[PR #1571] [MERGED] test(clients): verify native Grok Build workflows #2981

New issue

Closed

opened 2026-10-17 21:12:31 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-17 21:12:31 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/morluto/rea/pull/1571
Author: @N0zoM1z0
Created: 10/10/2026
Status: ✅ Merged
Merged: 10/10/2026
Merged by: @N0zoM1z0

Base: main ← Head: test/grok-native-client-20261010

📝 Commits (1)

6473c40 test(clients): verify native Grok Build workflows

📊 Changes

3 files changed (+560 additions, -0 deletions)

View changed files

📝 docs/testing.md (+20 -0)
📝 package.json (+1 -0)
➕ scripts/verify-grok-client.mjs (+539 -0)

📄 Description

REA already supports Grok Build configuration, but lacked a native client workflow check. Add an optional verifier using the official Grok Build CLI, its default lazy search_tool/use_tool interface and a loopback OpenAI completions/SSE model fixture.

The lane guards setup writes, preserves unrelated settings/backups, checks idempotence, loads the entire skill and validates all 139 advertised input schemas through native discovery. When discovery is still partial, the native agent diagnoses REA registration and retries. A larger Unicode-path analysis produces about 738 KB of Evidence: Grok retains the complete JSON, and a native terminal query returns the selected export facts, subject and complete artifact digest to the next model request. The retained file is checked against REA's named output schema. This proves artifact recovery without claiming that all offloaded graph facts enter model context.

Validation:

Official Grok Build 1.0.50 Linux x64: source ordinary chat and actual skill/discovery/analysis/artifact-query flows passed.
npm run check:fast passed.
Production-only package ordinary chat and large-result recovery also passed. All four source/package flows have verified empty descendant lists; a Linux run-token scan found no remaining client/server processes.

Native foreign configuration discovery is disabled; personal skill discovery uses an explicit isolated directory. The model declares a one-million-token context with synthetic inference/usage. Live xAI APIs, native Windows and default OS-home skill discovery remain unverified. No Hopper/Ghidra verification was performed; provider implementations and analysis contracts are unchanged.

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/morluto/rea/pull/1571 **Author:** [@N0zoM1z0](https://github.com/N0zoM1z0) **Created:** 10/10/2026 **Status:** ✅ Merged **Merged:** 10/10/2026 **Merged by:** [@N0zoM1z0](https://github.com/N0zoM1z0) **Base:** `main` ← **Head:** `test/grok-native-client-20261010` --- ### 📝 Commits (1) - [`6473c40`](https://github.com/morluto/rea/commit/6473c40ee7d62c45cde4603d2864f895c9a6acd7) test(clients): verify native Grok Build workflows ### 📊 Changes **3 files changed** (+560 additions, -0 deletions) <details> <summary>View changed files</summary> 📝 `docs/testing.md` (+20 -0) 📝 `package.json` (+1 -0) ➕ `scripts/verify-grok-client.mjs` (+539 -0) </details> ### 📄 Description REA already supports Grok Build configuration, but lacked a native client workflow check. Add an optional verifier using the official Grok Build CLI, its default lazy `search_tool`/`use_tool` interface and a loopback OpenAI completions/SSE model fixture. The lane guards setup writes, preserves unrelated settings/backups, checks idempotence, loads the entire skill and validates all 139 advertised input schemas through native discovery. When discovery is still partial, the native agent diagnoses REA registration and retries. A larger Unicode-path analysis produces about 738 KB of Evidence: Grok retains the complete JSON, and a native terminal query returns the selected export facts, subject and complete artifact digest to the next model request. The retained file is checked against REA's named output schema. This proves artifact recovery without claiming that all offloaded graph facts enter model context. Validation: - Official Grok Build 1.0.50 Linux x64: source ordinary chat and actual skill/discovery/analysis/artifact-query flows passed. - `npm run check:fast` passed. - Production-only package ordinary chat and large-result recovery also passed. All four source/package flows have verified empty descendant lists; a Linux run-token scan found no remaining client/server processes. Native foreign configuration discovery is disabled; personal skill discovery uses an explicit isolated directory. The model declares a one-million-token context with synthetic inference/usage. Live xAI APIs, native Windows and default OS-home skill discovery remain unverified. No Hopper/Ghidra verification was performed; provider implementations and analysis contracts are unchanged. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-17 21:12:31 +00:00