| FMRS | 67 / 100 · C | 79 / 100 · B |
|---|
| Reliability | 11 / 20 | 12 / 20 |
|---|
| Security and permissions | 12 / 20 | 18 / 20 |
|---|
| Maintenance | 13 / 20 | 16 / 20 |
|---|
| Documentation | 17 / 20 | 18 / 20 |
|---|
| Setup experience | 14 / 20 | 15 / 20 |
|---|
| Best for | - Developers and support engineers doing Windows crash dump and driver/kernel triage
- Users of MCP clients (Claude Code, GitHub Copilot, Claude Desktop, Cursor, Windsurf, Cline) who want natural-language control of WinDbg
- Teams that need to manage and automate multiple cdb/kd sessions (dumps, remote, kernel) in parallel
- Environments that must filter sensitive data before debugging output leaves the machine
| - Engineering teams scaling AI coding agents while controlling spend
- Platform teams that need auditable, rollback-aware agent execution records
- Organizations using multiple agents (Claude, Codex, Gemini) under one governed flow
|
|---|
| Not for | - Non-Windows platforms: dump analysis needs cdb.exe and kernel debugging needs kd.exe
- Users expecting an automatic fix: it runs real debugger commands and lets the model reason about output
- Users without organizational permission to use AI tools in a managed enterprise environment
- Users who have not installed the debugging tools (cdb.exe / kd.exe)
| - Developers who just want a simple code-completion helper without run governance
- Teams unwilling to accept local JSONL run records with HMAC signing
- Environments without a supported adapter (Claude/Codex/Gemini CLIs, direct provider, or verifier-only)
|
|---|
| Required permissions | - Execute local processes on the Windows host (cdb.exe / kd.exe and the Python package)
- Read local dump files (.dmp/.mdmp/.hdmp) and target directories
- Open TCP, named-pipe, or COM connections for user-mode remote debugging
- Open KDNET, named-pipe, or serial connections for kernel debugging (waits for and breaks into the target machine)
- Optional: filter tool arguments and output via --filter-script; configure a custom CDB_PATH and symbol path via _NT_SYMBOL_PATH
| - Read/write access to the configured repo root (writable scope restrictable via --allow-path/--deny-path)
- Ability to invoke the chosen agent CLI (Claude/Codex/Gemini, etc.) and its underlying model account
- Local disk writes for run records, receipts, and share outputs (share/run-receipt., etc.)
- Execution of user-configured verifier commands (e.g. npm test)
|
|---|
| Risks and side effects | - Closing a kernel session resumes the target machine, which may affect it if it is processing critical state
- Attaching to live (remote or kernel) sessions interrupts the target and may affect production or shared environments
- Without --filter-script, sensitive data in dumps may leave the machine as part of tool output
- Using MCP and AI tools in enterprise environments may be restricted by organizational policy; confirm permissions first
- The default symbol path uses the Microsoft symbol server, so debugging may access external networks
| - Verifier commands are executed; poorly chosen verifier commands carry security risk despite preflight policy checks
- Cost and token figures may be estimates (labeled with provenance) and should not be treated as settled accounting
- Receipt integrity must be verified before use as trustworthy evidence; missing rollback evidence is flagged as EVIDENCE_BOUNDARY
- Policy defaults come from martin.config.yaml and can be overridden by CLI flags; misconfiguration can loosen safety boundaries
|
|---|
| Supported clients | Claude Code, VS Code, Claude Desktop, Cursor, Windsurf, Cline | Claude Code, Codex, Gemini CLI, Cursor, VS Code, GitHub Copilot, Continue |
|---|
| Tools | 10 | 0 |
|---|