petalas-skills

Send independent adversaries to break a change or design, keep only findings proven by a run, and report them without changing the code.

1 skill · 1 command

Design non-trivial code from caller usage, types, signatures, and module boundaries before implementation, then implement against the chosen sketch when requested.

1 skill · 1 command

Generate several independent candidates for the same non-trivial artifact, judge them against a concrete rubric, and synthesize one verified result.

1 skill · 1 command

Design and execute an evidence-driven local workflow for a large task when no focused playbook fits.

1 skill · 1 command

Create or update a personal mode skill from the user's stated preferences and workspace-scoped conversation history.

1 skill · 1 command

Vet a perf measurement (limiter, tuning, limits, errors, repeatability, relevance, and whether the work happened) before you report or act on it. Use when you run a benchmark or report a speedup or regression you measured.

1 skill · 1 command

Find what a change could break outside its diff, identify the safety facts it depends on, and prove those facts with executable evidence when practical.

1 skill · 1 command

Restate the last message in plain human language, with no jargon.

1 skill · 1 command

Create focused Git commits that follow Conventional Commits without attributing authorship or co-authorship to AI tools.

1 skill · 1 command

Record host-neutral model and reasoning preferences for internal agent roles using only choices available in the current environment.

1 skill · 1 command

Find the mistakes agents keep repeating in this repo and make each one impossible. Try architecture first, then types, then a lint whose error names the fix, then a test, and write docs last. Prove each check fails on a real past mistake. Repeat this each time the user corrects you.

1 skill · 1 command

Generate a project-local skill that formally verifies the code where a bug costs the most and tests reach the least, in any language, with whichever verification tool fits. Use for "model-check this race", "prove this invariant", or when concurrency, retries, state machines, parsers, or unsafe code need more than tests.

1 skill · 1 command

Create a project-local skill that drives an app through its real user surface and captures proof.

1 skill · 1 command

Use when starting a software task that needs rigor: a bug, feature, refactor, investigation, performance issue, or any multi-step change to code that will be kept. Picks one playbook, routes to the matching principle skills and power-of-ten, and proves the result on the real artifact. Skip for casual questions, throwaway scripts, or when the user opts out.

1 skill · 1 command

Explain code plainly by combining how it works with the evidence for why it has that shape. Use for teaching a change, subsystem, or unfamiliar code path.

1 skill · 1 command

Generate or edit images through a ChatGPT subscription, prepare reusable prompts, and inspect saved artifacts. Use when the user names GPT Image 2.5 or asks for image generation through their ChatGPT plan. Checks exact-model requirements before generation; subscription model selection is not guaranteed.

1 skill · 1 command

Explain how code works: subsystem architecture, runtime flow, ownership, placement, and layering. Use why for historical motivation.

1 skill · 1 command

Restate in your own words what you think the user's goals are and what problem they're trying to solve.

1 skill · 1 command

Catch the user up on what changed in this chat since their last message, in plain words and small visuals. List every item waiting on them with what they need to decide it from the reply alone.

1 skill · 1 command

Catch the user up like ketchup, with a short subtitled video of slides for what happened and the waiting calls as text under it.

1 skill · 1 command

Audit one project's knowledge system on request: check for a routing table that maps file patterns to docs, a When You Struggle capture process, and a project-specific LEARNINGS.md; report pass or fail, then set the system up or migrate an older layout only with approval. Never runs proactively.

1 skill · 1 command

Scan a chosen directory tree for projects with agent instructions files, run the knowledge-audit quick check on each, and report which are current, need an update, or need setup. Read-only discovery across projects; it never edits any project without a separate explicit instruction.

1 skill · 1 command

Bring a project's formal verification back in line with the code after the code changes: find the commits that touched verified code, update the model, harness, or properties to describe the code as it is now, rerun, and say whether each failure is drift in the verification or a bug in the code. Use for "is the model still true to the code", or after merges and refactors that touched verified code without rechecking it.

1 skill · 1 command

Audit a project verification skill against source and every mapped user-facing feature, then apply proven local corrections.

1 skill · 1 command

Apply when writing or changing code that will be committed, tested, or run more than once. Hold every touched path to the ten language-agnostic rules, fix encountered violations within scope, verify affected behavior, and document justified deviations. Skip throwaway scripts, one-off commands, prototypes, and pure config or documentation edits; honor narrower project scoping.

1 skill · 1 command

Apply when two or more fixes that share one premise have failed the same gate. Write the premise down and test it directly before you write another fix that assumes it.

1 skill · 1 command

Apply when wiring validation, error handling, or framework adapters. Concentrate guards at system boundaries (CLI, config, network, external APIs); trust internal types and keep business logic in pure functions.

1 skill · 1 command

Apply to any non-trivial work, not just bulk work: edits, migrations, analyses, checks. Build the tool that does it or proves it (codemod, script, generator, or a skill your subagents follow) instead of working by hand. The tool is the artifact a reviewer can rerun.

1 skill · 1 command

Apply when you catch yourself writing the same instruction a second time, or notice a recurring correction. Encode the rule as a lint, metadata flag, runtime check, or script instead of more text.

1 skill · 1 command

Apply when facing a novel UI interaction or architectural decision with no precedent in the codebase. Build 2-3 competing prototypes and compare side by side before choosing.

