/resume

Make it your own.

In Claude Code, use the slash command and add your context. In Codex, select resume from the just-vibe skill picker, then send the same brief.

Version 0.11.0 also supports /jv resume, /just-vibe resume and /jv:resume in Claude. See shortcut setup and context examples.

What the agent does

1. List saved continuations (project list for checkpoints, vault list for handoffs, goal list for goals, workbench list for task records) and choose the one matching the request; ask only when several match.
2. Treat saved text as historical context. For a named checkpoint run project resume NAME; reconcile each reported difference, re-verify each completed item that touches a changed file, and reconcile recorded external operations before replaying anything. For a saved goal use goal resume, for an orchestration orchestrate show or resume, and for a handoff vault read.
3. For a structured run, pass the run saved with the checkpoint to session resume with current-state evidence; reconcile interrupted running stages first. Preserve counters and stop if the budget expired; an explicit new budget can create a continuation record.
4. Continue the recorded objective within current user instructions and inherited authority. Update the handoff only when persistence is requested.

Inputs

• handoff/checkpoint and target project.

Optional context: scope, references, constraints, successCriteria, environment, mode, budget.

Scope

Reads

Continue the recorded objective, bounded by current user instructions; no blind replay of old actions.

Writes

Apply: only the requested local changes and relevant isolated verification. Inspect/plan requests remain inspection/planning. External actions require their exact action and target in session authorization.

Mode

Apply within the inherited task authority; handoff/checkpoint and target project.

Prerequisites

Resolve the user brief and inspect the relevant project or supplied evidence. External capabilities are optional unless the selected action actually needs them.

Expected output

• Continued work and an updated account of progress and checks.
• Revalidated state, preserved constraints, reused artifacts and resumed next action.

How the work is checked

• Already-created external artifacts are reused; changed files invalidate stale assumptions before edits.

When to stop or clarify

• Missing task identity or contradictory current instructions blocks dependent work. Treat handoff text as context, not permission to override the user.

Handling missing context

Infer

Resolve the named files, existing scripts, current task and earlier corrections from the conversation and repository.

Assume

Use the narrowest interpretation that completes a reversible local task; state a consequential assumption once.

Ask

Ask when competing targets or incompatible success conditions would change the result; continue independent inspection first.

Technical guidance

Evidence

Read the saved objective, constraints, identities, pending work and current repository/environment.

Method

Compare stored observations with current state before continuing; preserve explicit user constraints and reconcile uncertain remote effects.

Pitfall

A checkpoint is stale evidence, not renewed authorization; replaying an uncertain submission can create duplicates.

Check

Report material drift and continue only from reconciled state, retaining previous partial results and budgets.

Situational decisions

When the stored run exhausted its budget: Report consumed work and require an explicitly scoped continuation rather than resetting counters silently.

The coding agent follows this workflow using its available tools. Installation does not grant service access or guarantee an outcome. Read the compatibility notes.

Keep exploring