Instrument code with NDJSON logs to debug issues via hypothesis testing.
npx skills add https://github.com/millionco/expect --skill debug-agent-millionco
Please help me install this Agent Skill. Skill: debug-agent Source: https://github.com/millionco/expect/tree/main/.agents/skills/debug-agent Command: npx skills add https://github.com/millionco/expect --skill debug-agent-millionco
This Skill enables rapid, evidence-based debugging by collecting runtime logs, forming hypotheses, and guiding reproducible investigations to identify root causes without guessing.
Start a debugging session by running the agent with a reproducible bug and let the system generate runtime logs to begin the evidence-based loop.
High-intent search queries and answers about installing and using this skill.
Evidence-based debugging with runtime logs gathers in-session data to systematically identify root causes without guessing. It instruments code with lightweight NDJSON logs, generates 3-5 hypotheses per bug report, and evaluates them in parallel against live data to prove the root cause.
Finding a root cause with NDJSON logs involves generating 3-5 precise hypotheses per bug report and evaluating them in parallel using live runtime data. The process enforces phased hypothesis testing and verifiable proof before applying fixes, iterating until the root cause is proven with cited log lines.
The best way to systematically reproduce a bug is to use a structured reproduction plan supported by runtime log instrumentation. After evaluating hypotheses and applying fixes, post-fix verification confirms the issue is resolved, ensuring instrumentation can be removed safely only after proof is established.
To capture runtime logs for debugging across components, instrument your runtime code with lightweight NDJSON logs. This strict instrumentation records hypotheses and outcomes during an in-session investigation, providing verifiable evidence and cited log lines to guide reproducible debugging.
Yes, you can use automated hypothesis generation for debugging reproducible bugs. The system automatically generates 3-5 precise hypotheses per bug report and evaluates them in parallel using live data captured from in-session runtime logs, enforcing verifiable proof before applying fixes.
Strategic-AutomationAutomate browser-based exploratory QA testing with evidence and bug reports.
Jesse-L-MDiagnose software bug root causes through structured symptom collection and hypothesis testing.
TheBushidoCollectiveDiagnose software bugs through evidence collection and hypothesis testing.
backengineeringInstrument binaries and trace runtime behavior using LLVM infrastructure.
