Plugin Claude Code
Tooling that keeps your repo out of the agent's context window. Search and edit code through a language server (or tree-sitter, or a fuzzy search built from the repo's own words), section-edit Markdown and config by heading, and read huge editor/build logs without pasting them into the chat. Answers come back as a…
Plugin Claude Code
Sits between Claude Code and your codebase and hands back a short file:line list instead of dumping source into the chat. It uses a real language server when it can (clangd / Roslyn / tsserver / pyright), tree-sitter when there's no toolchain, and a fuzzy search built from your repo's own naming when you don't know…
MCP server Claude CodeCodexCursor +2
MCP server "vs-search" as configured in JSungMin/vs-token-safer. Runs ${CLAUDE_PLUGIN_ROOT}/server/index.js with node.
Instructions file
Instructions for JSungMin/vs-token-safer, covering vs-token-safer — claude rules, first, orient (every session), what's true, layout and conventions (inherited — non-negotiable).
Agent
Delegated, token-isolated code LOCATOR for C/C++ (clangd), C#/.NET (Roslyn), JS/TS (tsserver), and Python (pyright) projects — no IDE needed. It LOCATES (symbols / references / definitions / files) via the official language-server index and returns ONLY a compact file:line table; it never reads or returns bodies.…
Command
Preview-only dead-code analysis — from a seed symbol, walk the call graph to a fixpoint and list what is provably DEAD (and what cascades), without deleting anything.
Command
Report how many tokens the vs-token-safer plugin has saved (vs forwarding the language server's raw index responses).
Command
Configure the vs-token-safer plugin (project path, backend, result cap). Writes /.vs-token-safer/config.json and tells you to /reload-plugins.
Command
Refresh the committable tree-sitter symbol index (.vts-index) when it has gone STALE — one incremental re-index so the syntactic tier answers CURRENT code, not the state it was built at. Use when searchsymbol shows "SYNTACTIC · STALE", after a big branch switch / pull / large refactor, or when the file:line answers…
Command
Stop the vs-token-safer dashboard server started by /vs-token-safer:viz (frees the port; nothing was ever transmitted).
Command
Open the vs-token-safer dashboard — a local (127.0.0.1, nothing transmitted) page showing tokens saved, language mix, per-tool savings, and the include graph as an interactive 3D force graph (Three.js, vendored locally — no CDN).
Plugin Claude Code
Token-efficient game-engine & build log analysis for Claude Code (CLI-first). Parses, deduplicates, and classifies huge Unreal/Unity/Godot/MSVC-UBT-MSBuild logs — search by severity/category, roll up by callsite, diff runs, locate file:line, and extract decisive scalar fields. No IDE required.
Agent
Delegated, token-isolated game-engine/build/structured log analysis. Hand it a log path (Unreal Saved/Logs, Unity Editor.log, Godot output, MSVC/UBT/MSBuild build output, or any .log/.jsonl) plus a question — "what errors", "what's flooding it", "what changed since last run", "track this scalar over time", "which…
Command
Scan local Claude Code transcripts and report (aggregate, local-only) how many raw log reads bypassed gamedev-log vs went through it — to find missed token savings. No log content in the output.
Command
Show or set Bash log-grep enforcement (block / warn / off) — controls whether raw grep/tail/cat over .log/.jsonl/Logs is intercepted and steered to gamedev-log.
Command
Analyze the game-engine/build log (Unreal Saved/Logs, Unity Editor.log, Godot output, MSVC/UBT/MSBuild, or any structured log) — parse, deduplicate, classify by severity/category, and read it token-efficiently via the gamedev-log CLI instead of dumping the raw file.
Hook
Runs before the agent uses a tool for Bash and Read tool calls, executing block-log-grep.mjs via node. From JSungMin/vs-token-safer.
Skill Claude CodeCodex
Analyze an editor, build, or structured trace log (Unreal Engine Saved/Logs, Unity Editor.log, Godot output, MSVC/UBT/MSBuild build output, JSONL / structured .jsonl traces, or any text log) token-efficiently. Use when the user mentions checking, reading, searching, summarizing, or diffing a log; investigating…
Skill Claude CodeCodex
Preview-only dead-code analysis for C/C++, C#/.NET, JS/TS, and Python via vs-token-safer. From one or more seed symbols, walk the official language server's call graph to a fixpoint and report what is provably DEAD (and what transitively becomes dead) — without deleting anything. Use whenever the user asks what code…
Skill Claude CodeCodex
Routing rules for code search in C/C++, C#/.NET, JS/TS, and Python projects via vs-token-safer — use the vs-search MCP tools (clangd/Roslyn/tsserver/pyright language-server index) or the vts CLI instead of Bash grep. Use whenever searching for a symbol, definition, function, variable, type, or finding…
Skill Claude CodeCodex
How to open/close and reason about the vs-token-safer local dashboard — an interactive page (savings trend, language mix, per-tool savings, and a 3D Three.js include-graph) served on 127.0.0.1. Use when the user wants to SEE/visualize what vs-token-safer has indexed or how many tokens it saved, or asks to…
Plugin Claude Code
Token-efficient tooling for large Unreal C++/Unity/.NET projects: search the codebase through JetBrains Rider's index instead of grep, and analyze huge editor logs — both at 99% fewer tokens.
Plugin Claude Code
Force Claude Code to use Rider's MCP symbol/reference/file index instead of Bash grep, with token-flood summarization. Faster and fewer tokens on large Unreal C++/.NET codebases. Auto-installs gamedev-log-analyzer.
MCP server Claude CodeCodexCursor +2
MCP server "rider-search" as configured in JSungMin/rider-mcp-enforcer. Runs ${CLAUDE_PLUGIN_ROOT}/proxy/src/server.js with node.