agentleFS
NVIDIAKnown publisherNVIDIA/NVFlare/research/auto-fl-research/AGENTS.md

NVFlare / auto-fl-research

AGENTS.md1k starsChanged 12 days agoApache-2.0View on GitHub

Start with program.md. It is the general research-org entry point for the agent. Then read the active task profile. If the human does not specify one, use tasks/cifar10/profile.md. Use the active task's mutation_schema.yaml only when program.md or the active task profile directs you to the hard mutation bounds, or when choosing a mutation axis. If anything here conflicts with program.md, follow program.md unless a human explicitly overrides it. If a task profile conflicts with generic program.md text on task budget,…

Trending

Less talk lately.about the whole repository

Dec 2025: 0 mentionsJan 2026: 1 mentionFeb 2026: 0 mentionsMar 2026: 0 mentionsApr 2026: 0 mentionsMay 2026: 1 mentionJun 2026: 3 mentionsJul 2026: 2 mentionsAug 2026: 0 mentionsSep 2026: 0 mentionsOct 2026: 3 mentionsNov 2026: 0 mentions
Dec 2025Jun 2026Nov 2026
Mentions a month of the NVIDIA/NVFlare repository that our weekly crawl found. There is no adoption data for this file yet.

Who's talking about NVIDIA/NVFlare

The biggest 5 of 25 mentions linking to the repository that our crawl found. Each line is their own headline.

See all 25

Talked about on GitHub (25 times).

On GitHub · 25

20 more on GitHub

The file

We split this AGENTS.md into the rules it gives an agent

4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Start with program.md. It is the general research-org entry point for the agent.

    line 5
  2. Distinctive:

    Do not change unless explicitly requested:

    line 23
  3. Very specific:

    Run the active task profile's validation command, with TASK_DIR set to the active task.

    line 43
  4. Very specific:

    Run the active task profile's smoke command.

    line 44

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

AGENTS.md

Entry point

Start with program.md. It is the general research-org entry point for the agent. Then read the active task profile. If the human does not specify one, use tasks/cifar10/profile.md.

Use the active task's mutation_schema.yaml only when program.md or the active task profile directs you to the hard mutation bounds, or when choosing a mutation axis. If anything here conflicts with program.md, follow program.md unless a human explicitly overrides it. If a task profile conflicts with generic program.md text on task budget, environment, metric, or edit surface, follow the task profile.

Mission

Improve this Auto-FL NVFlare harness without breaking the federated contract.

Files you may edit

Preferred mutation files:

  • task-local client.py
  • shared tasks/shared/custom_aggregators.py
  • task-local job.py
  • task-local model.py for registered architecture variants under the active parameter cap

Do not change unless explicitly requested:

  • shared data/* or task-local data bridge files

Hard invariants

You must preserve all of the following unless a human explicitly asks for a protocol upgrade:

  • flare.init()
  • while flare.is_running():
  • input_model = flare.receive()
  • flare.send(output_model)
  • model.load_state_dict(input_model.params, strict=True)
  • compute_model_diff(...)
  • output_model.params_type == ParamsType.DIFF
  • output_model.meta["NUM_STEPS_CURRENT_ROUND"]
  • the optional if flare.is_evaluate(): branch
  • the same selected model_arch must be instantiated on server and clients for a run
  • model_arch and max_model_params are fixed budget fields unless the campaign is explicitly labeled as an architecture subcampaign

Required workflow after every edit

  1. Run the active task profile's validation command, with TASK_DIR set to the active task.
  2. Run the active task profile's smoke command. For non-CIFAR tasks, pass the task-specific SMOKE_ARGS or run bash scripts/run_iteration.sh ... with the active task budget.
  3. Record the mutation in results.tsv
  4. Summarize the mutation in templates/mutation_report.md

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.