agenticat

Captures the current session into a continuation prompt for a fresh one, or resumes from a pasted handoff. Use when context is running low, when ending a session mid-task, or when told to 'hand off' or write a handoff. `/handoff reusable` instead maintains a persistent multi-session task-runner state file (`docs/handoff-state.md`) that a fresh session reads, advances a few tasks via agent pairs, and re-saves.

Authors and normalizes tasks into `docs/todo.md`. Use when writing a todo list, capturing audit findings, or turning loose notes into pickup-cold-ready tasks.

Reconciles `README`, `docs/*`, `CLAUDE.md`, and agent/skill/plugin prompt files with what the code actually does: fixes stale, missing, or overpromising claims and verifies every quoted command/flag/path. Use after a change alters documented behavior, when a tool's output shape or fields change, or to sweep all docs.

Tunes a Rust `Cargo.toml` or `.cargo/config.toml`: profiles, features, workspaces, dependency hygiene. Use when cutting compile time or binary size, trimming deps, or configuring a workspace.

Boots and drives Android AVDs and iOS simulators from the CLI (adb, `xcrun simctl`, Flutter `integration_test`, Alchemist goldens). Use when running headless app tests, verifying screenshots, or debugging emulator boot/GPU issues.

Suggests project/crate/plugin names and checks availability across registries (crates.io, npm, PyPI, AUR, GitHub, domains; needs `bun`). Use before naming, renaming, or publishing anything.

Diffs a program against a reference or spec and writes the gaps as tasks. Use when matching, catching up to, reaching parity with, or surpassing another tool. Writes the gap list into `docs/todo.md`.

Audits a skill (or all skills) for routing gaps: it holds the right answer but buries it where an agent won't reach it. Targets boundary/negative probes (can it do X, what can't it do, is Y possible) that route to the how-to and miss the answer. Use after authoring or editing a skill, or to sweep all skills.

Playwright toolkit for testing local web apps: write Python scripts to navigate, click, fill forms, capture screenshots, and read browser console logs against a running dev server or static HTML. Use when verifying frontend behavior or debugging UI for a local webapp.

Askama Rust templating reference (0.16).

Maud Rust `html!` macro reference (0.27).

Modern-ratatui knowledge pack for writing, refactoring, or reviewing Rust TUI code on ratatui or crossterm. Use when touching TUI widgets, layout, styling, or event loops; writing TestBackend render tests; or resolving ratatui version upgrades and API questions.

Defensive Bash and shell script quality: strict-mode `set` flags, traps, idempotent mutation of live systems (deploy, firewall/sshd/sudoers edits, systemd units, ssh remote-exec), ShellCheck rc-file and `# shellcheck disable=` discipline, bats-core tests (`.bats`, `run`, PATH-stub mocking). Use when writing, hardening, reviewing, or testing shell scripts, or wiring shellcheck/bats into CI.

Clean, idiomatic Rust 2024 conventions: ownership, error handling, async, unsafe, traits, iterators, performance. Use when writing or reviewing Rust, running clippy, or checking 'idiomatic rust'.

Threat modeling with STRIDE, attack trees, security-requirement extraction, mitigation and control-library selection (defense-in-depth), and compliance mapping (PCI-DSS/HIPAA/GDPR/OWASP-ASVS). Use when doing a security review, choosing mitigating controls, hardening a design, or building a threat-model doc.

Language-agnostic clean-code baseline: naming, functions, error handling, comments, duplication. Use before writing or reviewing code in any language, when no language-specific clean-* skill applies.

Clean, idiomatic Flutter/Dart conventions: feature-first architecture, Riverpod, freezed, go_router, const correctness. Use when writing or reviewing Dart, widgets, `BuildContext`, `build_runner`, or state management.