novotnyllc/roundhouse/fleet-auth

Claude Code and agent skill by novotnyllc/roundhouse

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Audit and deliberately reconcile configured authentication artifacts across machines without exposing secrets. Use for Codex auth.json, xurl auth.yml, GitHub or tool sessions, matching credential hashes, native auth health, reauthentication, or encrypted credential installation.

Install

npx vskill@latest install novotnyllc/roundhouse/fleet-auth

SKILL.md

Fleet Authentication

Set SKILL_DIR to the absolute directory containing this loaded SKILL.md and CLI="$SKILL_DIR/../../scripts/roundhouse"; the shell working directory is not the skill directory. Run the auth inventory section first. Show configured artifact name, path, strategy, owner, mode, size, mtime, SHA-256, link status, and native verification result. Do not read credential contents into the conversation, logs, JSONL, command arguments, or task prompts. Read "$SKILL_DIR/../../references/agent-settings-and-auth.md" for the distinct Claude CLI/Desktop and Codex credential boundaries.

Honor each configured strategy:

  • chezmoi: delegate declarative state to agent-utilities:fleet-chezmoi.
  • reauth: run the tool's native login on that machine.
  • encrypted-install: resolve the configured secret reference only after the

user approves exact source, target hosts, and destination path.

  • ignore: report state and make no change.

Before mutation, seal the exact artifact operation with "$CLI" seal-plan DRAFT SNAPSHOT PLAN, verify live host identity, recapture auth inventory, and require "$CLI" verify-preconditions PLAN CURRENT-SNAPSHOT to succeed. Obtain separate user approval for the exact sealed operation. Reject symlink destinations and capture current metadata for rollback. Fetch the secret directly into a mode-0600 temporary file on the target, validate type/size, then atomically rename it into the user-owned parent directory. Never copy through a world-readable location. Run the configured native verification command; on failure restore the prior file or remove the new file. For a local encrypted-install, use "$CLI" apply-plan PLAN PLAN-ID OUTPUT; it resolves the configured op:// reference directly into a private same-directory temporary file, replaces atomically, verifies, and rolls back on failure. Remote targets over SSH use "$CLI" apply-ssh-plan PLAN PLAN-ID OUTPUT. Apply recaptures trusted preflight itself and preserves authoritative partial output after a failed operation or postcondition when post-inventory remains available. SSH uses bounded connection/keepalive timeouts and verifies the configured native hostname/user. Native Windows auth mutation is not supported by this release; reauthenticate interactively on that host rather than copying credentials through a task. SSH reauthentication is also rejected because login needs a visible interactive terminal or browser. Complete the one-time login on the target, then re-run inventory; apply-ssh-plan remains available for noninteractive encrypted installs.

Pathless per-machine or native-store artifacts are status-only sessions; inventory does not infer their credential backend. Their status comes only from the configured native verification command and an unhealthy login is marked as reauth_required with a manual host action. A full Claude login is required once per Remote Control host; do not substitute a setup token or copy Claude state. Codex file-backed auth.json is portable only when the user explicitly chooses encrypted-install; otherwise keep Codex authentication per-machine.

Matching SHA-256 proves identical bytes, not valid authentication. Prefer per-machine least-privilege credentials for unattended work. For Windows, Codex uses a visible saved-project task as described in "$SKILL_DIR/../../references/codex-remote-control.md", which defers the task's model and effort choice to railyard:model-routing; Claude reports unsupported. Never route secrets through WSL or another machine as a bridge.

Privilege-lane records are status-only for this skill. Use "$CLI" privilege-status HOST SNAPSHOT and the shared Codex/Claude vocabulary prepare-privilege-enrollment, verify-privilege-plan, submit-privilege-plan, and lookup-privilege-result (the optional CA lane adds prepare-privilege-identity, preview-privilege-upgrade, and preview-privilege-revocation), but never put an auth artifact, credential, secret reference, encrypted file, token, private key, or secret-backed template in a lane request: the lane carries package tokens and versions, nothing else. Preserve the readiness states ready, needs_one_time_approval, disabled, legacy, unsupported, drifted, and unreachable, and the operation-result states partial and rejected, without fallback. Never ask for or relay a sudo or Administrator password; the host's single OS approval (roundhouse privilege-enroll HOST) is the only privileged credential interaction and it belongs to the owner at the host's own prompt.

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/novotnyllc/roundhouse/fleet-auth/badge)](https://verified-skill.com/skills/novotnyllc/roundhouse/fleet-auth)

Something wrong with this skill or its page? Tell us.