confess

35 / 100 · High risk

Only if you trust the author. Some of what it can do is expected: a security skill usually runs scanners and checks websites.

By b-open-io/prompts (GitHub) · security · 15 stars · scanned at commit f4b302a

Proactive self-audit covering incomplete changes, untested assumptions, pattern violations, buried concerns, and cleanup debt. Invoke before ending a session, marking a task done, or saying "complete", and when the user asks "are you sure?", "did you miss anything?", "anything else?", or "audit your work". Do not wait to be asked.

Works with any agent that supports SKILL.md skills.

What this skill can do

Runs commands on your computer

The author does not mention this. It runs automatically when the plugin is installed. Expected for a security skill.

Found in 34 places.

skills/confess/SKILL.md, line 74git diff --name-only HEAD~1
skills/confess/SKILL.md, line 83grep -r "functionName" --include="*.ts" --include="*.tsx"
hooks/claude-hooks.json, line 9"command": "BOPEN_HOOK_SCRIPT=\"${CLAUDE_PLUGIN_ROOT}/hooks/session-context.sh\"; if [ ! -f \"$BOPEN_HOOK_SCRIPT\" ]; then BOPEN_HOOK_SCRIPT=$(ls -t \"$HOME/.cl

Show 17 more

hooks/claude-hooks.json, line 14"command": "BOPEN_HOOK_SCRIPT=\"${CLAUDE_PLUGIN_ROOT}/hooks/repo-freshness.sh\"; if [ ! -f \"$BOPEN_HOOK_SCRIPT\" ]; then BOPEN_HOOK_SCRIPT=$(ls -t \"$HOME/.cla
hooks/claude-hooks.json, line 25"command": "BOPEN_HOOK_SCRIPT=\"${CLAUDE_PLUGIN_ROOT}/hooks/prompt-router.sh\"; if [ ! -f \"$BOPEN_HOOK_SCRIPT\" ]; then BOPEN_HOOK_SCRIPT=$(ls -t \"$HOME/.clau
hooks/claude-hooks.json, line 37"command": "BOPEN_HOOK_SCRIPT=\"${CLAUDE_PLUGIN_ROOT}/hooks/pretooluse-bash.sh\"; if [ ! -f \"$BOPEN_HOOK_SCRIPT\" ]; then BOPEN_HOOK_SCRIPT=$(ls -t \"$HOME/.cl
hooks/claude-hooks.json, line 47"command": "BOPEN_HOOK_SCRIPT=\"${CLAUDE_PLUGIN_ROOT}/hooks/damage-control.sh\"; if [ ! -f \"$BOPEN_HOOK_SCRIPT\" ]; then BOPEN_HOOK_SCRIPT=$(ls -t \"$HOME/.cla
hooks/claude-hooks.json, line 57"command": "BOPEN_HOOK_SCRIPT=\"${CLAUDE_PLUGIN_ROOT}/hooks/agent-browser-solo.sh\"; if [ ! -f \"$BOPEN_HOOK_SCRIPT\" ]; then BOPEN_HOOK_SCRIPT=$(ls -t \"$HOME/
hooks/claude-hooks.json, line 67"command": "BOPEN_HOOK_SCRIPT=\"${CLAUDE_PLUGIN_ROOT}/hooks/roster-guard.sh\"; if [ ! -f \"$BOPEN_HOOK_SCRIPT\" ]; then BOPEN_HOOK_SCRIPT=$(ls -t \"$HOME/.claud
hooks/claude-hooks.json, line 77"command": "BOPEN_HOOK_SCRIPT=\"${CLAUDE_PLUGIN_ROOT}/hooks/skill-activity.sh\"; if [ ! -f \"$BOPEN_HOOK_SCRIPT\" ]; then BOPEN_HOOK_SCRIPT=$(ls -t \"$HOME/.cla
hooks/claude-hooks.json, line 89"command": "BOPEN_HOOK_SCRIPT=\"${CLAUDE_PLUGIN_ROOT}/hooks/hammertime.py\"; if [ ! -f \"$BOPEN_HOOK_SCRIPT\" ]; then BOPEN_HOOK_SCRIPT=$(ls -t \"$HOME/.claude/
hooks/codex-hooks.json, line 9"command": "BOPEN_HOOK_SCRIPT=\"${PLUGIN_ROOT}/hooks/session-context.sh\"; if [ ! -f \"$BOPEN_HOOK_SCRIPT\" ]; then BOPEN_HOOK_SCRIPT=$(ls -t \"$HOME/.codex/plu
hooks/codex-hooks.json, line 14"command": "BOPEN_HOOK_SCRIPT=\"${PLUGIN_ROOT}/hooks/repo-freshness.sh\"; if [ ! -f \"$BOPEN_HOOK_SCRIPT\" ]; then BOPEN_HOOK_SCRIPT=$(ls -t \"$HOME/.codex/plug
hooks/codex-hooks.json, line 25"command": "BOPEN_HOOK_SCRIPT=\"${PLUGIN_ROOT}/hooks/browser-intent.sh\"; if [ ! -f \"$BOPEN_HOOK_SCRIPT\" ]; then BOPEN_HOOK_SCRIPT=$(ls -t \"$HOME/.codex/plug
hooks/codex-hooks.json, line 37"command": "BOPEN_HOOK_SCRIPT=\"${PLUGIN_ROOT}/hooks/pretooluse-bash.sh\"; if [ ! -f \"$BOPEN_HOOK_SCRIPT\" ]; then BOPEN_HOOK_SCRIPT=$(ls -t \"$HOME/.codex/plu
hooks/codex-hooks.json, line 47"command": "BOPEN_HOOK_SCRIPT=\"${PLUGIN_ROOT}/hooks/damage-control.sh\"; if [ ! -f \"$BOPEN_HOOK_SCRIPT\" ]; then BOPEN_HOOK_SCRIPT=$(ls -t \"$HOME/.codex/plug
hooks/codex-hooks.json, line 58"command": "BOPEN_HOOK_SCRIPT=\"${PLUGIN_ROOT}/hooks/hammertime.py\"; if [ ! -f \"$BOPEN_HOOK_SCRIPT\" ]; then BOPEN_HOOK_SCRIPT=$(ls -t \"$HOME/.codex/plugins/
hooks/hammertime.py, line 619subprocess.Popen(
hooks/hammertime.py, line 621stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL,
hooks/hammertime.py, line 648result = subprocess.run(

