sharing
Skill4 starsChanged 54 days agoMITView on GitHub
Prime on Eidet's layers and packs before changing them — MemoryLayer types and the Local/Shared/Base priority ladder, LayerService mounting and applicability, the immutable LayerScope handed to recall, MarkdownPackFormat's YAML+markdown round-trip and its import provenance clamp, LayerSyncService and ILayerSource, and the AGENTS.md export. Use when the task touches pack export/import, mounting or unmounting a layer, cross-repo scope, the .eidet markdown format, or ExportService. Not for seeding from a repo's own files (see intake), not for backup/restore or remote sync.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- BuilderIO/agent-nativetheir own version, first seen Oct 8, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- BuilderIO/agent-native★ 7.7ksame name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository12 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- canonSkill
- enrichmentSkill
- intakeSkill
- looseendsSkill
- maintenanceSkill
- memorySkill
- memorytoolSkill
- portalSkill
- qualitySkill
- recallSkill
- writepathSkill
The file
Files
Sharing (layers & packs) — priming
Canonical spec: docs/domains/sharing.md — read it for the layer model, the pack format, all
invariants, key files, and gotchas. Terms of record: UBIQUITOUS_LANGUAGE.md § Namespacing & layers,
Sharing. Threat: STRIDE.md T-7.
Layers stack namespaces (Local rw on top; Shared/Base ro below) and resolve into the recall scope. Packs are the human-readable transport that auto-mounts as a layer on import.
Core invariants (get these right)
- Writes always land in the Local layer — Shared/Base contribute to recall but never accept writes.
- Clamp declared provenance on import. A poisoned pack can write
provenance=userStated; anything above thePacktrust floor is clamped back down (T-7). Lower-or-equal origins pass through. Unknownprovenance never crosses the wire — omit it on export so a foreign install applies its own default and the clamp holds it at the pack floor.LayerScopeis resolved once at the boundary and is immutable; the read pipeline never learns about mounting.- Applicability has three routes: universal (empty
ApplicableRepos), explicit repo, or package-dependency match (the auto-mount path). - The pack format is a published contract (ScribeGate + plain markdown viewers): every field must
round-trip and defaults are omitted. A new
MemoryEntryfield needs a wire decision. - Mounting is idempotent, and a legacy
bundle:layer id is reused rather than forked. - Loose Ends and Memory files are never exported.
Key files / reuse
src/Eidet.Core/Services/LayerService.cs— mount/applicability/scope.src/Eidet.Core/Layers/LayerScope.cs— the scope snapshot +NonLocalDeBoost.src/Eidet.Core/Services/MarkdownPackFormat.cs— the format and the import clamp.src/Eidet.Core/Layers/ILayerSource.cs— add a transport here (fileis the only scheme today).
Gotchas
GetMountedLayersAsync("")is an intentional "all layers" wildcard, not a bug.- Cross-repo recall is off by default — a layer that "isn't working" is usually a
CrossRepo: falsecall. AutoMountByDependenciesAsyncmutates the layer (appending repo ids), not the repo.- Pack import mints ids under the pack id, not the importing repo — imports are a separate namespace.
- The de-boost is a multiplier, not a filter.
AGENTS.mdexport is a one-way rendering; regenerate rather than hand-edit.
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
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.

