careful

careful is a skill for Claude Code, Codex from panda850819/verbs. It costs 67 tokens per session (848 once invoked), scanned C, original, MIT.

A safety checkpoint for commands that could delete, overwrite, or alter production, shared infrastructure, live agent files, or unfamiliar code.

In plain words
What is it for?
Use it before destructive Git commands, risky file removal, overwriting files outside the project, or other actions that could damage shared state.
Why use it?
It prevents a high-risk action from running before you explicitly approve it, while reversible work can continue.

Skill for Claude CodeCodex

Written for Claude Code and Codex: user-invocable in frontmatter, but also reads ~/.codex or $CODEX_HOME. Also seen: reads .claude/ paths; mentions Codex.

Part of the verbs plugin — 8 skills shipped together

Good fit Use it before destructive Git commands, risky file removal, overwriting files outside the project, or other actions that could damage shared state.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/panda850819/verbs/careful
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add panda850819/verbs --skill careful
Clone the repo
git clone --depth 1 https://github.com/panda850819/verbs

Made for: Claude Code, Codex.

Or install verbs, the plugin that ships this one along with the rest of its 8 skills.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for careful

README.md
[![agentmods](https://agentmods.dev/badge/skills/panda850819/verbs/careful.svg)](https://agentmods.dev/skills/panda850819/verbs/careful)
Your own site
<a href="https://agentmods.dev/skills/panda850819/verbs/careful"><img src="https://agentmods.dev/badge/skills/panda850819/verbs/careful.svg" alt="Measured on agentmods" height="20"></a>
Per session 67 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 848 The whole file, excluding the scripts and references it only reads on demand.
Security scan C 1 finding. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00067 $0.00848
Opus 5 $0.00034 $0.00424
Sonnet 5 $0.00013 $0.00170
Haiku 4.5 $0.00007 $0.00085

Measured 7d ago against content hash 2553867f4ce4, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-07, from the pricing page.

Security

Grade C, and why

careful scanned grade C with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 7d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Recursive force deletehighDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

- `rm -rf` on an **unscoped / non-reinstallable** path: anything outside the current project, a path with a glob/variable that could expand wrong, or removal of source / data / config that isn't trivially regenerable
skills/engineering/careful/SKILL.md · 82 lines

How it starts

The opening of the file, as written. The whole thing — 82 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Careful Mode

Ordinary model caution can still proceed without an answer. Careful blocks the listed destructive or high-risk action until explicit confirmation arrives.

On Invoke

Announce: "CAREFUL mode ON. Will confirm before destructive actions."

While Active

Before executing any of the following, pause and ask the user for explicit confirmation:

Git

  • git push --force, git reset --hard, git clean -f
  • git branch -D (force delete)
  • git checkout . or git restore . (discard all changes)
  • git rebase on shared branches
  • Any push to main/master

Filesystem

  • rm -rf on an unscoped / non-reinstallable path: anything outside the current project, a path with a glob/variable that could expand wrong, or removal of source / data / config that isn't trivially regenerable
  • Deleting more than 3 files at once
  • Overwriting files outside the current project

Exemption (does NOT trigger the gate, regardless of where the path lives): removal of a directory whose basename names a trivially-reinstallable artifact — node_modules, .next, dist, build, target, .cache, .turbo, __pycache__, or a lockfile-regenerable deps dir. Key off the artifact NAME, not project membership: rm -rf /anywhere/node_modules is exempt because reinstall restores it. Do NOT use "is this the current project?" as the test — you often cannot resolve the cwd vs the target path, and an absolute foreign-looking path must not re-trigger the gate. The only conditions are: (1) basename is a regenerable artifact above, and (2) the path is explicit, no glob/variable that could expand wrong. Multi-path: if the command removes more than one path, EVERY path must independently satisfy (1) and (2) — a single foreign or non-artifact path (e.g. rm -rf node_modules ../../prod-data) re-arms the gate for the whole command. The gate is for irreversible / shared-state damage, not routine cleanup.

External

  • Any API call that mutates external state (POST/PUT/DELETE to production)
  • Publishing packages (npm publish, cargo publish)
  • Deploying to production environments

Read the full file on GitHub · 82 lines

Files

What ships with it

2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 7d ago First seen · 82 lines · 67 tokens per session scan C 2553867f4ce4

Subscribe to this mod's changes

careful is a skill published in the GitHub repository panda850819/verbs (5 stars, last pushed 6d ago), licensed MIT. It adds 67 tokens to every session and 848 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

backfill

Reconstruct an OKF bundle by event-sourcing a repository's history (git log and Claude session transcripts). Use when creating an .okf/ bundle for an existing repository that predates this skill, or when resuming an interrupted backfill session. Triggers on: "reconstruct the OKF bundle", "backfill the knowledge…

scaccogatto/okf-skills · 80 tokens

okf

Author, maintain, and consume Open Knowledge Format (OKF) knowledge bundles — portable markdown + YAML frontmatter that both humans and agents read. Use when capturing project knowledge (services, APIs, schemas, metrics, runbooks, decisions) into an OKF bundle, when updating one after code or docs change, or when a…

scaccogatto/okf-skills · 127 tokens

validate

Check that an Open Knowledge Format (OKF) bundle is conformant with the v0.2 spec (§11). Use when asked to validate, lint, or check an OKF bundle, or before committing changes to one. Runs a deterministic Python checker — not an eyeball pass. Also migrates a v0.1 bundle to v0.2 in place with --migrate.

scaccogatto/okf-skills · 82 tokens

visualize

Render an Open Knowledge Format (OKF) bundle as a single self-contained, interactive HTML graph (viz.html) — concepts as nodes coloured by type and sized by body length, markdown links and bundle-internal sources as edges, a wiki-style detail panel with rendered markdown, v0.2 trust/lifecycle/provenance metadata, and…

scaccogatto/okf-skills · 111 tokens

callee-create-agent

Create project-defined Callee Role, Script, Human, Sequential, Loop, and Router agents in Markdown or YAML. Use when the user asks to generate, scaffold, compose, or author a Callee agent or deterministic workflow.

baldaworks/callee · 49 tokens

callee-run-agent

Run and combine project-defined Callee agents and deterministic workflows for coding work. Use when the user asks to delegate investigation, review, implementation, testing, or a named workflow through Callee.

baldaworks/callee · 42 tokens