How do I install the Nopeek skill?

Run npx skillmds@latest add spences10/nopeek-2 in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Nopeek safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Nopeek?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Nopeek free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Nopeek?

spences10 (@spences10) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Nopeek by spences10 — Agent Skill

Name: Nopeek
Author: spences10

File contents

nopeek — Secret Safety for Agent Sessions

Secure proxy between coding agents and your secrets. The agent knows key names, never key values. For choosing secret-safety tooling alongside recall, MCP, and database tools, use ecosystem-guide.

Why

Agent tool output can be sent to model providers and retained outside your machine. If a CLI prints a token, that token may enter conversation context or logs. nopeek reduces that risk.

Quick Start

All nopeek commands are safe inside agent sessions — they never output secret values, only key names.

1. Scan and migrate CLIs

pnpx nopeek init

Detects installed cloud CLIs (aws, hcloud, kubectl, etc.), checks their auth config, and migrates to profile-based auth where needed.

2. Load .env secrets

pnpx nopeek load .env
pnpx nopeek load .env --only DATABASE_URL,API_KEY

Injects values into the session via CLAUDE_ENV_FILE (when available) or outputs a source command. Only key names appear in output — never values.

3. Store arbitrary keys

pnpx nopeek set MY_API_KEY --from-env
pnpx nopeek set STRIPE_KEY --value "sk_live_..."

When installed in Claude Code, stored keys auto-load on each session via the SessionStart hook.

4. Verify

pnpx nopeek status
pnpx nopeek list

What the hooks do

Hook	Event	Purpose
session-load.sh	SessionStart	Loads stored keys + CLI profiles into `CLAUDE_ENV_FILE`
redact-output.sh	PreToolUse(Bash)	Wraps cloud CLI output through secret pattern redaction

Limitations

Redaction is best-effort. Regex patterns catch known formats (AWS keys, bearer tokens, Stripe keys, private key headers, connection strings) but cannot catch every possible secret.
This is a safety net, not a guarantee. The primary defense is loading secrets via CLAUDE_ENV_FILE so they never appear in output at all.
No PostToolUse output redaction in Claude Code. The PreToolUse hook rewrites commands to filter output before the agent sees it.
Commands with pipes or redirections are not wrapped. The redaction hook skips commands containing |, >, or < to avoid breaking complex command semantics.
Requires jq. The redact hook will not fire without jq installed.
Waiting on anthropics/claude-code#39882 — PreApiCall/PostApiCall hooks would enable full payload redaction before data leaves your machine.

Audit .env files

pnpx nopeek audit

Scans current directory for .env files containing secrets and checks .gitignore coverage.

Troubleshooting

Keys not loading on session start:

Check pnpx nopeek status — are keys stored?
For Claude Code hook installs, verify the plugin is installed: /plugin list
Try /reload-plugins

Cloud CLI still showing secrets:

The redaction hook only fires for commands starting with known CLI names (aws, hcloud, kubectl, doctl, gcloud, az, terraform)
Commands with pipes or redirections are skipped to avoid breaking semantics
Run pnpx nopeek load .env to make secrets available as env vars, then use $VAR_NAME in commands

---name: nopeek-2description: Secret-safe agent sessions via nopeek CLI. Use when user asks about secrets, .env loading, credential safety, or keeping API keys out of conversation context.---# nopeek — Secret Safety for Agent SessionsSecure proxy between coding agents and your secrets. The agent knows key names, never key values. For choosing secret-safety tooling alongside recall, MCP, and database tools, use `ecosystem-guide`.## WhyAgent tool output can be sent to model providers and retained outside your machine. If a CLI prints a token, that token may enter conversation context or logs. nopeek reduces that risk.## Quick StartAll nopeek commands are safe inside agent sessions — they never output secret values, only key names.### 1. Scan and migrate CLIs```bashpnpx nopeek init```Detects installed cloud CLIs (aws, hcloud, kubectl, etc.), checks their auth config, and migrates to profile-based auth where needed.### 2. Load .env secrets```bashpnpx nopeek load .envpnpx nopeek load .env --only DATABASE_URL,API_KEY```Injects values into the session via `CLAUDE_ENV_FILE` (when available) or outputs a source command. Only key names appear in output — never values.### 3. Store arbitrary keys```bashpnpx nopeek set MY_API_KEY --from-envpnpx nopeek set STRIPE_KEY --value "sk_live_..."```When installed in Claude Code, stored keys auto-load on each session via the SessionStart hook.### 4. Verify```bashpnpx nopeek statuspnpx nopeek list```## What the hooks do| Hook | Event | Purpose || ---------------- | ---------------- | ------------------------------------------------------- || session-load.sh | SessionStart | Loads stored keys + CLI profiles into `CLAUDE_ENV_FILE` || redact-output.sh | PreToolUse(Bash) | Wraps cloud CLI output through secret pattern redaction |## Limitations- **Redaction is best-effort.** Regex patterns catch known formats (AWS keys, bearer tokens, Stripe keys, private key headers, connection strings) but cannot catch every possible secret.- **This is a safety net, not a guarantee.** The primary defense is loading secrets via `CLAUDE_ENV_FILE` so they never appear in output at all.- **No PostToolUse output redaction in Claude Code.** The PreToolUse hook rewrites commands to filter output before the agent sees it.- **Commands with pipes or redirections are not wrapped.** The redaction hook skips commands containing `|`, `>`, or `<` to avoid breaking complex command semantics.- **Requires jq.** The redact hook will not fire without jq installed.- **Waiting on anthropics/claude-code#39882** — PreApiCall/PostApiCall hooks would enable full payload redaction before data leaves your machine.## Audit .env files```bashpnpx nopeek audit```Scans current directory for .env files containing secrets and checks .gitignore coverage.## Troubleshooting**Keys not loading on session start:**- Check `pnpx nopeek status` — are keys stored?- For Claude Code hook installs, verify the plugin is installed: `/plugin list`- Try `/reload-plugins`**Cloud CLI still showing secrets:**- The redaction hook only fires for commands starting with known CLI names (aws, hcloud, kubectl, doctl, gcloud, az, terraform)- Commands with pipes or redirections are skipped to avoid breaking semantics- Run `pnpx nopeek load .env` to make secrets available as env vars, then use `$VAR_NAME` in commands

Frequently asked questions

Nopeek