pavlo-terminal-commands（巴甫洛终端命令）

Skill: Terminal Commands

Safe terminal command patterns for AI coding agents. Prevents common pitfalls like shell quote misparses and hanging terminals.

Python Execution

Always write Python scripts to a temp file and run from there.

Do

cat > /tmp/script.py << 'PYEOF'
import json

data = {"key": "value"}
print(json.dumps(data, indent=2))
PYEOF
python3 -u /tmp/script.py

• Use python3 -u (unbuffered) to ensure output appears immediately.
• Use a heredoc with quoted delimiter (<< 'PYEOF') so the shell does not expand variables inside the script.
• Use /tmp/script.py or another temp path.

Do Not

# ❌ Never do this — shell misparses inner quotes and hangs on dquote>
python3 -c "
import json
data = {'key': 'value'}
print(json.dumps(data, indent=2))
"

• Never pass multi-line Python via python3 -c "...".
• The shell misparses inner quotes (single inside double, nested strings) and hangs waiting for input (dquote>).

General Terminal Rules

• Disable pagers. Commands like git, less, man may invoke a pager that blocks the agent. Always add flags to disable:

- git --no-pager log instead of git log - git --no-pager diff instead of git diff - Pipe to cat as a fallback: some-command | cat

• Quote variables. Use "$var" instead of $var to handle paths and values with spaces.
• Avoid printing credentials. Never echo secrets, tokens, or API keys to the terminal unless absolutely required.
• Check exit codes. Use && to chain dependent commands so failures stop the chain: command1 && command2 && command3