balandongiv
35 verified skills
code-readability-best-practices
Review and refactor code for top-down readability by reorganizing functions, grouping related helpers, and rewriting noisy comments without changing behavior.
tools
good-function-design
Write small, readable, testable Python functions with clear names and explicit inputs and outputs.
development
handoff
Create operational handoff markdowns that capture what was completed, what is stopped, what remains pending, which artifacts matter, and which commands the next agent should run next.
data-ai
beginner-moodle-programming-question
Generate beginner-friendly Moodle programming questions with tags clarification, partial answer preload scaffolding, auto-created test cases, actionable hints, and feedback about common beginner mistakes.
testing
repo-readme-writing
Write clear, structured, beginner-friendly README files for GitHub repositories that explain purpose, setup, and usage.
documentation
beginner-moodle-programming-question
Generate beginner-friendly Moodle programming questions with tags clarification, partial answer preload scaffolding, auto-created test cases, actionable hints, and feedback about common beginner mistakes.
testing
experiment-runbook-discipline
Plan, launch, monitor, and document long-running experiments or validation sweeps with smallest-real-data smoke scopes, fresh experiment prefixes, live status artifacts, rolling logs, and promotion to full runs only after explicit pass criteria are met.
testing
good-function-design
Write small, readable, testable Python functions with clear names and explicit inputs and outputs.
development
python-class-and-filename
Create focused Python classes and choose matching snake_case module filenames when adding or refactoring class-based code.
development
subject-outlier-review
Review per-subject performance to identify likely outliers, distinguish bad data from difficult but valid cases, and document whether subject exclusion is justified before any filtered rerun.
documentation
code-readability-best-practices
Review and refactor code for top-down readability by reorganizing functions, grouping related helpers, and rewriting noisy comments without changing behavior.
tools
good-description-writing
Write precise, specific descriptions for agent skills, tools, and functions so that AI agents can route to them accurately.
tools
handoff-resume
Resume repository work from an existing handoff markdown by reading it first, executing pending tasks in order, and marking completed tasks with `Done? = Yes` only when concrete evidence exists.
tools
handoff-resume
Resume repository work from an existing handoff markdown by reading it first, executing pending tasks in order, and marking completed tasks with `Done? = Yes` only when concrete evidence exists.
tools
experiment-runbook-discipline
Plan, launch, monitor, and document long-running experiments or validation sweeps with smallest-real-data smoke scopes, fresh experiment prefixes, live status artifacts, rolling logs, and promotion to full runs only after explicit pass criteria are met.
testing
implementation-aligned-planning
Turn ambiguous, half-baked, or outdated plans into living planning docs aligned with code, config, paths, naming contracts, execution flow, caches, EDA outputs, and debugging navigation.
development
hyperparameter-search-strategy
Choose efficient hyperparameter search strategies for finding optimal parameter sets or parameter pairs, favoring random search, Bayesian optimization, successive halving, evolutionary methods, or population-based training over brute-force grids. Use when an experiment, detector, or training pipeline must tune parameters under compute and evaluation-cost constraints.
testing
good-description-writing
Write precise, specific descriptions for agent skills, tools, and functions so that AI agents can route to them accurately.
tools
handoff
Create operational handoff markdowns that capture what was completed, what is stopped, what remains pending, which artifacts matter, and which commands the next agent should run next.
data-ai
implementation-aligned-planning
Turn ambiguous, half-baked, or outdated plans into living planning docs aligned with code, config, paths, naming contracts, execution flow, caches, EDA outputs, and debugging navigation.
development
hyperparameter-search-strategy
Choose efficient hyperparameter search strategies for finding optimal parameter sets or parameter pairs, favoring random search, Bayesian optimization, successive halving, evolutionary methods, or population-based training over brute-force grids. Use when an experiment, detector, or training pipeline must tune parameters under compute and evaluation-cost constraints.
testing
manuscript-results-curation
Curate manuscript results directly in LaTeX from experiment artifacts, including tables, plots, graphs, images, and detailed scientific interpretation tied to concrete outputs.
testing
mne-blink-report-html
Generate an MNE Report HTML that plots blink windows one by one from epoch-based blink data, including true positives, false negatives, false positives, or all blinks, with configurable padding before and after each blink. Use when the user wants the same kind of per-blink sanity-check plotting as the Strategy C Approach 3 report or asks for reusable blink-report plotting from MNE epoch files and blink tables.
development
manuscript-results-curation
Curate manuscript results directly in LaTeX from experiment artifacts, including tables, plots, graphs, images, and detailed scientific interpretation tied to concrete outputs.
testing
python-class-and-filename
Create focused Python classes and choose matching snake_case module filenames when adding or refactoring class-based code.
development
template-skill-name
One-sentence description of what this skill does and when to use it.
tools
intellij-line-debugging
Turn complex execution paths into line-by-line IntelliJ IDEA debugging flows with serial tutorial entrypoints, smallest-real-input reruns, exact breakpoint order, and deliberate stepping into editable local dependencies.
development
real-data-validation-promotion
Validate data or ML pipelines on the smallest real dataset scope first, then promote to staged batches and full sweeps with the same code path, editable local dependencies, artifact checks, and honest residual-risk reporting.
development
real-data-validation-promotion
Validate data or ML pipelines on the smallest real dataset scope first, then promote to staged batches and full sweeps with the same code path, editable local dependencies, artifact checks, and honest residual-risk reporting.
development
strategy-impact-log
Record and track strategy proposals, code changes, performance metrics, issues encountered, and their cumulative effects on final results to maintain a durable audit trail of what was tried, what worked, and what didn't.
development
intellij-line-debugging
Turn complex execution paths into line-by-line IntelliJ IDEA debugging flows with serial tutorial entrypoints, smallest-real-input reruns, exact breakpoint order, and deliberate stepping into editable local dependencies.
development
subject-outlier-review
Review per-subject performance to identify likely outliers, distinguish bad data from difficult but valid cases, and document whether subject exclusion is justified before any filtered rerun.
documentation
strategy-impact-log
Record and track strategy proposals, code changes, performance metrics, issues encountered, and their cumulative effects on final results to maintain a durable audit trail of what was tried, what worked, and what didn't.
development
repo-readme-writing
Write clear, structured, beginner-friendly README files for GitHub repositories that explain purpose, setup, and usage.
documentation
template-skill-name
One-sentence description of what this skill does and when to use it.
tools