xrhstosmour

agent-models

Research available models for any provider, rank them per agent role, and update all model references across agent configs and opencode.json. Works with OpenCode, Claude Code, Codex, and any agentic environment.

capture-knowledge

Extracts generalizable, reusable knowledge from the current conversation and adds it as notes to an external "second brain" knowledge-base repository, following that repository's own conventions and scrubbing sensitive/personal/company data. Opens a feature branch, commits, and a PR for review, never pushes or merges without approval. Activate with: "capture knowledge", "update knowledge base", "update notes", "update second brain", "save this to my notes", "remember this for later", each with a target path, e.g. "update second brain in `~/Developer/notes`".

diagnose

Disciplined diagnosis loop for hard bugs and performance regressions. Reproduce → minimise → hypothesise → instrument → fix → regression-test. Use when user says "diagnose this", "debug this", reports a bug, says something is broken or throwing or failing, or describes a performance regression. Not for a standalone log search or a bare Grafana/Loki link with no bug-report framing, see `search-grafana-logs` for that.

eli5

Explain a concept, error, or piece of code in dead-simple terms, by default as if to a 5 year old, or matching a specific audience/expertise level the user names. Use when the user says "ELI5", "explain like I'm 5", "explain like I am/to <someone>", "I don't understand this", "break this down for me", "dumb this down", "explain this simply", asks to explain code or a concept they don't understand, asks for an explanation written as if they already have expertise in the topic, or asks to go deeper or simpler on an explanation just given.

humanize

Rewrite AI-sounding text so it reads as natural human writing, in English or Greek. Detects the source language, strips buzzwords/filler/structural tells specific to that language, and preserves facts, names, numbers, quotes, code, and the requested register. Use when the user says "humanize this", "make this sound less AI", "make this sound more human", "sound less robotic", "remove AI tells", "de-AI this", "rewrite this so it doesn't sound like ChatGPT", "ανθρωποποίησε αυτό το κείμενο", "κάνε το πιο ανθρώπινο", "να μην ακούγεται σαν να το έγραψε AI", or pastes text and asks it to sound less robotic/generated.

Web & Frontend Listed

interface-design

Craft-first interface design for dashboards, admin panels, SaaS apps, tools, settings pages, and data interfaces. Use when designing, building, reviewing, or refining product UI where visual hierarchy, component craft, design tokens, states, or design-system consistency matter.

manage-github-pr

Use for GitHub pull requests, creating a new PR with a full safety-gated workflow, branch checks, commit splitting, structured description, or approving, commenting on, and editing an existing PR. Not for a full multi-agent code review, see `review-github-pr` for that.

manage-phabricator-task

Create and edit Phabricator tasks via the official Phabricator MCP server: new tasks, or updating an existing task's status, title, description, owner, priority, project tags, or subscribers, added indirectly via @-mention, resolved from a username where possible. Triggered when the user explicitly mentions Phabricator or "phab": "create phab task/ticket/issue", "update/edit phab task", "reassign/close/reopen task T<id>", "change priority on T<id>", or "/manage-phabricator-task".

read-phabricator-task

Use when a `Phabricator` link like `https://phabricator.<sub>.<domain>/T<id>` appears, or when reading, searching, or analyzing Phabricator tasks via the official Phabricator MCP server. Not for creating or editing tasks, use the `manage-phabricator-task` skill for that.

read-sentry-issue

Use when a Sentry issue/event link appears, or when working with Sentry error tracking via the official Sentry MCP server, listing organizations/projects/issues, or fetching event/stacktrace details for an issue.

resolve-github-pr-comments

Review GitHub PR review comments, assess validity, propose or make fixes, create fixup commits, push, and reply with SHA links.

review-github-pr

Multi-agent GitHub PR code review orchestrator. Fetches a PR by URL or branch name, spawns specialist sub-agents for architecture and quality review, then synthesizes findings into a structured report. Activate with: "review my PR", "review this PR", "/review-github-pr `<url_or_branch>`".

search-grafana-logs

Use for Grafana/Loki links, "search Loki", "check Grafana", log queries, or investigating an error/failure that likely shows up in logs.

seo

Technical SEO, content quality/E-E-A-T, schema markup, sitemap, image SEO, and AI-search (GEO/AEO) analysis for any URL or page content. Use when the user says "SEO", "SEO audit", "technical SEO", "Core Web Vitals", "schema markup", "structured data", "JSON-LD", "sitemap", "robots.txt", "E-E-A-T", "content quality", "alt text", "image SEO", "AI Overviews", "GEO", "AEO", "llms.txt", "AI citations", or asks to audit/optimize a page or site for search engines.

technical-analysis

Produce a structured technical analysis with method-level change list, notes, and conservative time estimates. Also find architectural deepening opportunities when user asks for architecture review or refactoring candidates.

caveman

Ultra-compressed communication mode. Cuts token usage by dropping filler, articles, and pleasantries while keeping full technical accuracy. Use when user says "caveman", "caveman mode", "talk like caveman", "be brief", "be terse", "keep it short", "less tokens", "save tokens", or invokes /caveman.

craft-design-prompt

Use when the user wants to design something in an AI design tool (v0, bolt.new, Lovable, etc.) or hands over a rough idea, before they've gone to the tool. Polishes the prompt using the project's existing design context.

handoff

Compact the current conversation into a handoff document so another agent session can continue the work. Use when user wants to hand off work, save session state, or prepare context for a fresh agent session.

Web & Frontend Listed

implement-design-from-export

Use when the user hands off a design export, HTML/CSS, or screenshot from an AI design tool and asks to build or match it in the project.

manage-github-issue

Use when creating, commenting on, or editing a GitHub issue, its title, body, labels, or assignees. Not for reading an issue, see the `read-github-issue` skill for that.

read-github-files

Use when reading GitHub commits, releases, or raw repo file/directory content via the API. Also covers creating a release, since releases have no dedicated manage skill of their own.

read-github-issue

Use when reading or viewing a GitHub issue, its body or comments. Not for creating or commenting, see the `manage-github-issue` skill for that.

read-github-pr

Use when reading or viewing a GitHub pull request, its description, diff, or files. Not for creating or editing, see the `manage-github-pr` skill for that.