agentos-artifact-fs
Select, prepare, operate, and retire ArtifactFS-backed Scout workspaces for fast read-heavy access to large or multiple Git repositories. Use when a First or Second Mate is delegating a scout investigation whose clone startup or cross-repository breadth justifies an isolated FUSE-enabled Crewmate profile; never use it for persistent Mates or ship work by default.
- License 许可
- MIT
- 所属仓库
- akua-dev/agentos
- 最近更新
- 2026-11-09
- Stars
- ★ 8
- Forks
- 0
01 前言 Front Matter
- name
- agentos-artifact-fs
- description
- Select, prepare, operate, and retire ArtifactFS-backed Scout workspaces for fast read-heavy access to large or multiple Git repositories. Use when a First or Second Mate is delegating a scout investigation whose clone startup or cross-repository breadth justifies an isolated FUSE-enabled Crewmate profile; never use it for persistent Mates or ship work by default.
02 资源文件 1
- 参考文档agents/openai.yaml210 B
03 SKILL.md 正文
Explore repositories with ArtifactFS
Use ArtifactFS only as an optional Scout workspace. Keep First Mate, Second Mate, ship Crewmates and ordinary scouts on native Git unless measured repository size or multi-repository breadth makes lazy hydration materially useful.
Select the profile
- Confirm the Assignment is
scoutand its durable output is a report, not a project change. Load$agentos-delegationbefore provisioning the worker. - Inspect repository count, size, expected files and write intensity. Prefer native Git for one modest repository or build-heavy, status-heavy and write-heavy work.
- Read ArtifactFS's current official README, consumer skill and releases. Treat prereleases as experimental. Pin an exact reviewed tag or commit and the resulting Scout image by digest.
- Ask before building an image, installing cluster support, mounting
credentials or granting FUSE-related Pod permissions. If the platform lacks
an already reviewed
/dev/fusepath, stop or use native Git; never improvise a privileged Pod, host runtime socket or host-wide policy exception.
Prepare the Scout
- Build
packages/agentos/resources/crewmates/default/images/artifact-fs/Dockerfileon top of the exact reviewed Crewmate image. Do not add ArtifactFS, Go or FUSE packages to the common AgentOS image. - Create one dedicated Crewmate Pod and ServiceAccount for the Assignment.
Keep database, repository and Kubernetes permissions bounded to that Scout.
Use the platform's reviewed device plugin, DRA driver or equivalent to expose
/dev/fuse. Add only the capability and AppArmor/seccomp exception proven necessary by a lifecycle test, and record them in the reviewed per-Agent overlay. Do not grant cluster-admin. - Keep the daemon in the same container and mount namespace as the Scout harness. Use one daemon for the Assignment's repositories, not one daemon or credential set shared across the Fleet.
- Use dedicated scratch state and mount roots keyed by Assignment ID. Keep repository credentials ambient through a scoped Git credential helper, SSH agent or mounted file. Never embed a token in the remote URL, command arguments, brief, image or logs.
- Register each repository with a credential-free HTTPS or SSH URL. Start the long-running daemon in its own named Herdr pane, then start the Scout harness in a separate pane with the declared mount as its workspace. Do not hide the daemon behind a repository-owned background shell script.
- Verify every mounted repository's commit, branch, status and ArtifactFS health before the Scout reads it. A placeholder mount, I/O error, dirty overlay or unexpected ref is a blocker, not a ready workspace.
Preserve the Scout boundary
- Treat the mounted commit as the base view and every local write as disposable overlay state. Do not open a pull request or promote overlay edits into ship work.
- Query only the files needed for the report. Lazy hydration saves startup only when the Scout does not immediately read or build the entire repository.
- Persist findings, decisions and blockers through Task, Assignment and Inbox state. The mount, cache and terminal history are not the durable report.
- If implementation is requested, complete the Scout report and create a ship Task with a native isolated Treehouse worktree. Do not reuse the ArtifactFS overlay as delivered source.
Retire cleanly
- Make the report durable and let the owning Mate accept it.
- Stop the Scout harness, then stop the ArtifactFS daemon and verify every mount is gone. Remove registered repositories and Assignment-scoped scratch state only after acceptance.
- Revoke or unmount repository credentials and remove the exceptional Pod.
- Close the Assignment and retire the Scout through the ordinary guarded delegation lifecycle.
Use $agentos-runtime for native Herdr, Pod and recovery operations. ArtifactFS
does not replace Git as delivered-code truth or PostgreSQL as Fleet truth.
04 同仓库更多 Skills
example-replacement
Reconcile the example replacement AgentOS distribution.
example-additive
Inspect the example additive AgentOS behavior.
agentos-upgrade
Upgrade one persistent AgentOS Mate or, from First Mate with exact Captain authority, every active persistent Mate in one Fleet to an exact published stable release, including its run-once released database migrations. Use for stable First- or Second-Mate updates, resumed post-replacement verification, and authorized rollback boundaries. Do not use for dogfood, Crewmates, provisioning, retired Mates, services, or another Fleet.
agentos-supervision
Reconcile and supervise an AgentOS Mate's direct reports across PostgreSQL, Kubernetes, Herdr, PVCs, and Git. Use at every First- or Second-Mate session start, after restart or compaction, while delegated work is active, when Inbox or runtime state changes, and for blocked, stale, stuck, failed, interrupted, or recovering agents.
agentos-secrets
Safely create, retry, rotate, take over, roll back, project, and revoke AgentOS-managed Kubernetes Secrets without retaining or printing credential bytes. Use whenever an AgentOS workflow moves credential files into Kubernetes, changes an existing managed Secret, verifies a projected credential, encounters conflicting Secret ownership or legacy annotations, or retires a Kubernetes-delivered credential.
agentos-runtime
Inspect, create, and operate AgentOS runtime pods, Kustomize overlays, Herdr sessions, Mise toolchains, agent worktrees, terminals, PVCs, Kubernetes health, attach paths and runtime recovery. Use for normal First- or Second-Mate Kubernetes operations, child-Agent workload creation, tool resolution, stuck-agent debugging, remote terminal access, pod or session recovery, and readiness diagnosis.