
octopus-platform-admin-cli team points list --json
octopus-platform-admin-cli team points topup --team-id <team-id> --points 1000 --idempotency-key <key> --json

team wallet * commands are removed because wallet balance is a ledger implementation detail, not the business available points.

LLM Model Catalog

Use llm-models-sync-models-dev to sync selected model metadata and prices from https://models.dev/api.json into the Host LLM model catalog.

octopus-platform-admin-cli llm-models-sync-models-dev \
  --models moonshotai/kimi-k2.7-code-highspeed,zhipuai/glm-5.2 \
  --dry-run \
  --json

octopus-platform-admin-cli llm-models-sync-models-dev \
  --models moonshotai/kimi-k2.7-code-highspeed,zhipuai/glm-5.2 \
  --json

The command prefers the live models.dev API endpoint by default. Use --dry-run to inspect the exact batch-upsert body before writing, and use --source-url only for controlled tests or a mirrored catalog. It upserts model catalog fields such as provider, model name, thinking capability, modalities, and token prices; provider credentials, transport policy, digital worker binding, and tenant employee assignment are managed by their own admin commands.

Release Ops Command Tree

The release operations surface uses workflow-oriented commands instead of raw swagger operation names.

octopus-platform-admin-cli release status --json
octopus-platform-admin-cli release revisions list --json
octopus-platform-admin-cli release trigger team-pods-preview <revision-id> --dry-run --json
octopus-platform-admin-cli release trigger team-pods-preview <revision-id> --confirm-revision <revision-id>
octopus-platform-admin-cli release runs list --json

release trigger * invokes GitHub CLI gh workflow run. Real dispatch requires --confirm-revision matching the positional revision ID. Use --dry-run to inspect the workflow and arguments without dispatching.

Cluster and runtime inspection commands use the generated Host admin SDK:

octopus-platform-admin-cli cluster list --json
octopus-platform-admin-cli team-pods status
octopus-platform-admin-cli team-pods status --cluster octopus-01
octopus-platform-admin-cli team-pods status --json
octopus-platform-admin-cli cluster pods distribution --cluster-key <cluster-key> --json
octopus-platform-admin-cli runtime fleet show --cluster-key <cluster-key> --json
octopus-platform-admin-cli runtime rollout status --cluster-key <cluster-key> --json
octopus-platform-admin-cli runtime team expedite --team-id <team-id> --cluster-key <cluster-key> --body-json '{"reason":"queue jump"}' --json

Use team-pods status as the primary team pod status command. It accepts either a cluster key or cluster name, defaults to active clusters, and prints a human-readable summary with latest revision, dev/preview/prod team pod versions, rollout progress, blockers, expedited teams, and pod distribution. Use --json when an agent needs stable structured fields. Do not use raw cluster pods commands unless you are drilling into pod-level details.

runtime team expedite is a queue-jump operation. It does not take a version argument; the backend snapshots the current fleet desired version.

Conversation Debugging

Use platform-admin conversation commands for global production investigation across teams and clusters:

octopus-platform-admin-cli conversation show --conversation-id <conversationId> --json
octopus-platform-admin-cli conversation detail --conversation-id <conversationId> --debug --json
octopus-platform-admin-cli conversation watch --conversation-id <conversationId> --follow

Use octopus-cli conversation start/watch instead for team-user harness runs. Do not exec into a runtime pod and run octopus conversation ... for manual testing or investigation.