Replace <output_efficiency> with <response_guidelines> in Grok CLI prompts
Grok CLI renames the system prompt tag <output_efficiency> to <response_guidelines>, strips trailing /** on sandbox allow paths, restricts session dirs to owner-only, and installs native arm64 on Apple Silicon (including Rosetta/x86_64 updaters).Why this loop
Grok CLI be71313 (Source-Revision 5d08d7e4123092567ccd584cd9f99afa2972065c) changes the system prompt contract: <output_efficiency> is renamed to <response_guidelines>. Harnesses, fixtures, or parsers that still key off the old tag will miss guidelines or write into a dead slot. The same cut hardens local surfaces: trailing /** is stripped from sandbox allow paths, so a path that previously globbed a tree now matches a prefix after normalization; session directories are restricted to owner-only permissions. On Apple Silicon the installer now prefers native arm64 even from Rosetta shells and x86_64 updaters, so a translated binary can be replaced without an explicit arch switch. Align prompt templates, sandbox allow lists, session dir modes, and the installed CLI arch with this revision before the next agent run.
Proposed actions
- Search the repo, prompt overlays, snapshots, and string assertions for the literal tag <output_efficiency> and rename it to <response_guidelines> so Grok CLI be71313 / Source-Revision 5d08d7e4123092567ccd584cd9f99afa2972065c still applies response guidelines.
- For each Grok CLI sandbox allow path, strip a trailing /** (the CLI now does this) and re-test that the remaining prefix still allows only the intended directories and denies writes/reads outside it.
- Update every mkdir/umask (and chmod existing session dirs) so Grok CLI session directories are owner-only (0700), matching restrict-session-directories-to-owner-only-permissions.
- On Darwin arm64, reinstall Grok CLI as a native arm64 binary even if the current shell is Rosetta or the updater is x86_64; do not keep a translated CLI.
Agent prompt
Paste into your agent or query via MCP (get_agent_prompt) — free, no extra AI costAGENTS.md / CLAUDE.md / GEMINI.md rule update
DevAgentRadar → Repo Harness Rule Patch
Goal: update our repository's permanent harness instructions based on this release.
Update the file that tool actually reads. Do not dump everything into one file.
- Claude Code → CLAUDE.md plus
.claude/(skills, hooks, settings, agents, commands). It does not read AGENTS.md natively; start CLAUDE.md with@AGENTS.md. - Codex, Copilot, Cursor, Factory Droid, Grok Build, Roo Code, Goose, OpenCode, Amp, Zed, Aider → AGENTS.md.
- Gemini CLI / Antigravity → GEMINI.md. AGENTS.md only if
context.fileNameis set. - Cursor glob-scoped rules →
.cursor/rules/*.mdc(plain.mdis ignored), not a second constitution. - Codex MCP →
.codex/config.toml, not.mcp.json. - Copilot extra instructions →
.github/copilot-instructions.md; custom agents →.github/agents/. - Windsurf →
.windsurf/rules(do not assume AGENTS.md). - Cline →
.clinerules. - Procedures → a skill (
SKILL.md). Enforcement the model must not skip → a hook. Isolated roles → subagents. - Do not fork the same rule into three tool files.
Context
Assistant: xAI Grok CLI (Grok Build) Proposal: Replace <output_efficiency> with <response_guidelines> in Grok CLI prompts Summary: Grok CLI renames the system prompt tag <output_efficiency> to <response_guidelines>, strips trailing /** on sandbox allow paths, restricts session dirs to owner-only, and installs native arm64 on Apple Silicon (including Rosetta/x86_64 updaters). Primary source: https://github.com/xai-org/grok-build/commit/be713136d2a69080743a3f6b3c72077057e5948f
Why it matters
Grok CLI be71313 (Source-Revision 5d08d7e4123092567ccd584cd9f99afa2972065c) changes the system prompt contract: <output_efficiency> is renamed to <response_guidelines>. Harnesses, fixtures, or parsers that still key off the old tag will miss guidelines or write into a dead slot. The same cut hardens local surfaces: trailing /** is stripped from sandbox allow paths, so a path that previously globbed a tree now matches a prefix after normalization; session directories are restricted to owner-only permissions. On Apple Silicon the installer now prefers native arm64 even from Rosetta shells and x86_64 updaters, so a translated binary can be replaced without an explicit arch switch. Align prompt templates, sandbox allow lists, session dir modes, and the installed CLI arch with this revision before the next agent run.
Suggested actions
- Search the repo, prompt overlays, snapshots, and string assertions for the literal tag <output_efficiency> and rename it to <response_guidelines> so Grok CLI be71313 / Source-Revision 5d08d7e4123092567ccd584cd9f99afa2972065c still applies response guidelines.
- For each Grok CLI sandbox allow path, strip a trailing /** (the CLI now does this) and re-test that the remaining prefix still allows only the intended directories and denies writes/reads outside it.
- Update every mkdir/umask (and chmod existing session dirs) so Grok CLI session directories are owner-only (0700), matching restrict-session-directories-to-owner-only-permissions.
- On Darwin arm64, reinstall Grok CLI as a native arm64 binary even if the current shell is Rosetta or the updater is x86_64; do not keep a translated CLI.
Config surfaces this release may change
- sandbox settings — check your repo before applying
After you finish
Do not report this as applied to DevAgentRadar. You cannot write the visitor's loop.
Tell the human: open https://devagentradar.com/proposals/xai-grok-cli-be71313-replace-output-efficiency-with-response-guidelines- and mark Applied, Skipped, or Failed. Proposal id: efbafa5b-dcca-4507-b10f-6c2834c4b22c
Instructions:
- Read the existing context files and settings for the tools in this repo.
- Draft an explicit Git diff. Update ONLY agent harness rules or tool configuration.
- Call out deprecated flags, obsolete habits, or changed permission boundaries.
- Do not touch application logic.
Start with the proposed diff and say which layer and which tool's file it belongs in.
Your loop
This browser · no sign-in · not shared as “you”After you run the prompt
Only you can mark this. Agents cannot write your loop.
Your decision stays on this device. A public tally appears after a few votes.