portable-claude-skills

Plugin portable-claude-skills from marketplace portable-claude-skills in ariaxhan/the-agent-library. Commands read from the repo's manifest — 1 month ago.

Skills (41)

• build-sites-as-specs

  Architect a website-generation system that structurally cannot produce slop, by compiling a validated specification into output instead of generating HTML directly. Use when building an AI site generator, a templating/CMS pipeline, a multi-tenant site builder, or any system where a model or template

• debug-code-systematically

  Debug code with a scientific method instead of guessing. Use when fixing bugs, regressions, crashes, stack traces, failing tests, unexpected behavior, or "not working" reports. Produces a reproduced failure, tested hypotheses, isolated root cause, fix, and regression check.

• design-api-contracts

  Design or review API contracts for HTTP/REST services and agent-facing endpoints. Use when creating routes, endpoints, status codes, pagination, error responses, versioning, OpenAPI specs, idempotency, or API behavior for AI clients. Produces a clear API contract and review checklist.

• design-loud-failures

  Design systems so a failure can never present as success. Use when building or reviewing pipelines, extractors, reconciliations, checkers, or any computation whose output humans act on; when missing inputs, unreadable files, or crashes could pass silently; when a tool "found" an explanation that fee

• orchestrate-coding-agents

  Coordinate multiple AI coding agents safely. Use when splitting implementation, review, testing, research, or refactor work across agents; when tasks can run in parallel; or when the user asks to orchestrate, spawn agents, delegate, parallelize, or manage tiered work. Produces scoped contracts, isol

• refactor-code-safely

  Refactor code without changing behavior. Use when the user asks to clean up, simplify, rename, restructure, extract, inline, move code, reduce duplication, or modernize internals while preserving behavior. Produces a scoped refactor with before/after checks.

• review-ai-code-quality

  Review AI-generated or recently changed code for common quality failures. Use when reviewing code, validating an implementation, checking "is this ready", auditing AI-authored changes, or looking for edge cases, validation gaps, error handling problems, duplication, and needless complexity. Produces

• secure-code-changes

  Review or implement code changes with security constraints. Use when touching auth, secrets, credentials, permissions, payments, PII, input validation, SQL queries, file uploads, LLM/prompt features, or when the user asks for a security review. Produces a threat-aware checklist, findings, and fixes.

• secure-keys

  Audit and lock down how AI tools hold API keys and credentials. Finds exposed keys across code, config files, .env files, environment variables, shell history, and git history; moves connections to OAuth / "Sign in with X" where one exists; migrates the rest into secure storage (macOS Keychain, a cl

• humanize-ai-prose

  Rewrite AI-generated text into genuinely human prose with seven ordered transform passes: add real named specifics, fix the structure, kill the tells (em-dashes, intensifiers, negative parallelism, meta-signposts), admit real limits, and read it aloud.

• train-a-taste-model

  Train a small model on your own archive that learns your personal taste and ranks new items the way you would, using implicit curation signals you already produced instead of hand-labeling thousands of examples.

• agent-tradition

  Give AI coding agents durable institutional memory: purpose, falsifiable beliefs, narrative lessons, and hook-enforced session accountability, so each session makes the next one smarter instead of starting blind.

• extend-before-inventing

  Search for the existing spine before building anything new. Use when about to create a file, module, helper, wrapper, config system, or process: "should I add a new X", "where should this live", "build a helper for this", "we need a system for this", or when reviewing work that added a parallel vers

• harden-rules-into-hooks

  Enforce the agent rules that must never break with hooks the agent can't bypass (secret scans, destructive-command guards, format gates) instead of prose instructions it quietly ignores under load.

• research-failures-first

  Before building anything non-trivial, research real-world failure modes from production post-mortems and issue trackers and produce a committed symptom→cause→fix→URL map, instead of donating your own debugging time or asking the model to hallucinate what could break.

• critique-vision-handoff

  Adversarially review vision handoff docs, seed docs, architecture briefs, or implementation plans before they reach coding agents. Use when the user asks to "critique this vision", "review my handoff", "stress test this spec", "break this plan", "find the holes", or check whether a planning document

• decompose-by-verification-boundary

  Split substantial work into units by verification boundary, not by file or feature. Use when planning implementation order, breaking down a big task, or sequencing a build: "how should I split this", "what order should I build this in", "break this down", "plan the implementation", "where do I start

• hand-off-before-context-death

  Write the session handoff while the verified context is still loaded, before the window or budget runs out. Use when context is running low mid-work, when pausing substantial work ("wrap up", "hand this off", "save state for the next session", "we'll pick this up later"), or when a session is stuck

• write-project-vision-handoff

  Create implementation-ready project handoff documents from planning context. Use when the user asks to "write a vision handoff", "create a seed document", "prepare this for a coding agent", "turn this plan into VISION.md", or transfer a planned system to an implementation agent. Produces VISION.md a

• brainstorm-with-entropy

  Generate genuinely novel ideas by injecting entropy, ranked by surprise instead of safety. Use when the user asks to brainstorm, generate ideas, find an angle, or name something: "what could I build/make/write", "I need fresh ideas", "help me think of", "I'm stuck on the obvious options". Triggers a

• debate-decisions

  A compact decision protocol for non-trivial choices, reviews, governance calls, and live disagreement. Use implicitly at decision points, even when nobody says "debate", whenever a choice is high-impact, hard to reverse, policy-bearing, or has multiple plausible paths that evidence alone cannot sett

