Browser automation with persistent page state, letting Claude run small incremental scripts to navigate, fill forms, screenshot, and scrape the web.
Rigorously verifies web and terminal UIs with screenshot/pixel-diff evidence plus two parallel read-only reviewers, then issues one GOOD/NEEDS WORK verdict.
Turns complex multi-step work into three persistent Markdown files — task_plan.md, findings.md, progress.md — with automatic context recovery after /clear.
Runs 3 vulnerability hunters and 2 PoC engineers in parallel so only findings with a proven attack path get reported.
A Manus-style skill that keeps complex, multi-step work on track using persistent task_plan.md, findings.md and progress.md files.
Detects an existing OpenClaw install and interactively migrates identity, channel credentials, scheduled tasks, and config into NanoClaw v2.
Enforces strict typing, modern toolchains, and TDD discipline whenever Claude writes Python, Rust, TypeScript, or Go.
A frontend discipline skill that forbids saying "done" until the build passes and the result is screenshot-verified in a real browser.
Lets Claude drive a real browser to research, click, fill forms, screenshot and extract data.
Keeps complex, multi-step work on track by persisting task_plan.md, findings.md and progress.md to disk and re-injecting the plan via hooks.
Explore codebases with tree-sitter AST parsing — index first, fetch details on demand, and cut token usage several-fold.
Splits a project's full claude-mem timeline into ISO weeks and writes one serialized narrative chapter per week, each inheriting the previous week's carry-forward.
Strips AI-generated code smells from your branch diff, but only after regression tests lock the current behavior.
An all-in-one Git skill that splits changes into atomic commits, drives safe rebases, and digs through history.
An expert Git agent that splits work into atomic commits, cleans history with rebase/squash, and digs through history to find when and why code changed.
Playwright-based browser automation with persistent page state, driven by small incremental scripts for navigation, form filling, screenshots and scraping.
A hypothesis-driven debugging loop that demands real runtime evidence, per-runtime references, and a clean-up journal before declaring done.
Splits a task into the smallest mergeable PRs and drives each one end to end: isolated worktree, evidence-backed implementation, PR creation, CI/Cubic verification loop, merge and cleanup.
Splits a task into atomic PRs, implements each in an isolated git worktree, loops until CI and the review bot pass, then auto-merges and cleans up.
Audits a UI against Dieter Rams' ten principles with cited evidence, then issues a NEW / REFINE / REDESIGN verdict plus a ready-to-run `/make-plan` handoff.