How do I install the Debugui LLM Reset skill?

Run npx skillmds@latest add xilinx/debugui-llm-reset in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Debugui LLM Reset safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Debugui LLM Reset?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Debugui LLM Reset free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Debugui LLM Reset?

xilinx (@xilinx) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Debugui LLM Reset by xilinx — Agent Skill

Name: Debugui LLM Reset
Author: xilinx

Debugui LLM Reset

Diagnoses embedded debug UI LLM context loss and replies that stop mid-sentence during Connect, board changes, or target retargeting. Use when switching targets starts a new Claude session, a transcript ends with session end, or live tools keep using the previous board.

xilinx 4d2e8e1 2.1 KB Updated Sep 17, 2026

File contents

Debug UI LLM Reset

Diagnose

Find the latest matching llm_*.log under the selected app's worklocal.
Inspect the final marker:
- [watchdog: ... turn abandoned] or [watchdog: ... claude exited] means the daemon hard-stopped the turn (process gone, or 600s silence).
- While the claude subprocess is still running, the UI keeps polling silently with the working indicator — no interim warning. Multi-tool turns can go minutes between stream-json lines while the model processes tool results.
- [session end] means llm_reset() terminated the process.
- No marker plus closed output means the Claude process exited independently.
Compare the transcript target with the next transcript and the daemon process timeline.
Check whether /settarget ran while the answer was active.

Retarget invariant

Changing the board target or device must not restart Claude. /settarget updates backend_status.json; the existing aiemcp process reads that file before each tool call and refreshes its in-process AieGdb configuration.

Preserve all three:

The Claude subprocess and conversation.
The browser transcript and pending response.
The in-process AieGdb object and tile/channel scope.

Refresh the AieGdb target, device, start column, AIE version, _reg_read, and _passthrough. Only explicit New chat or process recovery may advance llm_generation.

Verify

Run:

python3 -m unittest src/tool/debug/test_schedule_debug_server_llm.py
python3 -m unittest src/tool/debug/test_aiemcp_retarget.py
python3 -m py_compile src/tool/debug/schedule_debug_server.py src/tool/debug/schedule_view.py

Confirm both same-target reconnects and real board changes preserve the active conversation, while the next aie_exec call uses the new target and device.

---name: debugui-llm-resetdescription: Diagnoses embedded debug UI LLM context loss and replies that stop mid-sentence during Connect, board changes, or target retargeting. Use when switching targets starts a new Claude session, a transcript ends with session end, or live tools keep using the previous board.---# Debug UI LLM Reset## Diagnose1. Find the latest matching `llm_*.log` under the selected app's `worklocal`.2. Inspect the final marker: - `[watchdog: ... turn abandoned]` or `[watchdog: ... claude exited]` means the daemon hard-stopped the turn (process gone, or 600s silence). - While the claude subprocess is still running, the UI keeps polling silently with the working indicator — no interim warning. Multi-tool turns can go minutes between stream-json lines while the model processes tool results. - `[session end]` means `llm_reset()` terminated the process. - No marker plus closed output means the Claude process exited independently.3. Compare the transcript target with the next transcript and the daemon process timeline.4. Check whether `/settarget` ran while the answer was active.## Retarget invariantChanging the board target or device must not restart Claude. `/settarget`updates `backend_status.json`; the existing `aiemcp` process reads that filebefore each tool call and refreshes its in-process `AieGdb` configuration.Preserve all three:1. The Claude subprocess and conversation.2. The browser transcript and pending response.3. The in-process `AieGdb` object and tile/channel scope.Refresh the AieGdb target, device, start column, AIE version, `_reg_read`, and`_passthrough`. Only explicit **New chat** or process recovery may advance`llm_generation`.## VerifyRun:```bashpython3 -m unittest src/tool/debug/test_schedule_debug_server_llm.pypython3 -m unittest src/tool/debug/test_aiemcp_retarget.pypython3 -m py_compile src/tool/debug/schedule_debug_server.py src/tool/debug/schedule_view.py```Confirm both same-target reconnects and real board changes preserve the activeconversation, while the next `aie_exec` call uses the new target and device.

Frequently asked questions