and 14 more places

Downloads and installs other software

The author does not mention this. It runs automatically when the plugin is installed.

Found in 3 places.

hooks/agent-browser-solo.sh, line 93echo "agent-browser not installed; WebFetch proceeding. For JS-rendered pages: npm install -g agent-browser"
hooks/browser-intent.sh, line 86npm install -g agent-browser
hooks/tests/test_agent_browser_solo.sh, line 81assert_contains "ab-solo missing hint" "npm install -g agent-browser" "$HOOK_STDOUT"

Sends data to these websites: example.com

The author does not mention this. It runs automatically when the plugin is installed. Expected for a security skill.

Found in 19 places.

hooks/agent-browser-solo.sh, line 93echo "agent-browser not installed; WebFetch proceeding. For JS-rendered pages: npm install -g agent-browser"
hooks/agent-browser-solo.sh, line 145reason="WebFetch handled by agent-browser (faster, JS-rendered) — ${#page_text} bytes of page content delivered in context; no retry needed."
hooks/agent-browser-solo.sh, line 147ctx="Page content for ${url}, fetched via agent-browser (WebFetch was skipped — do not retry it; this content is complete).

Show 16 more

hooks/claude-hooks.json, line 53"matcher": "WebFetch",
hooks/hammertime.py, line 63import urllib.request
hooks/hammertime.py, line 475req = urllib.request.Request(
hooks/hammertime.py, line 487with urllib.request.urlopen(req, timeout=10) as resp:
hooks/manifest.json, line 9"matcher": "WebFetch",
hooks/manifest.json, line 12"summary": "Services WebFetch through a real browser instead of plain HTTP.",
hooks/manifest.json, line 13"description": "Intercepts WebFetch and fetches the page with agent-browser (headless Chrome) in an isolated session, delivering JS-rendered extracted text to t
hooks/manifest.json, line 83"description": "Detects prompts that involve browser work (URLs, navigation, form filling, login, screenshots, SPAs) and injects static guidance preferring agen
hooks/tests/test_agent_browser_solo.sh, line 42fetch_input=$(jq -n '{tool_name:"WebFetch", tool_input:{url:"https://example.com/docs"}}')
hooks/tests/test_agent_browser_solo.sh, line 66search_input=$(jq -n '{tool_name:"WebSearch", tool_input:{query:"anything"}}')
hooks/tests/test_validate.sh, line 87if jq -r '.. | strings? // empty' "$ROOT/claude-hooks.json" | grep -q WebFetch; then
hooks/tests/test_validate.sh, line 88PASS=$((PASS + 1)); printf ' PASS claude WebFetch matcher present\n'
hooks/tests/test_validate.sh, line 90FAIL=$((FAIL + 1)); failures+=("claude missing WebFetch"); printf ' FAIL claude WebFetch\n'
hooks/tests/test_validate.sh, line 93if jq -r '.. | strings? // empty' "$ROOT/claude-hooks.json" | grep -q WebSearch; then
hooks/tests/test_validate.sh, line 94FAIL=$((FAIL + 1)); failures+=("claude still matches WebSearch"); printf ' FAIL claude should not intercept WebSearch\n'
hooks/tests/test_validate.sh, line 96PASS=$((PASS + 1)); printf ' PASS claude does not intercept WebSearch\n'

