By yasyf
Run the slop-cop CLI on a file (or stdin) to detect LLM-generated prose tells, and report the violations. Bootstraps the prebuilt binary into the plugin's persistent data dir on session start.
A Claude Code plugin marketplace of small, focused skills. Each plugin installs independently.
| Plugin | What it does | Prerequisites |
|---|---|---|
slop-cop | Check a file or text for LLM-generated prose tells and report the violations. | None; a SessionStart hook bootstraps its prebuilt binary. |
codex | Get a second opinion from OpenAI's Codex CLI on hard debugging or design problems. | The codex CLI on PATH. |
repo-bootstrap | Scaffold a new repo with proven conventions: agent docs, Claude Code settings, guard hooks, plus an opinionated Python packaging layer. | uv (for the hooks and the Python layer); gh recommended. |
llm-prompts | Guidance for writing effective LLM prompts and agent instructions, refreshed with current per-provider model behaviors. | None; slop-cop recommended for the post-edit prose check. |
Add the marketplace, then install the plugins you want:
/plugin marketplace add yasyf/cc-skills
/plugin install slop-cop@skills
/plugin install codex@skills
/plugin install repo-bootstrap@skills
To try it from a local checkout before publishing:
/plugin marketplace add ~/Code/cc-skills
Wraps the slop-cop CLI. Ask the agent to
"check this file for slop" (or name a path) and it runs slop-cop check,
auto-detecting the input language and masking non-prose regions, then reports
the violations grouped by category. It only rewrites the file if you ask. A
SessionStart hook fetches the host-matched binary from the latest
yasyf/slop-cop release into the plugin's persistent data dir, so no Go
toolchain is needed and there is no first-call download stall.
A second-opinion escape hatch for when you're stuck after a couple of failed
attempts. The skill gathers full context, hands it to codex exec, and returns
a structured summary you can verify. It needs the OpenAI Codex CLI installed and
authenticated on your machine.
Scaffolds a new repo with conventions that work out of the box, so you skip the
first day of setup. Every repo gets a base layer: agent docs
(AGENTS.md/CLAUDE.md/STYLEGUIDE.md), a README skeleton, Claude Code settings,
semble code search, and
capt-hook guard hooks. Python projects
also get an opinionated packaging layer: uv with the uv_build backend, a Click
CLI, loguru, pytest, strict pyright, Great
Docs published to GitHub Pages, and
tag-driven PyPI releases via trusted publishing. Say "bootstrap a new repo" or
"scaffold a new Python package".
Packages the team's prompt-writing guidance as a skill: positive framing,
contrastive examples, XML tag structure, reasoning-first output, and the current
per-provider knobs (effort/verbosity/thinking) for Claude, GPT-5.x, and Gemini.
Deeper per-provider notes live under its reference/ folder. repo-bootstrap
installs a companion capt-hook nudge that points prompt edits back at this skill.
MIT. See LICENSE.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
npx claudepluginhub yasyf/cc-skills --plugin slop-copDetect and fix LLM-generated prose tells via the slop-cop CLI. Ships a self-review skill for the agent and bootstraps the binary on first use.
Maintain a Claude-written summaries sidecar in any repo: a stdlib-only template module (summaries.py — whole-file staleness gate, sanitizing, key matching) that consumer render scripts commit and read, plus a config-driven headless refresh skill that rewrites the sidecar whole from real commit/release data with reuse-before-rewrite and a strict flattery law. Consumers describe their groups, raw-material recipes, and render command in .github/summaries.config.json; the gh-profile plugin is the reference consumer.
Create or non-destructively refresh a fancy GitHub profile README (the <username>/<username> repo) from real data — repos, stars, languages, releases, recent activity — with marker-delimited sections kept fresh by a committed cron updater, flattery gates that hide unimpressive numbers, a snake animation, and an opt-in daily Claude refresh that summarizes recent commits and releases into the activity/shipped lines. Requires the gh CLI authenticated with repo + workflow scopes; the AI banner comes from the gen-image plugin.
Generate project images — mascot logos, README banners (dark/light), social cards, arbitrary illustrations — via the OpenAI Images API with local <1MiB compression. Used standalone or by repo-bootstrap and gh-profile.
Get a second opinion from OpenAI's Codex CLI when stuck on difficult debugging, code analysis, or architecture problems, or generate images with Codex's $imagegen skill. Requires the `codex` CLI on PATH.
MCP server that saves 98% of your context window with session continuity. Sandboxed code execution in 11 languages, FTS5 knowledge base with BM25 ranking, and automatic state restore across compactions.
Complete AI coding workflow system. Self-correcting memory + persistent FTS5-indexed research wikis + auto-research loop + multi-LLM council on a single SQLite store. 33 skills, 8 agents, 22 commands, 37 hook scripts across 24 events. Cross-agent via SkillKit.
Open-source, local-first Claude Code plugin for token reduction, context compression, and cost optimization using hybrid RAG retrieval (BM25 + vector search), reranking, AST-aware chunking, and compact context packets.