anntnzrb/agents

skill-creator

Use when creating, editing, validating, benchmarking, or packaging agent skills, SKILL.md files, or skill triggers.

autoreview

Use when reviewing git diffs, commits, or pull requests for bugs, invariant violations, regressions, or security flaws.

amz-live

Use when searching Amazon products, comparing live prices, filtering ratings, finding deals, or shortlisting items.

artificial-analysis-live

Use when querying Artificial Analysis or comparing AI model speed, latency, throughput, or provider token pricing.

deepswe-live

Use when querying DeepSWE benchmark scores, leaderboards, pass@1 rates, model efficiency, or confidence intervals.

ship

Use when asked to ship, get an issue or PR merge-ready or merged, or file a discovered problem for later.

github

Use when managing GitHub via gh CLI, issues, PRs, Actions, releases, or stacked PRs; not for contribution guidelines.

autommit

Use when asked to autommit, split changes into atomic commits, or create unattended git commits from local diffs.

tdd

Use when fixing a bug via TDD or reproducing a defect with a cheap failing regression test before writing the fix.

agent-browser

Use when automating a live browser or agent-browser for login, forms, clicks, screenshots, or UI tests, not for crawls.

flight-live

Use when searching live flights, comparing airfare across dates, filtering nonstop routes, or ranking travel options.

git-worktrees

Use when creating, removing, or managing Git worktrees for isolated branch development, parallel tasks, or testing.

grep-app

Use when searching public GitHub code with Grep.app for real-world API, regex, or config examples, not for local code.

livebench-live

Use when querying LiveBench releases, categories, subtasks, model scores, leaderboards, or cost per successful task.

mneme

Use when denoising meeting transcripts, extracting action items, or searching audio notes, not for coding sessions.

n8n

Use when creating, inspecting, activating, exporting, or managing n8n workflows via the REST CLI or n8n MCP tools.

nix-clan-2605

Use when configuring Clan 26.05 inventory, services, vars, or deployments, not for generic Nix flakes or derivations.

recall

Use when asked to recall my work, catch up, or reconstruct context from agent sessions and git, not for code history.

reddit

Use when browsing or searching Reddit, r/ subreddits, thread comments, or u/ user activity, not for general web search.

ui-ux-pro-max

Use when designing or reviewing web or mobile UI/UX, accessibility, responsive layouts, styling, or components.

vals-live

Use when querying Vals benchmarks, vals.ai, Vals Index, SWE-bench, Terminal-Bench, model scores, or cost per test.

webpaste

Use when sharing or fetching text, screenshots, recordings, or archives via anonymous URLs; not for clipboard copying.

odoo-ops

Use when developing, testing, or operating Odoo 17 modules, replica DBs, XML views, Server Actions, or safe RPC.

python

Use when writing Python, editing .py files, configuring pyproject.toml, using uv, basedpyright typing, or pytest tests.

react-best-practices

Use when building or optimizing React or Next.js rendering, data fetching, bundles, or Server Actions, not styling.

pstack-principles

Use before refactoring, debugging, design, or reporting a measured speedup, applying pstack engineering principles.

blast-radius

Use when asking what a change could break elsewhere, or vetting a risky diff before it ships, beyond the callers.

correct

Use when the user keeps correcting agents for the same repo mistakes and wants each made structurally impossible.

firecrawl

Use when scraping URLs, crawling or mapping sites, parsing docs, or querying Firecrawl indexes, not for deep research.

how

Use when explaining how a subsystem works, tracing runtime code flow, or critiquing design, not for git history.

swarm

Use when fanning out parallel subagents across task slices, races, or best-of attempts, not for live web search.

technical-writing

Use when writing developer docs, RFCs, architecture guides, PR descriptions, or commit messages; not changelogs.

unslop

Use when asked to deslop text, remove AI writing patterns, or clean AI-generated code, not for general refactoring.

verification-skill

Use when an app lacks repeatable behavioral verification or its project-local verification skill needs an audit.

why

Use when investigating why code was written a certain way via git blame, PRs, or issues, not for how it works today.

paseo

Use when delegating to Paseo subagents, handing off tasks, getting second opinions, or setting Paseo heartbeats.

kestra

Use when interacting with Kestra instances through kestractl or querying Kestra documentation through MCP.

market-hunter

Use when finding or vetting software subscriptions, keys, or accounts on G2A, Kinguin, Plati, Z2U, or FunPay.

ecuabet

Use when given an Ecuabet match URL or ID for live odds, EV, incidents, or feed-health analysis.

rust-script

Use when creating or debugging a single-file Rust script with Cargo, rust-script, or an embedded manifest.