1 skill · 1 command

Apply when product, UX, or feature-scope tradeoffs come up. Choose user delight over implementation convenience; ship fewer polished features over more rough ones.

1 skill · 1 command

Apply before you measure, and before you trust, report, or act on a number you measured: a speedup, a regression, a throughput, a latency, or an eval result. Find what limits it, and rule out that it measured something other than the work you think.

1 skill · 1 command

Apply when debugging. Trace each symptom back to the first place the system went wrong and fix it there. A guard is a fix only when the state it handles is legitimate.

1 skill · 1 command

Apply before writing logic: choosing core types and data structures, sequencing scaffold-vs-feature work, asking what concurrent actors share. Get the data structures right so downstream code becomes obvious.

1 skill · 1 command

Apply when context is filling up: large outputs, long files, repeated reads, fan-out planning. Route bulk to subagents; keep summaries in the main thread, not raw payloads.

1 skill · 1 command

Apply when refactoring, evaluating diff size, or tempted to add abstractions, layers, or signal threading. Bias toward deletion and the smallest change that solves the problem.

1 skill · 1 command

Apply when reversible local work can proceed without a permission pause. Keep agents autonomous inside the workspace while reserving external communication and consequential actions for the user.

1 skill · 1 command

Apply when designing commands, lifecycle steps, or processing loops that run amid crashes, restarts, and retries. Converge to the same end state regardless of partial prior runs.

1 skill · 1 command

Apply when reviewing or shaping code that's hard to trace. Count layers between question and answer, and hidden state in the reader's head; collapse one-caller wrappers and shrink mutable scope.

1 skill · 1 command

Apply when writing stateful logic, or when code branches a lot or repeats a shape assumption across files. Encode the domain in a structure instead of scattered conditionals.

1 skill · 1 command

Apply during planned rewrites and migrations with explicit phase boundaries. Converge on the target architecture; don't preserve smooth intermediate states with throwaway compatibility code.

1 skill · 1 command

Apply after completing a task, before declaring done. Verify against the real artifact (run the feature, read the actual value, inspect the diff), not a proxy, self-report, or 'it compiles.'

1 skill · 1 command

Apply when integrating a new requirement into an existing design. Redesign as if the requirement had been a foundational assumption from day one, instead of bolting it on.

1 skill · 1 command

Apply when introducing a new internal API while old callers still exist. Migrate callers and delete the old API in the same wave instead of preserving compatibility layers.

1 skill · 1 command

Apply when concurrent actors might write to the same file, branch, key, or state object. Eliminate the sharing first; serialize structurally only when one shared writer is a real invariant.

1 skill · 1 command

Apply to multi-step work such as sweeps, migrations, and runs of similar edits. Break work into small units that each end in a verifiable state, check each before the next, and order authorized delivery so the sequence proves itself.

1 skill · 1 command

Apply when sequencing an addition, refactor, or rewrite. Remove dead code, redundant validators, and stub references first, then build on the simpler base.

1 skill · 1 command

Apply when you write, change, or keep a test. Assert what the code's users can observe, so that the test fails when that behavior breaks and keeps passing when only the internals change.

1 skill · 1 command

Apply when designing types, reviewing a function signature, or writing code in any statically-typed language. Make illegal states unrepresentable, brand semantic primitives, parse external data at boundaries, exhaust variants, derive from authoritative schemas.

1 skill · 1 command

Reconstruct recent solo-development context from the current host's accessible task history, repository history, local docs, and live workspace state, then return a concise current-state brief.

1 skill · 1 command

Review the active session through independent lenses and propose durable skill improvements for user approval.

1 skill · 1 command

Use when the user asks for a regression test, or when a bug has an obvious cheap local test seam. Do not use for general TDD or test-first feature development.

1 skill · 1 command

Review source-code comments for noise, stale explanations, suppressions, and constraints that should be encoded in code, then apply only authorized local cleanups.

1 skill · 1 command

Find the highest-value behavior-preserving refactor that can be safely bounded, protected by tests or equivalent evidence, and validated.

1 skill · 1 command

Make a task deeper, not bigger, then rewrite the prompt.

1 skill · 1 command

Keep a local, reviewable TSV decision trail for long-running, unattended, or multi-phase work.

1 skill · 1 command

Check a TLA+ model of a system that moves between states, such as a product flow, a protocol, or a service, and draw it as a Mermaid statechart. The report gives the shortest path to each state that breaks a rule. Use to show a bug as a path through states.

1 skill · 1 command

Fan out independent subagents across parallel slices or races, drain every required result, and return one consolidated local report.

1 skill · 1 command

Layered technical-writing standard: Diataxis structure, Google developer style sentences, STE instruction rules, Global English syntax. Use for $technical-writing or when writing or reviewing docs, RFCs, readmes, PR descriptions, or commit messages.

1 skill · 1 command

TypeScript best practices. Use when reading or editing any TypeScript or JavaScript file (.ts, .tsx, .js, .mjs, .cjs).

1 skill · 1 command

Cut AI tells from any writing. Use when writing or editing prose a person will read.

1 skill · 1 command

Investigate why code has its current shape using read-only repository history, issue tracking, user-scoped documents, observability, error tracking, and product data.

1 skill · 1 command

Audit git worktrees and local development caches, then remove only targets proven unused and recoverable.

1 skill · 1 command