personal-skills

Plugin personal-skills from marketplace personal in Allmight97/agents. Commands read from the repo's manifest — 21 days ago.

Skills (23)

• audit-human-era

  Audit a repo's CI, hooks, checks, and workflows for steps built around human speed or gestures, and for checks that do not prove what they claim. Report only. Use when the owner asks for a human-era, process, or validation audit.

• code-review

  Review a PR, branch, diff, or worktree for correctness, requested behavior, regressions, and missing proof. Use for code review or a merge verdict.

• codebase-design

  Vocabulary and principles for module interface design: module, interface, depth, seam, adapter, locality. Use when designing or consolidating a module, placing a dependency seam, or weighing testability.

• diagnose

  Diagnose and fix a reported bug, failing or flaky test, or performance regression by reproducing it and testing the cause. Use for debugging requests.

• field-depth

  Do a task the way a field-leading expert would: pick Speed, Altitude, Depth, or Range, then run the Depth opener before acting.

• grill-me

  Pressure-test a plan, design, or decision by asking the open questions that change it, with a recommendation for each. Use when the user asks to be grilled or to pressure-test a choice.

• handoff

  Write a handoff file so a fresh agent can continue this work; saved to the OS temp directory.

• impeccable

  Design, build, critique, or polish a frontend interface: layout, typography, color, accessibility, interaction, motion, browser-checked iteration. Use for UI changes and UX critique; commands include craft, critique, audit, polish.

• improve-codebase-architecture

  Scan code for architecture friction (shallow modules, leaky seams, untestable interfaces) and recommend bounded refactor candidates.

• issue-hygiene

  Draft, publish, label, refresh, or close a GitHub issue so its body is resume-ready and every write is authorized. Use when an issue is written, edited, labeled, or closed, or when a PR an issue references merges.

• parallels-windows-ops

  Audit, operate, or tune a local Parallels Windows VM: prlctl access ladder, measured resource tuning, Mac/Windows integration hygiene.

• pre-pr-gut-check

  Gut-check a finished branch before its PR opens or goes ready: architecture, whittle, code-review, security, UI, and guidance lenses, then apply the fixes that earn their keep.

• ratchet-files

  Ratchet source file size so every file stays small enough for an agent to read whole: find every file over a token budget, split oversized files by concept, and set up the pre-commit ratchet that stops them from growing. Use when a commit is blocked because a file is over the size budget, when asked

• retro

  Retrospective on a coding session: environment, check, and guidance improvements with session evidence; nothing changes until you pick candidates.

• security-best-practices

  Apply stack-specific security guidance (Python, JavaScript/TypeScript, Go web stacks) for a security review, secure implementation, or vulnerability fix.

• security-threat-model

  Build a threat model of this repository: assets, trust boundaries, attacker capabilities, ranked abuse paths, and mitigations.

• task-compass

  Set the primary outcome and proportionate proof for a mixed or rambling request, and park the rest.

• to-issues

  Split a settled plan or parent GitHub issue into child issues, one per vertical pull request, linked with native sub-issue and blocked-by relationships. Use when the user asks to split work into issues or tickets.

• visual-brief

  Explain substantial material as one HTML brief: diagram, comparison, or expandable evidence, readable without the chat.

• wait-what

  Re-pitch the last message in short plain sentences, with a little context first.

• wayfinder

  Map large or foggy work into a GitHub roadmap issue, or advance an existing roadmap by its number. Use when work spans sessions and scope or design decisions are still open.

• whittle

  Simplify an implementation or audit it for accidental complexity, keeping required behavior. Use when the user invokes Whittle by name.

• writing-for-agents

  Write or audit guidance that agents read: SKILL.md files, AGENTS.md or CLAUDE.md, issues, specs, and handoffs. Use when the task is instruction clarity, rule placement, or skill triggers.

Similar plugins