by engasnm111
lnwjud — local AI-agent runtime & MCP gateway
# Add to your Claude Code skills
git clone https://github.com/engasnm111/lnwjudLast scanned: 8/26/2026
{
"issues": [],
"status": "PASSED",
"scannedAt": "2026-08-26T04:39:14.313Z",
"npmAuditRan": false,
"pipAuditRan": true,
"promptInjectionRan": true
}See how lnwjud compares with popular alternatives.
lnwjud is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by engasnm111. lnwjud — local AI-agent runtime & MCP gateway. It has 157 GitHub stars.
Yes. lnwjud passed SkillsLLM's automated security scan — a dependency vulnerability audit plus prompt-injection heuristics — with no high-severity issues. You can read the full report in the Security Report section on this page.
Clone the repository with "git clone https://github.com/engasnm111/lnwjud" and add it to your Claude Code skills directory (see the Installation section above).
lnwjud is primarily written in TypeScript. It is open-source under engasnm111 on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh lnwjud against similar tools.
No comments yet. Be the first to share your thoughts!
⚠️ Third-Party Software Notice
This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM is an informational directory and does not control or maintain the underlying repository.
Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.
Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.
v4.62.1 is the current source/release-candidate version. The latest public build is always available from GitHub Releases. Development artifacts from dev are for testing before the public release is published.
tools/call now bypasses the SDK layer's eager output-schema validation so a real isError: true result reaches lnwjud unchanged; successful structured output and the MCP result envelope are still validated, with regression coverage for the exact SDK-layer failure path reported in Issue #53.image content without duplicating the full Base64 payload into text/structured metadata, preventing successful captures from being lost behind oversized tool-result JSON.mcp_call now preserves child MCP image and text content blocks instead of flattening the child CallToolResult into JSON text, so screenshots from Serena/custom MCP servers can reach the model as images.vision:capture_window now accepts natural app.name selectors, prefers a visible non-minimized matching HWND when apps expose several helper windows, and reports minimized/hidden-window states directly instead of collapsing them into generic Operation failed errors.chatgpt.com OAuth callbacks—including newly created Plugin/App callbacks shaped as /connector/oauth/<redirect_id>—complete DCR + Authorization Code + PKCE through a one-time browser handoff to an ephemeral 127.0.0.1 Desktop approval listener, so workspace members press Connect without copying a PIN. The public gateway never grants trust from the callback URI alone; the 6-digit PIN remains only as a fail-closed fallback for non-ChatGPT OAuth clients.MODE / WORK status card.--url on later starts/updates, refusing to silently switch the ChatGPT endpoint if it drifts. Users do not need to buy/register their own domain; custom domains remain optional, and re-saving the ngrok authtoken intentionally resets the remembered origin for account/domain changes.2026-07-28, while lnwjud's own inbound MCP contract remains unchanged. External MCP definitions saved in Settings are applied live, so adding or changing Serena/custom servers does not require restarting lnwjud.OFF / LITE / FULL / ULTRA modes default to OFF, resolve Current Goal > Workspace > Global, and use exact bundled Ponytail skills rather than relying on discovery ranking. Active modes require the exact bundled primary skill before code mutation; FULL/ULTRA durable coding goals additionally require a fresh bundled Ponytail review before completion. Full Bypass does not bypass this correctness gate, while explicit session suppression remains available without changing persisted policy.(platform, architecture) tuples. Current pins include OpenAI tunnel-client 0.0.14, ripgrep 15.2.0, and Windows Poppler 26.07.0-0, with checksum/provenance/version checks before packaging.pdftotext where supported; the bundled Poppler auto-installer remains Windows x64-only and is hidden on unsupported hosts.