imnotdev25

@imnotdev25 source repo

6 skills
0 followers
8 hours ago last updated

▌
Keshav Three Pass Reading · imnotdev25 bundle
S. Keshav's three-pass paper reading method with AI-assisted passes and structured outputs (Summary, Problem, Method, Results, Takeaways). Teaches when to skim, when to dig deep, and how to use AI as a reading companion without outsourcing judgment.

0
installs
▌
File Management · imnotdev25 bundle
Workspace Hygiene

0
installs
▌
Trace Code Work · imnotdev25 bundle
Maintain auditable, dated coding-session records for implementation, debugging, refactoring, and code review. Use whenever Codex changes source code, tests, configuration, schemas, dependencies, build or deployment code, or resumes a coding project that has prior session logs. Create and update code-logs/YYYY-MM-DD/DECISIONS.md, FLOW.md, EXPLORE.md, and QUIZ.md; record why approaches and libraries were chosen, trace caller-to-callee execution paths and modified segments, remember explored files/functions, and enforce a user recall quiz before further project changes after a long completed session.

0
installs
▌
Research Pipeline · imnotdev25 bundle
Drives a HUMAN-LED, multi-stage research workflow (ideation -> literature review -> problem framing & novelty -> hypothesis & planning -> experimentation -> validation -> reporting) in which Claude acts strictly as an implementation engine while the researcher does the thinking. Use this whenever the user wants to start, structure, or advance a research project, paper, thesis chapter, or study. Trigger it for things like "I have a research idea", "help me frame this", "find / screen papers on X", "write a problem statement", "is this novel?", "help me design an experiment", or "let's write a paper" -- and especially when the user wants a rigorous, fully provenance-tracked process rather than a quick answer. The skill keeps a workspace where every link, term, assumption, inference, and decision is written down as it happens, and it counter-questions the user instead of thinking for them. This is the opposite of an autonomous "AI scientist": the human owns every intellectual commitment; Claude owns retrieval, s

0
installs
▌
Generate Knowledge · imnotdev25
Generate a self-contained markdown knowledge base (functions, classes, variables, flow) for ANY codebase into a <repo>/.agents-kg/ folder. Use at the start of work on an unfamiliar repo, when .agents-kg/ is missing or stale, or when asked to 'create the docs', 'generate knowledge', 'build a code reference', 'make a knowledge base', or 'document the whole repo'. Dynamically scouts the repository structure and fans out parallel agents, one per main area. Trigger: /generate-knowledge

0
installs
▌
Engineering Discipline · imnotdev25 bundle
Enforces a strict engineering-discipline workflow for any coding project — docs-first, modular code (100-LOC file ceiling), strict types, fail-loud errors, semantic logging, behavior-tests, dependency tracking, dated changelogs, and infra separation. Use this skill whenever the user types `/init-project` (bootstrap a new project) or `/existing-project` (onboard / audit a current repo), and also whenever the user asks to start a new codebase, scaffold a repo, set engineering standards, audit project structure, build a dependency map, write a FILE_TREE.md or DEPENDENCY.md, or work on a project under these rules. Trigger even when the slash commands aren't used verbatim — phrases like "start a new project", "set up a repo properly", "make this codebase clean", or "I want to add a feature to my project" should pull this skill in. The skill always interviews the user before producing code or docs; never assume tech stack, never skip the docs-before-code step.

0
installs

6 published skills