Claude Skills by saadshahd

Use when adding or changing a skill, hook, or runtime file in this repo — including deciding which unit a new capability should be.

Rewrite a rule so anyone applies it the same way. Use when a stated rule or claim takes the reader's judgment to apply.

Turn words with two readings into one confirmed ask. Use when the user's words admit two readings that would build different things.

Plan the skills an ask needs, invoke each as its condition arrives. Use when the ask has multiple aspects that different skills carry, or the user named multiple skills.

Build the thing the user can't specify, choices surfaced. Use when the user needs something to react to before they can choose.

Draw out what the user holds but can't say. Use when the user knows something they cannot yet express.

Put one settled thing in the user's head, plainly. Use when the user does not hold something already settled.

Freeze outside-the-repo state into one snapshot. Use when the work depends on the current state of a service, database, queue, third party, or live logs.

Make the user defend a proposal they committed to. Use when a proposal is committed but its decisions are undefended.

Find the route to guaranteed failure, negate it. Use when a goal keeps not being reached head-on.

One committed verdict on something that exists, cited. Use when something exists and a claim about it needs a verdict.

Extract movement from a statement known to be false. Use when nobody can see past an obvious answer.

Re-derive something written into the shortest version that still covers everything. Use when something already written should get shorter and nothing it covers can be lost.

Restate the problem off its forced framing. Use when a framing feels forced and every option inside it loses.

Spawn agents and verify their output — route work, never do it. Use when the work splits into pieces agents can carry and the session should route and check, not execute.

Restate the problem in a distant domain's terms. Use when a problem circles inside its own vocabulary.

Define how unsupervised work is judged before it runs. Use when work will run without a human watching, optimizing toward something.

Assemble cases from context, induce the widest sound rule. Use when a real case shows a class no rule yet decides, or a rule decides only the cases it came from.

Drop the jargon and speak plainly, from this message on.

Shape output for a human working through agents. Use on every response, out to and including casual conversation, even when the user did not ask for brevity.

Use when the user asks to explain or understand a code change, diff, branch, or PR — or to catch up on a change an agent built before reviewing or shipping it.

Use when the user wants knowledge delivered to an audience — slides, a deck, a diagram, an explainer for others, teaching or talk material — or shares existing content that isn't getting through.

Use when writing prose a human will read — docs, READMEs, PR descriptions, commit messages, code comments, chat replies, explanations. Not for code, identifiers, API names, error strings, or established technical terms.

Grind on one goal concurrently in a Solo workspace

Use when the user wants to understand a system by interacting with it themselves — asks for a debugger, visualizer, simulator, stepper, playground, or command center as an aid to their own understanding — or when reading and explanation haven't built intuition for how something behaves. Not for user-facing product features (a docs playground, a UI component, a customer demo).

Use in a repo that has sound taste rules installed (a `.claude/sound/` directory), before writing or editing code — at the start of a coding task, and again when the work reaches a surface it was not planned to touch.

Use when reviewing a staged diff or a commit range for soundness against the project's taste rules, including a pre-commit pass or an explicit "sound review".

Use when installing or refreshing sound taste rules in a project — an explicit "sound setup", starting work in a repo with no .claude/sound/, or a taste-not-defined nudge. Also use to re-run after a stack change (new framework, new package in a monorepo).

Implement tasks from an OpenSpec change. Use when the user wants to start implementing, continue implementation, or work through tasks.

Archive a completed change in the experimental workflow. Use when the user wants to finalize and archive a change after implementation is complete.

Enter explore mode - a thinking partner for exploring ideas, investigating problems, and clarifying requirements. Use when the user wants to think through something before or during a change.

Propose a new change with all artifacts generated in one step. Use when the user wants to quickly describe what they want to build and get a complete proposal with design, specs, and tasks ready for implementation.

Assesses team fitness and composes agent teams. Use when "set up a team", "team for this", "should I use agents", "design a team", "how many agents", "agent team".

Generate a project-level CLAUDE.md from stack detection and user-selected rule categories. Use when starting a new project, onboarding a repo, or when the user says "seed claude.md", "create project rules", "set up CLAUDE.md", "configure this project for me", or wants to establish coding conventions.

Use when asking "code like [expert]", "what would [expert] say", "idiomatic", "best practice", "panel", "debate", or needing domain guidance. Triggers on expert names, style requests, tradeoff questions, or "stuck on".

Turn rough ideas into iron-clad work orders. Use when request is vague like "add a button", "make it better", "fix the thing". Triggers on ambiguous or underspecified requests.

Start autonomous iteration loop. Triggers on "loop", "keep going", "continue until done", "implement feature", "fix all", "loop status", "cancel loop".

Bridge WHAT (intent) to HOW (implementation). Use when spec is clear but approach is not. Triggers on "shape this", "how should I build", "implementation approach".

Structured thinking framework with session strategy. Auto-applied to every task. Detects session type (Build/Debug/Plan/Reflect) and asks engagement level (Autonomous/Collaborative/Guided). Triggers on "how confident", "verify this", "think through", "what could go wrong".