Claude Code mod
jev-skill-suggestion
Keeps the skill listing out of the context window and lets TypeSafe's Jev, a System One decision model, pick at most one skill per prompt from the skills' descriptions, the way TypeSafe's skill-suggestion cookbook does: one request ranks every skill and asks whether the prompt needs a skill at all, a second re-reads the top three with their SKILL.md and can reject them all. The winner's SKILL.md is attached to the prompt, so the skills can be hidden from the model altogether (/jev-skill-suggestion:setup makes them user-invocable-only and /context counts them at 0). Falls back to the engine's built-in classifier when no key is set.
Install
git clone https://github.com/AnatoliyShk/bulgolingo.git
claude --plugin-dir ./bulgolingo/.claude/skills/jev-skill-suggestionThe repo has no marketplace file, so load the plugin folder directly with --plugin-dir.
What it can access
- Reach
- L3Network: network, drives Claude, reads files, reads env vars, draws
- Sees
- every prompt, compaction, the jev-skill-suggestion:setup skill, every skill
- Hooks
prompt.attachmentprompt.submitsession.endsession.compactskill.prompt
From a static scan of the mod's code by awesome-claude-code-mods. It shows what the code asks for, not a security review.
How Claude Code mods work
- A mod is a plugin with a JavaScript or TypeScript module that runs inside Claude Code (version 2.1.287 or later). It can draw panes beside the transcript, a band above the prompt, buttons and toasts, restyle the spinner and tool rows, and watch or rewrite tool calls and prompts.
- It installs like any plugin: add its marketplace with /plugin marketplace add owner/repo, then /plugin install name@marketplace.
- Check what a mod can reach before installing: it runs with your permissions. Reach goes from drawing only, to reading files, to writing files or running processes, to the network. Run claude plugin validate on a clone to list its hooks and calls without running it.
- Turn one off from the Installed tab in /plugin, or start Claude Code with --safe-mode to skip every customization for a session.
Details
- Repository
- AnatoliyShk/bulgolingo
- Folder
.claude/skills/jev-skill-suggestion- Author
- claude-code-templates
- Last updated
- 30 October 2026
- Listed in
- awesome-claude-code-mods
FAQ
What is jev-skill-suggestion?
Keeps the skill listing out of the context window and lets TypeSafe's Jev, a System One decision model, pick at most one skill per prompt from the skills' descriptions, the way TypeSafe's skill-suggestion cookbook does: one request ranks every skill and asks whether the prompt needs a skill at all, a second re-reads the top three with their SKILL.md and can reject them all. The winner's SKILL.md is attached to the prompt, so the skills can be hidden from the model altogether (/jev-skill-suggestion:setup makes them user-invocable-only and /context counts them at 0). Falls back to the engine's built-in classifier when no key is set. It is a mod for Claude Code, from AnatoliyShk/bulgolingo on GitHub.
What can the jev-skill-suggestion mod access?
Its widest reach is "network" (network, drives Claude, reads files, reads env vars, draws). It sees every prompt, compaction, the jev-skill-suggestion:setup skill, every skill. Like any mod, it runs with your permissions.
How do I install jev-skill-suggestion?
Clone AnatoliyShk/bulgolingo and start Claude Code with --plugin-dir pointing at the mod's folder (.claude/skills/jev-skill-suggestion).
More context & usage mods
jev-compact-trigger
HAR5HA-7663/jev-compact
Companion to jev-compact: watches context use after every turn and starts a compaction at the configured percentage. Separate plugin on purpose — the engine skips a plugin's own session.compact hook when that plugin raises the compaction, so the trigger must come from outside jev-compact for the pruning to run.
k9s
holdennguyen/claude-code-setup
A live, interactive k9s terminal in a pane (k9s runs under a private tmux server, so kube auth and context come from the shell that launched Claude Code), plus two tools so the model can read the k9s screen and send it keys. While k9s runs, Bash calls that invoke kubectl are refused and the model is pointed at those tools; kubectl works as usual otherwise. One server per Claude Code process, a watchdog and a start-up sweep keep orphans from surviving, and an idle stop keeps k9s from polling the cluster unseen.
jev-compact
HAR5HA-7663/jev-compact
Context compaction that prunes instead of summarising: rules drop exploration output and hook noise locally, Jev ranks the rest against a budget, a note says what was removed, and removed outputs are archived for the brain. Falls back to the built-in summary when it cannot do better.
kamehameha
Aurumdev952/claude-code-kamehameha
Context usage as a pixel-art showdown: a hero laughs off a kamehameha that grows with the context window, strains harder and harder, and at 95% it's GAME OVER, please compact
jev
gecm0/jev-mod
Adds $.jev: typed Choice, Score and Noul judgments from TypeSafe's Jev, with a timeout and validated answers, for other mods to call.
kc-usage-band
kcsujeet/kc-claude-kit
A row of pills above the prompt: model, git branch, context fill, 5h and 7d rate limits with reset times, input/output/cache tokens, and session cost, with usage figures turning amber from 70% and red from 90%
Browse the most popular Claude Code mods, all context & usage mods, or the skills and MCP server directories.