agentleFS
NVIDIAKnown publisherNVIDIA/aistore/AGENTS.md

aistore

AGENTS.md1.9k starsChanged 16 days agoMITView on GitHub

Apply to these directories and all their subdirectories: Before changing, building, or testing CLI/iShard, read CLI or iShard instructions. Elsewhere, check for local instructions. Flag conflicts and cite the relevant document before work. Consult these references as needed for the task: Run from the repository root in Bash, using the Go version in go.mod and Makefile targets. Choose commands for the task. Remaining prompts default to no cloud backends, loopback devices, or tracing. Use make kill before redeploying, or make…

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

9 rules. About 0 are common, 6 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    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. Distinctive:

    Before changing, building, or testing CLI/iShard, read CLI or iShard instructions.

    line 32
  2. Very specific:

    Run from the repository root in Bash, using the Go version in go.mod and Makefile targets.

    line 46
  3. Very specific:

    Use make kill before redeploying, or make restart with existing configs.

    line 56
  4. Distinctive:

    Use existing test infrastructure for behavior and regressions.

    line 60
  5. Very specific:

    Run formatting and spelling checks.

    line 79
  6. Distinctive:

    Before using Delve, read debugging.md.

    line 88
  7. Distinctive:

    Read implementations and callers for each codepath's contracts and lifetimes.

    line 95
  8. Distinctive:

    Do not redesign shared contracts around a single use.

    line 97
  9. Distinctive:

    Before committing, follow Commit Messages and contributor requirements for human authorship, sign-off, and disclosure.

    line 103

≈ “Common” is an estimate: at least 1 in 5 of the 36,368 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

Core AIStore Agent Instructions

Scope and References

Apply to these directories and all their subdirectories:

ais/        proxy/target nodes, integration tests
api/        Go client API/SDK
cmd/        node, CLI, utility executables
cmn/        shared configuration, types, utilities
core/       object/cluster metadata, internal APIs
ec/         erasure coding
ext/        ETL, dSort, downloader
fs/         mountpaths, object file layout
hk/         periodic housekeeping callbacks
ios/        disk I/O stats, OS storage interfaces
memsys/     slab/SGL memory management
mirror/     local object replicas
nl/         progress/completion notifications
reb/        cluster-wide rebalance
res/        resilvering across local mountpaths
space/      LRU eviction, storage cleanup
stats/      metrics tracking and export
sys/        CPU, memory, load, processes
tracing/    OpenTelemetry distributed tracing
transport/  intra-cluster object streaming
volume/     volume configuration and metadata
xact/       background jobs, registry, implementations

Before changing, building, or testing CLI/iShard, read CLI or iShard instructions. Elsewhere, check for local instructions. Flag conflicts and cite the relevant document before work.

Consult these references as needed for the task:

  • CONTRIBUTING.md: Human responsibility, disclosure, licensing, and DCO.
  • overview.md: Architecture, design, and traffic patterns.
  • terminology.md: Core abstractions and vocabulary.
  • getting_started.md: Local cluster setup.
  • development.md: Debugging, scripts, and performance comparisons.

Build and Verification

Run from the repository root in Bash, using the Go version in go.mod and Makefile targets. Choose commands for the task.

# Local playground: 3 targets, 3 proxies, 4 mountpaths per target.
MODE=debug make cli deploy <<< $'3\n3\n4'

Remaining prompts default to no cloud backends, loopback devices, or tracing.

Use make kill before redeploying, or make restart with existing configs. Repeat required MODE, TAGS, and GORACE settings on rebuilds/restarts.

Validate affected behavior against a running local AIStore cluster. Run focused tests for the changed subsystem; CI provides broader regression coverage. Use existing test infrastructure for behavior and regressions; remove temporary validation tests that have no upstream.

Replace TestName with a relevant test and confirm it runs; no matches is not verification. Set AIS_ENDPOINT and BUCKET for the test cluster. Match deployment TAGS and test TEST_TAGS to changed code; omit unused tags.

AIS_ENDPOINT=http://localhost:8080 BUCKET=ais://tmp \
  TESTS_DIR=ais/test RE='^TestName$' make test-run

For concurrency changes, enable race detection in both daemons (GORACE) and affected tests (TEST_RACE=true for make test-run).

For performance-sensitive changes, measure the affected path against a baseline under comparable conditions. Use relevant microbenchmarks for isolated code and aisloader for cluster workloads. Confirm measurements are produced.

Run formatting and spelling checks; for core Go changes, also gofmt changed files and lint. Install a missing linter with make lint-update-ci (CI version). Activate a Python virtual environment before make fmt-check, which installs Ruff.

make fmt-check spell-check
make lint-scoped

Before using Delve, read debugging.md. Track checks, results, bug reproduction, and incomplete verification for human review; the human contributor owns claims and disclosure.

Engineering Contracts

  • Preserve user data. Test changes to integrity, persistence, and transport.
  • Read implementations and callers for each codepath's contracts and lifetimes. Preserve them; update all affected callers when changing a contract. Do not redesign shared contracts around a single use.
  • Preserve API compatibility. Change APIs only when explicitly instructed; follow maintainer decisions on deprecation.
  • Prioritize correctness over performance. Understand synchronization primitives and their responsibilities before changing concurrent code or critical sections.
  • Follow Coding Style.
  • Before committing, follow Commit Messages and contributor requirements for human authorship, sign-off, and disclosure.

External Sources

AIStore is MIT-licensed. Agents must not copy or adapt third-party code, regardless of license, without copyright-holder permission and explicit maintainer approval. Consult external documentation for interfaces and behavior; implement using AIStore's existing patterns and contracts.

Download now (ZIP)

Trending

Steady: 0 new projects took it in the last 4 weeks, none the 4 weeks before.thin data

Week of Aug 30: 0 new projectsWeek of Sep 6: 0 new projectsWeek of Sep 13: 0 new projectsWeek of Sep 20: 0 new projectsWeek of Sep 27: 0 new projectsWeek of Oct 4: 0 new projectsWeek of Oct 11: 0 new projectsWeek of Oct 18: 0 new projects
Aug 2026Sep 2026Oct 2026
New projects taking a version of this file each week, over the 8 weeks our latest run read: only 0 in all, so this is a hint, not a trend.

Who's talking about NVIDIA/aistore

The biggest 5 of 28 mentions linking to the repository that our crawl found. Each line is their own headline.

See all 28

Talked about on GitHub and Bluesky (28 times).

On GitHub · 27

22 more on GitHub
  • Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.

    Of the AGENTS.md files we read, the 28,693 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.

  • In this repository3 other files it gives its agents.

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.