Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →brycewang-stanford/software-engineering-research
Guide to software engineering research topics and methodologies
Passed 22 of 22 test cases with the skill, and 21 of 22 without it.
1 fixed21 passed either way0 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
A test case the skill fixed. Same model, same prompt, with and without the skill.
In a research paper discussing software metrics, I need to cite the original 1976 publication that introduced Cyclomatic Complexity. Who authored this metric?
| Checked | Without skill | With skill |
|---|---|---|
| Attributes cyclomatic complexity to McCabe. | Missed | Passed |
The instructions your model reads when the skill loads.
Navigate the landscape of software engineering research, including key subfields, methodologies, datasets, benchmarks, and top venues.
| Subfield | Key Topics | Major Venues |
|---|---|---|
| Software Testing | Test generation, fuzzing, mutation testing, flaky tests | ISSTA, ICST, ASE |
| Program Analysis | Static analysis, abstract interpretation, symbolic execution | PLDI, POPL, OOPSLA |
| Software Maintenance | Code refactoring, technical debt, code smells, evolution | ICSME, MSR, SANER |
| SE for AI/ML | ML pipeline testing, data quality, model debugging | ICSE-SEIP, FSE |
| AI for SE | Code generation, bug detection, program repair | ICSE, FSE, ASE |
| Distributed Systems | Consensus, fault tolerance, scalability, microservices | SOSP, OSDI, EuroSys |
| Cybersecurity | Vulnerability detection, malware analysis, privacy | IEEE S&P, CCS, USENIX Security |
| HCI in SE | Developer tools, IDE usability, code comprehension | CHI, CSCW, VL/HCC |
| Empirical SE | Mining repositories, developer surveys, controlled experiments | ESEM, MSR, TOSEM |
Testing a specific hypothesis with treatment and control groups:
Example: Does AI code completion improve developer productivity?
Design:
- Participants: 60 professional developers
- Treatment: IDE with AI code completion enabled
- Control: IDE with AI code completion disabled
- Task: Complete 5 programming tasks of varying difficulty
- Metrics: Task completion time, code correctness, lines of code
- Analysis: Mixed-effects linear model with participant as random effect
Threats to validity:
- Internal: Learning effect (counterbalance task order)
- External: Lab setting may not reflect real development
- Construct: "Productivity" operationalized as speed + correctness
test cases passed, vs 21 / 22 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.