Skip to content

Commit e596a07

Browse files
docs: add empty system prompt to codex command for speed (#1300)
## Summary - Add `-c system_prompt=''` to the Codex example command to skip the default system prompt, reducing tokens and latency for commit message generation Follow-up from #1295 — @max-sixty asked about using empty system prompts for speed. **opencode**: Does not currently support disabling the system prompt. The system prompt is always injected, and even the plugin system has a safety fallback that restores the original if emptied. There are open feature requests ([#1242](anomalyco/opencode#1242), [#1894](anomalyco/opencode#1894), [#5005](anomalyco/opencode#5005)) but nothing available yet. **Codex**: Has an undocumented `system_prompt` config key that can be overridden via `-c system_prompt=''`. There are feature requests for dedicated `--system-prompt` flags ([#6023](openai/codex#6023), [#11588](openai/codex#11588)) but not yet merged. Co-authored-by: worktrunk-bot <[email protected]> Co-authored-by: Claude Opus 4.6 <[email protected]>
1 parent 8a0404d commit e596a07

2 files changed

Lines changed: 4 additions & 4 deletions

File tree

‎docs/content/llm-commits.md‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -33,10 +33,10 @@ command = "CLAUDECODE= MAX_THINKING_TOKENS=0 claude -p --model=haiku --tools=''
3333

3434
```toml
3535
[commit.generation]
36-
command = "codex exec -m gpt-5.1-codex-mini -c model_reasoning_effort='low' --sandbox=read-only --json - | jq -sr '[.[] | select(.item.type? == \"agent_message\")] | last.item.text'"
36+
command = "codex exec -m gpt-5.1-codex-mini -c model_reasoning_effort='low' -c system_prompt='' --sandbox=read-only --json - | jq -sr '[.[] | select(.item.type? == \"agent_message\")] | last.item.text'"
3737
```
3838

39-
Uses the fast mini model with low reasoning effort. Requires `jq` for JSON parsing. See [Codex CLI docs](https://developers.openai.com/codex/cli/).
39+
Uses the fast mini model with low reasoning effort and an empty system prompt for faster output. Requires `jq` for JSON parsing. See [Codex CLI docs](https://developers.openai.com/codex/cli/).
4040

4141
### Other tools
4242

‎skills/worktrunk/reference/llm-commits.md‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -19,10 +19,10 @@ command = "CLAUDECODE= MAX_THINKING_TOKENS=0 claude -p --model=haiku --tools=''
1919

2020
```toml
2121
[commit.generation]
22-
command = "codex exec -m gpt-5.1-codex-mini -c model_reasoning_effort='low' --sandbox=read-only --json - | jq -sr '[.[] | select(.item.type? == \"agent_message\")] | last.item.text'"
22+
command = "codex exec -m gpt-5.1-codex-mini -c model_reasoning_effort='low' -c system_prompt='' --sandbox=read-only --json - | jq -sr '[.[] | select(.item.type? == \"agent_message\")] | last.item.text'"
2323
```
2424

25-
Uses the fast mini model with low reasoning effort. Requires `jq` for JSON parsing. See [Codex CLI docs](https://developers.openai.com/codex/cli/).
25+
Uses the fast mini model with low reasoning effort and an empty system prompt for faster output. Requires `jq` for JSON parsing. See [Codex CLI docs](https://developers.openai.com/codex/cli/).
2626

2727
### Other tools
2828

0 commit comments

Comments
 (0)