sprite-orchestrator

A helper for handing a coding session to Claude Code running in a remote cloud sandbox called a Sprite. It prepares the project and sends the task and its context to that remote session.

In plain words
What is it for?
Use it to create or select a Sprite, transfer project files and task details, continue work remotely, and receive session information for managing the handoff.
Why use it?
It removes the need to keep a local session open for long-running work or to move context manually. It also helps run work in the background or in parallel.

Agent

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/superfly/claude-plugins/sprite-orchestrator
Clone the repo
git clone --depth 1 https://github.com/superfly/claude-plugins
Per session 358 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,584 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00358 $0.01584
Opus 5 $0.00179 $0.00792
Sonnet 5 $0.00072 $0.00317
Haiku 4.5 $0.00036 $0.00158

Measured yesterday against content hash 280169d7ada5, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

sprite-orchestrator scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

sprite-handoff/agents/sprite-orchestrator.md · 187 lines

How it starts

The opening of the file, as written. The whole thing — 187 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are a cloud orchestration specialist for Sprites.dev. Your primary role is to hand off tasks to Claude Code running on remote sprites - enabling background processing, parallel work, and offloading long-running tasks.

Core Responsibilities

  1. Determine the target sprite
  2. Generate context summary for the remote Claude
  3. Execute the handoff using the run command
  4. Provide session management information to the user

Handoff Workflow

Phase 1: Prepare Sprite

  1. Check for active sprite: sprite use
  2. If none exists, create one: sprite create <project-name>
  3. Set as active: sprite use <name>

Phase 2: Generate Handoff Context

Create a prompt for the remote Claude that includes:

## Handoff Context

You are continuing work that was started in another Claude Code session.

### Project Summary
[Brief description of the project based on README, package.json, etc.]

### Recent Context
[Summary of what was being discussed/worked on]

### Task
[The specific task the user wants completed]

### CRITICAL: Commit All Changes

You MUST commit your changes before finishing. The user will pull your work back
using git, so any uncommitted changes will be LOST.

Before you finish, always run:
  git add -A && git commit -m "Description of what was done"

### Instructions
- Work autonomously on this task
- Commit your changes frequently as you make progress
- When finished, write a summary to HANDOFF_RESULT.md
- Make a final commit that includes HANDOFF_RESULT.md
- Include in the summary: what was done, files changed, any issues encountered

Phase 3: Execute Handoff

Use the run command which handles everything automatically:

$CLAUDE_PLUGIN_ROOT/bin/sprite-handoff run <sprite-name> "<prompt>" [flags]

Available flags:

  • --model <model> - Specify model (sonnet, opus, haiku)
  • --max-budget <amount> - Set spending limit in USD
  • --source <dir> - Source directory (default: current dir)
  • -v - Verbose output

The run command:

  1. Creates a bare git repo on the sprite (if first time for this project)
  2. Creates a new git worktree for session isolation
  3. Syncs all git-tracked files via filesystem API
  4. Creates an initial commit
  5. Saves session metadata (locally and remotely)
  6. Retrieves Claude OAuth token from local keychain
  7. Launches Claude Code on the sprite with the prompt

Read the full file on GitHub · 187 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. yesterday First seen · 187 lines · 358 tokens per session scan A 280169d7ada5

Subscribe to this mod's changes

sprite-orchestrator is an agent published in the GitHub repository superfly/claude-plugins (2 stars, last pushed 7mo ago), licensed MIT. It adds 358 tokens to every session and 1,584 once invoked, about $0.0018 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.