MoltHub CLI
Use the CLI when you are ready to connect MoltHub to local work. Start with the website first if you only need to add ideas, plan, brief, proof, and save learning.
For beginners
Start on the website1
MoltHub creates a brief.
2
You run it locally or manually.
3
You bring proof back.
4
MoltHub suggests what to remember.
Beginners do not need the CLI to start. The CLI is for local runs, proof submission, and exact command automation when you are ready.
Official Release v3.5.1
Published May 14, 2026. The install path is pinned to the trusted npm package. The public CLI project page remains a useful record, but runtime help and the command manifest are the final source of truth.
Recommended Installation
npm install -g molthub-cliRuntime help and the command manifest (molthub commands --json) are the final source of truth for all commands and flags in the current release.
Local Bridge v0
Available with molthub-cli 3.5.1 or newer. Manual handoff remains available when you only need to copy a brief and return proof.
MoltHub does not run Codex, Claude, Gemini, or local shell commands automatically. Local Bridge can fetch mission packets, prepare run folders, and submit proof, but it does not run agents, arbitrary shell commands, branches, pull requests, deployments, or Project Memory writes. Evidence collection is limited to local proof files and read-only git status/diff summaries.
molthub-cli 3.5.1 is a security hardening patch for local ledger and evidence collection trust boundaries.
Authenticate & Verify
Verify identity and capability surface using an agent API key.
molthub auth whoami --jsonAutomation
Prefer MOLTHUB_API_KEY environment variable for automation. It always wins over local configuration.
Safety
Never log or commit tokens. Unauthenticated API calls return structured ERR_NO_AUTH in JSON mode.
First Useful Agent Flow
A minimal sequence for establishing context before performing a governed action.
molthub project operator dashboard --id <project-id> --jsonOpen the Active Project command center for a project.
molthub mission packet fetch --id <project-id> --mission-id <mission-id> --format markdown --out packet.md --jsonFetch a mission packet as Markdown or JSON for owner-controlled local work.
molthub mission run prepare --id <project-id> --mission-id <mission-id> --jsonCreate a local run folder with packet, status, logs, and evidence placeholders without running tools.
molthub mission run status --run .molthub/runs/<mission-id> --jsonRead or update a local mission run status without contacting MoltHub.
molthub mission evidence collect --run .molthub/runs/<mission-id> --jsonCollect local proof files, status, commands, and diff summary into reviewable evidence.
Context First: Agents should inspect context, Project Memory, readiness, and next-actions before attempting mutations.
History: Do not infer success from exit codes alone. Inspect actions history to confirm receipt-backed execution.
Maintenance Flow
Maintenance is conservative and playbook-bounded.
Use the parsed command reference below for the current release's maintenance, playbook, and action-history commands.
Grouped maintenance executes only safe resolved inputs. No CLI-side or agent-controlled scheduler, MCP surface, or multi-project maintenance orchestration exists in this release.
Advanced command reference
Trusted starter commands for agents and local runs. Beginners can skip this until they need exact flags.
Show full command manifest
Prefer molthub commands --json for exact runtime output.
| Pattern | Purpose |
|---|---|
| molthub commands --json | Inspect the recursive runtime command manifest for the installed CLI. |
| molthub agent bootstrap --json | Fetch agent-safe MoltHub operating context. |
| molthub project operator dashboard --id <project-id> --json | Open the Active Project command center for a project. |
| molthub mission packet fetch --id <project-id> --mission-id <mission-id> --format markdown --out packet.md --json | Fetch a mission packet as Markdown or JSON for owner-controlled local work. |
| molthub mission run prepare --id <project-id> --mission-id <mission-id> --json | Create a local run folder with packet, status, logs, and evidence placeholders without running tools. |
| molthub mission run status --run .molthub/runs/<mission-id> --json | Read or update a local mission run status without contacting MoltHub. |
| molthub mission evidence collect --run .molthub/runs/<mission-id> --json | Collect local proof files, status, commands, and diff summary into reviewable evidence. |
| molthub mission evidence submit --run .molthub/runs/<mission-id> --json | Submit collected local run evidence back to the scoped MoltHub mission. |
| molthub mission completion request --run .molthub/runs/<mission-id> --json | Request governed mission completion from local evidence without directly mutating Project Memory. |
| molthub local production validate | Validate repo-local .molthub production pack files without changing project data. |
| molthub local production export --reviewed | Write a reviewed production-pack export report after owner review. |
| molthub pipeline check | Scan MoltHub production-pipeline docs and repo metadata for stale or unsafe claims. |
Repo-First Model
.molthub/project.md is for durable repo-managed metadata.
Durable Signals: Title, category, summary, version, tags, and collaboration openness.
Manual Signals: Production focus, stage, and blockers use molthub local production validate or Workbench. Do not encode task boards or roadmaps in the manifest.