• delegate-to-worker-models

  Delegate implementation to cheap or small models safely by pre-deciding everything. Use when splitting work across agent lanes, picking a model tier, or writing a worker prompt: "can a small model do this", "delegate this", "fan this out", "spawn workers for these", "which model for this task". Enco

• dream-solution-space

  Expand a problem's solution space before committing. Use when the user asks to "dream", "brainstorm deeply", "find a better frame", "generate competing approaches", "stress test options", or when the obvious answer feels too easy. Produces grounded perspectives, adversarial critique, integrity score

• experiment-on-rules

  Treat prompts, rules, workflows, habits, or methodology claims as hypotheses and test them with evidence. Use when the user asks to "run an experiment", "prove this rule", "test this methodology", "validate this workflow", "is this actually working", or decide whether to keep, change, or kill an ope

• extract-invariants

  Turn a borrowed metaphor, analogy, or cross-domain inspiration into a testable mechanism, or prove it is decoration and discard it. Use when a design leans on language from another field ("immune system", "evolution", "market", "quorum", "crystallization"), when someone proposes copying how another

• forge-autonomous-work

  Run a bounded autonomous work loop that generates competing approaches, implements iteratively, runs acceptance checks, stress-tests the result, learns from the output, and either ships or reports why it cannot. Use when the user asks to "forge this", "run autonomously", "take this end to end", "ite

• extract-source-to-notes

  Ingest one external source into structured notes. Use when the user drops a Medium article, blog post, paper, URL, transcript, or pasted text and asks to "extract this", "process this article", "distill this URL", "pull the insights from this", or capture one source with provenance. Produces raw pro

• hunt-research-sources

  Run a wide research source hunt and return a ranked findings report. Use when the user asks to "hunt for sources", "go find sources on X", "scan the web for", "do a research sweep", "find the best sources", or compare many sources before deciding what matters. Produces a sourced report with accepted

• manage-a-knowledge-base

  Run a personal knowledge base or second brain as a system, not a junk drawer. Use for any stage of a notes collection: "process my inbox / triage my captures", "find contradictions across my notes", "synthesize / weave these notes together", "prune / clean up my ideas list", "archive stale notes / d

• audit-website-quality

  Audit a website or page against the full quality policy bank (SEO, accessibility, performance, responsive layout, security, content/anti-slop, link integrity, and structured data) and produce graded findings. Use when reviewing a site before launch, auditing an AI- or template-generated site, checki

• build-and-release-ios-app

  Draft skill for building and releasing iOS or macOS Catalyst apps. Current coverage is the App Store Connect API key TestFlight upload path; expand before publishing into the full lifecycle from app/design import through build, test, metadata, screenshots, TestFlight, and App Review submission.

• prep-site-for-ai

  Make a website legible to LLMs, AI search engines, and autonomous agents (structured data (JSON-LD), llms.txt, semantic server-rendered HTML, machine-readable facts, crawler access, and provenance signals) so AI systems quote you correctly instead of skipping or hallucinating about you. Use when pre

• create-portable-skill

  Create, audit, or package a portable Claude/Codex skill. Use when building a new skill, converting an existing workflow into a reusable skill, reviewing whether something is a real skill or just a rule/reference/kit, or preparing a skills repository catalog.

• audit-the-premise

  Stress-test the assumption baked into a task before spending real effort on it. Use when a request asserts something is broken, true, already decided, or unchanged before you act on it: "fix the broken X", "since Tuesday it's been down", "just flip the existing thing", "the data already has Y". Trig

• reality-audit

  Audit a "done" report or near-final artifact against reality. Use when an AI agent or a person reports work as finished, or when something substantial is about to be sent, published, submitted, or shipped. Separates corroborated structure from self-reported metrics, hunts real defects before release

• test-the-measuring-stick

  Check whether a score, grade, or ranking actually has teeth before you trust it. Use when relying on any quality number: an AI grading its own or another AI's output, a rubric score on writing, a relevance ranking, a benchmark result. Triggers include "is this quality score real", "does my grader/ev

• verify-adversarially

  Earn "done" by splitting verification into two roles: a fresh-context agent attacks the work, and the coordinator adjudicates every finding. Use before declaring a subsystem complete, before trusting a green report, after delegated work returns, or on "verify this", "is it actually done", "attack th

• commission-work

  Create or refine a commission before substantial delegated work. Use when starting a non-trivial coding, research, writing, or planning effort; when the user asks to "scope this", "write a commission", "shape the work", "make the brief", or when an agent needs authority, boundaries, outcomes, and ve

• keep-a-confidence-ledger

  Maintain one honest map of what is actually true about every subsystem in a project. Use when landing or verifying work changes a subsystem's state, at session close, or when someone asks "what's the state of X", "is X done", "how far along are we", "what's actually working". Replaces percentages an

• synthesize-learnings

  Distill accumulated notes and lessons from many sessions into durable insight. Use when scattered learnings, journal entries, meeting takeaways, or decisions have piled up across time and need consolidating: "run a retrospective across my notes", "what have I learned", "clean up and promote my lesso

• write-session-chronicle

  Write an honest session chronicle after substantial work. Use when wrapping up coding, research, writing, planning, or agent work; when the user asks to "write the chronicle", "record what happened", "wrap this up", "leave a handoff", or preserve decisions/failures for the next session. Produces a c

Similar plugins