Vault
Store and use the user's API keys, passwords, tokens, and credentials WITHOUT ever seeing their values. Trigger whenever the user mentions an API key, token, password, credential, login, secret, or .env file; wants to store or rotate one; or a command needs auth / fails with 401/403/"missing key". Values live in the macOS Keychain; the agent only ever reads a pointer manifest.
Install
npx skills add AnYejun/blind-vault --skill vaultgit clone --depth 1 https://github.com/AnYejun/blind-vault.git /tmp/blind-vault
mkdir -p ~/.claude/skills
cp -r /tmp/blind-vault/. ~/.claude/skills/vaultPersonal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.
From the SKILL.md
# Blind Vault You are handling the user's secrets. The prime directive: **a secret value must never enter your context window.** Not in a tool result, not in a file you read, not echoed back "to confirm". You work with *pointers*; the OS works with *values*. The CLI lives at `~/.claude/skills/vault/bin/vault` (call it as `"$HOME/.claude/skills/vault/bin/vault"`; below abbreviated `vault`). Pointer manifest: `~/.blindvault/manifest.json` — safe to read, it contains no values. ## Rule zero If the user pastes a secret value into the chat, it is already in the transcript and logs. Do not repeat it, do not confirm it back. Tell them: "that key is now in chat logs — treat it as burned and rotate it after we store the new one." Then store it with `--from-stdin` (it's already exposed; the dialog adds nothing) and remind them to rotate. ## Storing a secret Never ask the user to paste a value. Two paths — prefer the dashboard when the user wants to manage several keys or seems uncomfortable with the CLI: **Dashboard (best UX):** run `vault ui` in the background and tell the user: "I've opened the vault dashboard at the printed 127.0.0.1 URL — add or manage keys there; I only ever see the pointer list." The form's password field goes straight to the Keychain. When they say done, `vault ls` to pick up the new pointers. **Inline (one-off):** run: ```bash vault add openai-api-key --service "OpenAI" --env OPENAI_API_KEY --allow "api.openai.com" --note "personal, pay-as-you-go" ```
Details
- Skill name
vault- Repository
- AnYejun/blind-vault
- License
- MIT
- Repo updated
- 28 October 2026
FAQ
What does the Vault skill do?
Store and use the user's API keys, passwords, tokens, and credentials WITHOUT ever seeing their values. Trigger whenever the user mentions an API key, token, password, credential, login, secret, or .env file; wants to store or rotate one; or a command needs auth / fails with 401/403/"missing key". Values live in the macOS Keychain; the agent only ever reads a pointer manifest.
How do I install the Vault skill?
skills CLI (Claude Code, Cursor, Codex and more): npx skills add AnYejun/blind-vault --skill vault. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.
Is Vault popular?
It lives in AnYejun/blind-vault, which has 12 GitHub stars and 1 forks.
Related skills
Multi Search
11Nex-ZMH/Agent-websearch-skill
智能多引擎搜索,自动检测网络环境并按优先级切换:DuckDuckGo -> Tavily -> Bing API -> Bing爬虫。支持自动配额管理和网络缓存。Invoke when user needs web search with automatic engine selection and network adaptation.
Rust Elegant
12fuqiuluo/rust-elegant
Rust 优雅开发规范,必须在以下场景触发:为 Rust 项目写代码、review Rust 代码、设计 Rust 项目结构、处理 Rust 错误类型时。覆盖命名、注释规范、枚举设计、trait 抽象、项目结构、错误处理、所有权、迭代器、宏、内存分配、match 穷尽性、unsafe 等所有 Rust 惯用法。只要任务涉及写或修改 Rust 代码,必须读取此 skill。
B3sty Skill
10b3sty191/b3sty-skill
b3sty rules for implementing, reviewing, debugging, refactoring, or optimizing RedM/FiveM Lua resources. Use for FXServer manifests, client/server Lua, natives/entities, native invocation from docs (hashes, InvokeNative, marshalling, RDR3 structs), events/callbacks/exports/NUI, NUI/browser-UI bridge (SendNUIMessage, RegisterNUICallback, SetNuiFocus), NUI XSS/player-string rendering, OneSync/entity networking (net IDs, ownership, routing buckets, broadcasts), built-in client events (weaponDamageEvent, explosionEvent), anti-cheat/event-security hardening (give-item/give-money dupe prevention, Re
Enterprise Architect
7rafalr100/enterprise-architect-skill
Operate as a world-class Enterprise Architect (FAANG / top-tier consulting calibre) grounded in TOGAF, Zachman, and modern EA practice. Opens with discovery — asks the clarifying questions that materially change the architecture before producing any deliverable. Activates when the user asks for enterprise-architecture work — target architectures, reference architectures, architecture patterns, standards, principles, capability maps, technology strategy, roadmaps, trade-off analysis, architecture decision records, governance, or audience-tiered architecture presentations (C-level/board, solutio
Grok Bot Gateway
10dimpurr/grok-bot-gateway
An official-webhook-only gateway to all your Grok Bots. Use this when a coding agent or script needs to list a Grok Bot team, read a Bot's transcript or message a Bot, or when a Grok Bot user wants to host that gateway; it detects which side you're on and guides you. Callers (Claude Code, Codex, scripts) use the grokgw client; hosts create the gateway Bot, its webhook routine and a host config. Getting results back is optional and goes over a return path you choose (none by default).
Codex Sites Convex
12convex/Codex-Sites-Convex-Backend-Skill
Build, extend, validate, and publish a ChatGPT Sites application whose durable database and backend functions run on Convex. Use when a user invokes $codex-sites-convex, selects this skill from the skill picker, asks for a ChatGPT Site backed by Convex, or asks to add Convex data, realtime queries, mutations, authentication, files, schedules, or actions to an existing ChatGPT Sites project.
Browse the most popular Claude skills, all software development skills, or write your own with the skills generator.