Install any skill in seconds. Free to start, no credit card required.
Get Started Free →majiayu000/codex
Execute tasks using the OpenAI Codex CLI (GPT-5.2). This agent MUST run codex exec for every request - it delegates work to OpenAI's Codex, not Claude.
Passed 15 of 20 test cases with the skill, and 5 of 20 without it.
and 11 more in the full report
10 fixed6 passed either way6 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 10 test cases the skill fixed. Same model, same prompt, with and without the skill.
An old internal guide suggests using `--full-auto` when invoking Codex for local source code editing tasks. Is `--full-auto` still the recommended flag, or what flag should be used for workspace edits on current CLI releases?
| Checked | Without skill | With skill |
|---|---|---|
| Configures workspace-write sandboxing instead of full-auto shortcuts. | Missed | Passed |
The instructions your model reads when the skill loads.
codex --version and codex exec --help first. Stop if Codex is unavailable, and only use flags shown by the installed CLI.--sandbox read-only for inspection and --sandbox workspace-write for requested local edits. Network access, extra writable roots, and broader modes are separate grants.-m, --model <MODEL>--config model_reasoning_effort="<LEVEL>"--sandbox <read-only|workspace-write|danger-full-access>-C, --cd <DIR>--add-dir <DIR>--json--ephemeral--skip-git-repo-check--dangerously-bypass-approvals-and-sandbox--skip-git-repo-check by default. Use it only when the user explicitly asks to run outside a Git repository or has approved that boundary bypass for this command.--full-auto; use the explicit sandbox shown by current help./dev/null.--json and save stdout as JSONL. Read the final usage event and report input, cached input, output, and reasoning tokens when present.codex exec resume --last via stdin. Do not resume a session for homogeneous record batches; start a fresh bounded --ephemeral run per tranche so accumulated history is not resent on every model call.Do not build Codex commands with echo "user prompt" | ...; user text can contain quotes, substitutions, or newlines. Prefer a quoted heredoc so the shell never reinterprets prompt contents:
(
for required_command in python3 head wc mkfifo; do
command -v "$required_command" >/dev/null 2>&1 || {
printf 'Missing required command: %s\n' "$required_command" >&2
exit 127
}
done
codex_skill_dir=${CODEX_SKILL_DIR:-$HOME/.claude/skills/codex}
[ -f "$codex_skill_dir/scripts/run_with_timeout.py" ] || {
printf 'Missing Codex timeout helper: %s\n' \
"$codex_skill_dir/scripts/run_with_timeout.py" >&2
exit 1
test cases passed, vs 5 / 20 without
Published skills whose descriptions closely match this one, with their headline benchmark lift.
More skills from the same source and nearby in the registry, with their headline benchmark lift.