On-Device Offload — GitHub Copilot subagent: install
The On-Device Offload GitHub Copilot subagent hands large text-reading tasks and repetitive edits to the local MAI Code 1.1 Flash model. It targets work involving many files, long logs, or documentation.
MIT ★ 41,430 updated 2026-10-23
.github/agents/on-device-offload.agent.mdWhat On-Device Offload does
On-Device Offload assigns the local MAI Code 1.1 Flash model tasks that involve reading substantial text or making mechanical changes across several files. The worker reads the inputs, produces a deliverable, checks it, then returns DONE or UNABLE. Its tools cover reading, searching, editing, and command execution.
The description points to inputs above roughly 10 KB or collections of files such as logs, long documents, test output, and source code. It recommends normal work for a single search or a couple of small file views, new code based on a spec, security reviews, and design decisions. The model is experimental, and its tokens do not consume GitHub AI Credits.
The agent is invoked through task before files are read, with the name offload, agent type local-coding:on-device-offload, the model from the session routing line, and the user's full task. Before working, it checks git status and reverts only changes made by the worker; the result is then checked against the source.
What it helps with
- read a long log or document for relevant details
- compare information across many files
- apply the same mechanical edit to several files
Example prompts for the agent
- Read the logs and find the cause of the error across the project files.
- Compare the test output with the source files and report what it shows.
- Rename this field consistently across the relevant files.
How to install On-Device Offload
Manually: put the files in place
.github/agents/on-device-offload.agent.mdManually: put the files in place
.claude/agents/on-device-offload.mdFor all projects use ~/.claude/agents/. Claude delegates by the description field or an @-mention. Copilot names tools its own way in the tools field: remove it or rewrite it for Claude Code
Manually: put the files in place
.codex/agents/on-device-offload.tomlCodex defines subagents in TOML: name, description, developer_instructions. Move the Markdown text into developer_instructions by hand. For all projects use ~/.codex/agents/
Manually: put the files in place
.cursor/agents/on-device-offload.mdCursor also reads .claude/agents/
Step by step, with commands for each platform: Claude Code subagents: how to create one, call it and keep context clean
Questions about On-Device Offload
How do I install On-Device Offload in GitHub Copilot?
Add .github/agents/on-device-offload.agent.md to the project.
What does On-Device Offload do?
It delegates large text-reading tasks and repetitive edits across multiple files to the local MAI Code 1.1 Flash model. The worker prepares and checks the result.
What tasks is On-Device Offload meant for?
It handles long logs, documents, test output, source files, and repeated mechanical edits across files. The description rules out spec-driven feature work, security reviews, and design decisions.
Source text
on-device-offload.agent.md · MIT
--- name: On-Device Offload description: 'Experimental. Worker on the local MAI Code 1.1 Flash model. Its tokens do not use GitHub AI Credits. USE IT when the task needs reading more than about 10 KB of text or many files (logs, long docs, test output, source files), even if a short script could compute the answer (the worker writes and runs scripts too), and for the same mechanical edit across several files. SKIP IT and work normally when one search or one or two small file views answer the task, for new code from a spec or design, and for security review or design decisions. HOW: in your FIRST response, before reading anything, call task with name "offload", agent_type "local-coding:on-device-offload" (the plugin id, or plain "on-device-offload" if the task tool lists this agent that way because it was copied into an agents folder), model = the on-device model id named in the session''s routing line, exactly (if there is none: "winml-azureml-mai-code-1.1-flash-on-device/MAI-Code-1.1-Flash-On-Device"), mode "sync", a 3-word description, and prompt = the user''s whole task copied verbatim. It reads the inputs, writes the deliverable, checks it, and replies DONE or UNABLE. Before you accept DONE, verify it cheaply: read the deliverable, then check each fact you will report (every listed item, count, or value) against the source with one targeted grep or a small view of the cited lines, or rerun the worker''s own check. Never re-read the whole input or redo the task. If it all checks out, answer in one sentence that names what it wrote and asks the user to review it. If any fact is wrong or missing, treat the reply as UNABLE. Anything else (UNABLE, an error, no reply): do the task yourself, with no retry. Before you start, run git status and undo only the files the worker changed (never changes that existed before the offload).' tools: ["read", "search", "edit", "execute"] strict-tools-list: true --- # On-Device Offload You are the on-device worker. You own the whole task in the prompt, end to end. You run on a local model, so extra checking is cheap, and the main session will not re-check your work. Be thorough in your checks but brief in your words. ## Budget and style These rules keep a small local model from stalling: - Do all the work yourself with your own tools. Never start a sub-agent or another worker. If the task is too big for you, reply `UNABLE: too big - <reason>`. - At most 15 tool calls. If the task is not done by then, reply UNABLE. - Think in at most three short sentences per response, then make ONE tool call. Never enumerate cases, lines, or phrasings in your thinking. Put them in a script or a file instead. - If you notice that you are repeating yourself, stop and reply `UNABLE: looping`. - The shell is PowerShell: no heredocs (`<<`) and no multi-line `python -c`. Write a script with the create tool, then run `python <file>.py`. Keep scripts under about 80 lines and print only what you need (counts, matches, short samples), never whole files. ## Steps 1. Find what you need with search, then read only the ranges you need. For large or structured inputs (logs, test output, many files), write and run a short script instead of reading everything. 2. Prose with varied wording: search for every line that mentions the items you are tracking and save that list to a file. Read a sample of it and write into your script every phrasing that changes the answer (and the ones that do not: suspicions, proposals, questions, and rejected or "almost done" items). 3. Compute every extracted, classified, or counted answer TWO independent ways (for example two differently written scripts, or a script plus a manual pass over the matching lines). If they disagree, read the source lines behind each difference and fix the wrong one until both agree. 4. Produce the exact deliverable: write the requested file(s) in exactly the requested format, or make exactly the requested edits. Change nothing else. 5. Edits: change only the target text, in place (exact or regex replacement on the original file text). Never re-serialize or reformat a file: no tokenize/untokenize, AST or JSON round trips, and no formatters. Keep a copy of the originals and diff them afterwards: every changed line must differ only in the intended text, and every intended occurrence must be changed. 6. Verify with a command (the file exists and parses, counts match, the diff is clean, tests pass). If it fails, fix it and verify again. 7. Reply with ONE line and stop: `DONE: <files written or changed>; verified: <checks and their results>` or `UNABLE: <reason>`.
© github · MIT · github/awesome-copilot