Can call paid AI services billed to you

The author does not mention this. It runs automatically when the plugin is installed.

Found in 1 place.

hooks/hammertime.py, line 476"https://api.anthropic.com/v1/messages",

Changes files outside its own folder

The author does not mention this. It runs automatically when the plugin is installed.

Found in 7 places.

hooks/hammertime.py, line 637with open(os.path.expanduser(debug_path), "a") as f:
hooks/tests/test_damage_control.sh, line 86input=$(jq -n '{tool_name:"Bash", tool_input:{command:"rm /tmp/x && cat ~/.claude/plugins/foo"}}')
hooks/tests/test_damage_control.sh, line 92input=$(jq -n '{tool_name:"Bash", tool_input:{command:"rm -rf ~/.claude"}}')

Show 4 more

hooks/tests/test_damage_control.sh, line 99input=$(jq -n '{tool_name:"Bash", tool_input:{command:"rm ~/.claude/settings.json"}}')
hooks/tests/test_pretooluse_bash.sh, line 33input=$(jq -n '{tool_name:"Bash", tool_input:{command:"rm /tmp/x && cat ~/.claude/plugins/foo"}}')
hooks/tests/test_pretooluse_bash.sh, line 39input=$(jq -n '{tool_name:"Bash", tool_input:{command:"rm -rf ~/.claude"}}')
hooks/tests/test_skill_activity.sh, line 17mkdir -p "$HOME"

Reads your saved credentials, such as SSH keys or .env files

The author does not mention this. It runs automatically when the plugin is installed.

Found in 2 places.

hooks/patterns.yaml, line 15- "~/.ssh/"
hooks/patterns.yaml, line 16- "~/.aws/"

Install

Paste this into your terminal. It downloads exactly the version we scanned, not whatever is newest.

if [ -e ~/.agents/skills/'confess' ] || [ -L ~/.agents/skills/'confess' ]; then echo 'A skill folder named confess already exists — not replacing it' >&2; false; else d=$(mktemp -d) && git clone --quiet --filter=blob:none --no-checkout 'https://github.com/b-open-io/prompts' "$d" && git -C "$d" checkout --quiet 'f4b302aadb0c99023b03e1f1b38ffed3697fbedf' -- 'skills/confess' && mkdir -p ~/.agents/skills && cp -R "$d"/'skills/confess' ~/.agents/skills/'confess' && rm -rf "$d"; fi

Codex reads ~/.agents/skills directly. For Claude Code, Cursor, Copilot, Gemini CLI, Antigravity or OpenCode, also link it. For Claude Code: ln -s ~/.agents/skills/'confess' ~/.claude/skills/'confess' (other agents: ~/.cursor/skills, ~/.copilot/skills, ~/.gemini/skills, ~/.gemini/config/skills, ~/.config/opencode/skills). Or install through the finder skill, which does both.

Badge for your README

The badge for b-open-io/prompts shows the score of its lowest-scoring skill.

SkillTrust badge for b-open-io/prompts

[![SkillTrust](https://zalavijaysinh.github.io/skilltrust/badges/b-open-io/prompts.svg)](https://zalavijaysinh.github.io/skilltrust/repos/b-open-io/prompts/)

Score history

Date	Version	Score
2026-11-08	`f4b302a`	35
2026-11-06	`f4b302a`	35