danielleit241
UserThe toolkit includes slash commands, sub-agents, skills, lifecycle hooks, shared rules, and project configuration. It tracks installed files with checksums so updates are repeatable and local edits are not overwritten silently.
Categories
Indexed Skills (22)
code-review
Use when reviewing a PR, commit, pending diff, or codebase area; receiving review feedback; checking a done/fixed/passing/complete claim; or preparing to commit, merge, ship, or hand off completed work. Produces findings-first verdicts, adversarial review, feedback triage, and verification gates. This is the only code review skill; other workflows reuse its verdict contract via the `code-reviewer` agent instead of inventing their own review logic.
frontend-design
Guidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults.
preview
Use when the user asks to preview, explain, visualize, ASCII-map, or create a self-contained HTML showcase for an idea, feature brief, file path, or plan directory before implementation. Builds grounded preview context with a read-only scout pass, routes only to explain, ascii, or showcase_html, and writes preview artifacts under visuals/. Do not use for implementation, review, shipping, or creating a new plan. Flag: --html.
problem-solving
Use when stuck after repeated failed attempts, when conventional approaches are not producing progress, when a problem needs reframing, when an elegant simplification may remove complexity, or when complex problems require systematic step-by-step reasoning with the ability to revise thoughts, branch into alternatives, or adjust initially unclear scope. Provides collision-zone thinking, inversion, pattern recognition, scale testing, simplification, and sequential thinking techniques.
collision-zone-thinking
Force unrelated concepts together to discover emergent properties - "What if we treated X like Y?" Use when conventional approaches feel inadequate and you need breakthrough innovation by forcing unrelated concepts together.
inversion-exercise
Flip core assumptions to reveal hidden constraints and alternative approaches - "what if the opposite were true?" Use when stuck on unquestioned assumptions or feeling forced into "the only way" to do something.
meta-pattern-recognition
Spot patterns appearing in 3+ domains to find universal principles. Use when noticing the same pattern across 3+ different domains or experiencing deja vu in problem-solving.
scale-game
Test at extremes (1000x bigger/smaller, instant/year-long) to expose fundamental truths hidden at normal scales. Use when uncertain about scalability, edge cases are unclear, or validating architecture for production volumes.
simplification-cascades
Find one insight that eliminates multiple components - "if this is true, we don't need X, Y, or Z." Use when implementing the same concept multiple ways, accumulating special cases, or complexity is spiraling.
skill-creator
Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.
strategic-compact
Use when a long session is approaching context pressure, crossing a major task or phase boundary, or recovering from a failed approach where old reasoning should be summarized before continuing.
using-skills
Use when starting any Claude Code session in this toolkit or before deciding whether a workflow skill, support checklist, routing skill, or technique skill applies. Injected by SessionStart to route tasks to the right skill before acting.
blindspot
Use when the user is unfamiliar with a repository area, asks what they may be missing, or needs unknown unknowns surfaced before making a decision. Gather read-only evidence through the `scout` agent, compare it with the user's framing, and return a Blindspot Report only. Do not design, plan, implement, or generate a brainstorm prompt.
brainstorm
Use when the user wants to clarify a vague idea, compare meaningful directions, or choose a product or technical approach before planning. Explore conversationally, gather repo or external evidence only when it changes the decision, and write the chosen direction to `brainstorm.md`. Do not create an implementation plan or edit product code.
cook
Use when the user asks to implement an approved `plan.md` or a clear bounded requirement. Execute the scoped change, verify it, inspect the resulting diff, and record meaningful decisions and verification in one `implement-notes.md`. Use scout, subagents, or independent review only when uncertainty, risk, or scope warrants them. Do not make commits, pushes, or releases implicit.
engineering-practices
Routes to six engineering support practices: testing strategy, security hardening, observability, migration safety, documentation/ADRs, and condition-based waiting. Use when planning tests, using --tdd, writing regression guards, or handling flaky tests; when work touches auth, authorization, user input, secrets, PII, payments, webhooks, file upload, SSRF-prone URL fetching, or external integrations; when production behavior needs logs, metrics, traces, alerts, or runbook evidence; when changing databases, schemas, data contracts, APIs, feature flags, or deprecating and removing old behavior; when public APIs, setup, commands, architecture, or decision rationale need durable docs or an ADR; or when tests rely on arbitrary sleeps, fixed delays, or timeout guesses instead of waiting for an observable condition.
fix
Use when repairing a concrete bug, regression, failing test, stack trace, or broken behavior. Diagnose only as far as needed to make a minimal verified repair, then write one `fixed.md` record. Use scout or independent review when the cause, blast radius, or risk is uncertain; do not impose incident ceremony on a local fix.
git-workflow
Routes git workspace and branch-integration practices. Use when starting feature work that needs an isolated workspace or git worktree, before executing a risky implementation plan, when the current workspace has unrelated changes that should not mix with the task, or when implementation is complete and verification passes and the branch needs a structured integration decision: merge locally, push and create a PR, keep the branch, or discard the work.
plan
Use when the user asks for an implementation plan, technical strategy, or phased approach before coding a non-trivial change. Produce one durable `plan.md` containing the design contract, ordered work, verification, risks, and status. Gather scout or research evidence only when uncertainty affects the plan; do not create separate phase files by default.
scout
Use when the user asks to scout a repository, git history, plan, diagnosis, research question, or pending diff before a decision or change. Select the appropriate mode and automatically dispatch the read-only `scout` agent; return its evidence report without editing or making the final implementation decision. Modes: --repo, --research, --diagnose, --review.
ship
Use when completed work is being prepared for merge, release, deployment, rollback, or user-facing handoff. Run a proportional live readiness checklist using the current diff, verification, review, docs, risks, and rollback needs. Do not create a readiness artifact; publishing, deploying, tagging, pushing, or destructive rollback actions always require explicit user approval.
source-grounding
Grounds implementation and research decisions in authoritative sources. Use when framework, library, API, CLI, cloud, model, protocol, or product behavior could be version-sensitive, recently changed, unfamiliar, or likely to be misremembered.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.