Menu
SkillsMP has collected 26 skills from Vistyy/agent-harness. Open a skill to review its source and details.
Showing 26 of 26 collected skills.
Review code or repository changes ruthlessly for bugs, quality regressions, objective fit, proof validity, maintainability, changed state, and final repo health. Use when the user asks for review or before claiming non-trivial work complete. Use this skill…
Handle Python runtime and typing semantics that static checks cannot fully prove: dynamic trust boundaries, exception chaining and cancellation, async/resource lifetime, Decimal/time/config/import pitfalls, dataclass mutability, and framework or SQL/JSON…
Shape non-trivial work before coding or review by preserving the full objective and choosing the simplest coherent end state and slices. Use when objective, scope, owner/interface, architecture, decomposition, cleanup, refactors, risk, or verification…
Route and manage bounded subagent delegation or reviewer handoffs. Use when deciding whether to delegate, preparing handoffs, routing explorer/planning_critic/implementer/quality_guard/final_reviewer/runtime_evidence/design_judge, or handling active-subagent…
Manage lightweight work state: delivery maps, queued items, active control sheets, detail notes, closeout notes, and cleanup. Use when work must survive turns, compaction, subagents, queue routing, or document lifecycle. Use this skill for memory/state only;…
Address actionable GitHub pull request review feedback. Use when the user wants to inspect unresolved review threads, requested changes, or inline review comments on a PR, then implement selected fixes. Use `gh` and the bundled GraphQL script whenever…
Use when a user asks to debug or fix failing GitHub PR checks that run in GitHub Actions. Use `gh` for PR metadata, patch context, Actions check status, and log inspection before implementing any approved fix.
Triage and orient GitHub repository, pull request, and issue work through `gh` and local git context. Use when the user asks for general GitHub help, wants PR or issue summaries, or needs repository context before choosing a more specific GitHub workflow.
Publish local changes to GitHub by confirming scope, committing intentionally, pushing the branch, and opening a draft PR with `git` and `gh`. Use when the user asks to commit, push, publish, or open a PR from the local checkout.
Preserve durable documentation truth by assigning one source of truth, validating claims against repo reality, and safely moving, merging, deleting, or reorganizing docs. Use when creating, editing, auditing, or restructuring durable docs; designing doc IA or…
Govern reusable workflow overlay changes: skills, AGENTS overlays, adapters, role/prompt contracts, validation checks, and reusable agent-policy ownership. Use when changing overlay structure or deciding where agent workflow policy belongs. Use this skill for…
Find root cause before fixing bugs, failing checks, build failures, performance issues, or unexpected runtime behavior. Use when symptoms need reproduction, diagnosis, or boundary tracing before repair. Use this skill for diagnosis only; it does not decide…
Steward persistent tests by choosing the right layer, admission standard, durability, fake-red safeguards, and cleanup. Use when adding, changing, deleting, reviewing, or relying on persistent tests, or when deciding whether a test is the right proof layer.…
Build durable, composable, agent-facing CLIs from APIs, SDKs, curl examples, web apps, admin tools, or local scripts. Use when the user wants a command available from any repo with stable help, auth, JSON, and smoke checks. Use this skill for durable tools…
Classify feedback or review findings before edits by deciding whether they are a surface fix, owner fix, separate backlog work, accepted temporary debt, or no-change. Use when comments, review findings, PR feedback, or user concerns need disposition before…
Implement or debug Flutter/Dart mobile code with repo-consistent architecture, state management, navigation, widget composition, and performance. Use when changing mobile app code or diagnosing Flutter behavior. Use this skill for Flutter mechanics only; it…
Choose appropriate `just` recipes in overlay-managed repos by discovering namespaces, selecting quality tiers, and avoiding raw tools. Use before running project format, lint, test, quality, dev, or runtime commands when `just` is available. Use this skill…
Drive persistent browser or Electron sessions through `js_repl` for fast iterative UI debugging. Use when repeated navigation, handles, app state, screenshots, or live UI inspection should survive across steps. Use this skill for interactive browser control…
Automate a real browser from the terminal through `playwright-cli` or the bundled wrapper script. Use for navigation, form filling, snapshots, screenshots, data extraction, or UI-flow debugging. Use this skill for CLI browser mechanics only; it does not own…
Capture desktop or system screenshots: full screen, app/window, active window, or pixel region. Use when the user explicitly asks for an OS screenshot or when tool-specific capture is unavailable. Use this skill for OS-level capture only; prefer browser,…
Create, edit, or analyze Svelte `.svelte`, `.svelte.ts`, and `.svelte.js` files with official docs lookup, autofixer workflow, and core Svelte posture. Use whenever Svelte source changes or Svelte-specific analysis is needed. Use this skill for Svelte…
Implement or refactor Tailwind CSS v4 styles: CSS-first config, class detection, variants, theming, and maintainable component patterns. Use when Tailwind class structure, config, theme tokens, or styling mechanics are touched. Use this skill for Tailwind…
Judge broad product-facing UI completion against project design context, rendered artifacts, and screenshot-led visual approval. Use when a UI change claims visual/product quality across screens, states, viewports, or clients. Use this skill for…
Run Python commands, dependencies, and tooling through `uv` in uv-managed projects. Use when project `just` recipes do not already own the Python action, and avoid raw python, pip, pytest, or manual virtualenvs. Use this skill for Python execution routing…
Choose honest proof for a work claim and reject proof that could pass while the claim is false. Use before claiming non-trivial work done or when tests, runtime evidence, screenshots, performance, security, CI, static checks, or repo inspection must prove…
Help the user define a concrete, measurable goal before starting work, especially when they ask to use the goal tool, create a goal, set an objective, clarify success criteria, or turn a fuzzy intention into a quantitative outcome. Use this skill for goal…