ghuntleyghuntley/loom/protocol/AGENTS.md

loom / protocol

AGENTS.md1.6k starsChanged todayMITView on GitHub

Follow root guidance. This dependency contains versioned wire validation and types shared by Marshall and Lilly, and the pure state-machine, cadence and single-writer in-memory view primitives every component shares (ADR-0140). ADR-0142 additionally places the outbound control WebSocket client in the named Control.Client and WebSocket.Connection modules. Keep codecs and state/value primitives pure; those transport modules may own sockets and use application-supplied credential callbacks and TLS trust. Keep the entire library free of Ash, database access, host mutations and durable credential storage.…

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

2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.

≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.

Common
Distinctive
Very specific

Distinctive:
Follow root guidance. This dependency contains versioned wire validation and types shared by Marshall and Lilly, and the pure …
line 3
Distinctive:
Never create atoms from peer input.
line 11

≈ “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

Protocol guidance

Follow root guidance. This dependency contains versioned wire validation and types shared by Marshall and Lilly, and the pure state-machine, cadence and single-writer in-memory view primitives every component shares (ADR-0140). ADR-0142 additionally places the outbound control WebSocket client in the named Control.Client and WebSocket.Connection modules. Keep codecs and state/value primitives pure; those transport modules may own sockets and use application-supplied credential callbacks and TLS trust. Keep the entire library free of Ash, database access, host mutations and durable credential storage. Never create atoms from peer input. Bound every collection and integer before passing data to either application.

Build //protocol:check //protocol:test and both applications' :check and :test (tools/precommit selects them) when this contract changes. Codecs accept exactly the current shape. Change a shared message by ordered, converged pushes (consumer tolerant, producer changes, decoder strict; ADR-0159), never by keeping a compatibility reader, and cover each step with protocol tests.

Download now (ZIP)

Jan 17, 2026Cross-Session Thread Loading & Replay (from Huntley Loom research) — daydemir/milhouse-go-legacy0 reactions
Jan 17, 2026Active System Verification Loops (from Huntley Loom research) — daydemir/milhouse-go-legacy0 reactions
Jan 17, 2026Meta-Loop Orchestration: Agents Spawning Sub-Loops (exploratory/major from Huntley Loom research) — daydemir/milhouse-go-legacy0 reactions
Jan 17, 2026Structured Audit Threads for Agent Decision Trails (from Huntley Loom research) — daydemir/milhouse-go-legacy0 reactions
Jan 17, 2026Server-Side LLM Proxy for Credential Isolation (from Huntley Loom research) — daydemir/milhouse-go-legacy0 reactions

8 more on GitHub

Jan 17, 2026Secret Wrapper Type for Auto-Redacting (from Huntley Loom research) — daydemir/milhouse-go-legacy0 reactions
Jan 17, 2026Error Classification with Transient/Quota/Permanent (from Huntley Loom research) — daydemir/milhouse-go-legacy0 reactions
Jan 17, 2026Thread Persistence with UUID7 & Offline-First Sync (from Huntley Loom research) — daydemir/milhouse-go-legacy0 reactions
Jan 17, 2026Post-Tool Hook State for Auto-Commit (from Huntley Loom research) — daydemir/milhouse-go-legacy0 reactions
Jan 17, 2026Snippet-Based File Editing (from Huntley Loom research) — daydemir/milhouse-go-legacy0 reactions
Jan 17, 2026Proactive OAuth Token Refresh (from Huntley Loom research) — daydemir/milhouse-go-legacy0 reactions
Jan 17, 2026Multi-Account LLM Pool with Failover & Cooling (from Huntley Loom research) — daydemir/milhouse-go-legacy0 reactions
Jan 17, 2026Explicit State Machine for Agent Orchestration (from Huntley Loom research) — daydemir/milhouse-go-legacy0 reactions

On Hacker News · 1

Jan 16, 2026If your name is not Geoffrey Huntley then do not use loom2 points · 1 comment

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 repository31 other files it gives its agents.
- AGENTS.mdAGENTS.md
- antithesis/AGENTS.mdAGENTS.md
- lilly/AGENTS.mdAGENTS.md
- marshall/AGENTS.mdAGENTS.md
- mila/AGENTS.mdAGENTS.md
- susan/AGENTS.mdAGENTS.md
- adr-maintenanceSkill
- antithesis-agent-browserSkill
- antithesis-debugSkill
- antithesis-documentationSkill
- antithesis-feature-workloadSkill
- antithesis-launchSkill
- antithesis-mutation-testingSkill
- antithesis-query-logsSkill
- antithesis-researchSkill
- antithesis-review-inputsSkill
- antithesis-setup-k8sSkill
- antithesis-setupSkill
- antithesis-skills-feedbackSkill
- antithesis-triageSkill
- antithesis-workloadSkill
- architecture-reviewSkill
- buck2Skill
- code-contractsSkill
- covenantSkill
- hegel-elixir-testsSkill
- localizationSkill
- monospace-webSkill
- nix-buildsSkill
- security-auditSkill
- test:triageSkill

Does it work?

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

Reviews can't be read right now.

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.

Check your setup, free