large-file-reader 0.3.1
Big files and connector results are read by Haiku first; Claude gets short notes and a path to the full thing. A side pane shows each hand-off to Haiku live. /input-router full lets the next one through whole
Install
/plugin marketplace add rome2o/tokenpower-claude /plugin install large-file-reader@tokenpower-claude
Run both lines inside Claude Code.
Hooks into
- Pane
- Tool calls
- Toasts
- Slash commands
Also in rome2o/tokenpower-claude
- mcp-schema-auditor 0.1.0 /mcp-cost shows how many tokens each MCP server's tool definitions cost in every request, ranked, and suggests which to disable with /mcp 1
- context-bloat-auditor 0.1.0 /context-audit lists the always-loaded items that cost the most tokens (CLAUDE.md files, memory, skills, agents), flags big items and duplicate names, and never reads the contents of your files 1
- context-meter 0.1.0 A live band above the prompt: what fills the context window by category, how close it is to compaction, this session's cost as Claude reports it, and today's tokens saved by the other mods 1
- compaction-pruner 0.1.0 Off by default. When you compact, shortens old large tool results with simple local rules and keeps the original text (secrets masked) in a folder you can open. It deletes nothing and makes no network or model calls 1
- effort-level-switcher 0.1.0 Reads each prompt, picks the effort level it needs with a four-question framework, and offers Switch to that level, or sets it by itself in auto mode 1
- subagent-model-router 0.6.0 Modified copy of an MIT-licensed Jev model router that uses Haiku in place of Jev. Haiku reads each subagent task at agent.spawn and answers in JSON with a tier and a confidence from 0 to 1; a task moves to a cheaper model only when that confidence is at least minDowngradeConfidence. Logs every p... 1
- prompt-cache-timer 0.3.0 Counts down until the chat's prompt cache goes cold, above the prompt. When a big chat goes cold, offers a handoff to a fresh chat. Keep warm pings the cache in the background so it does not go cold 1
- context-handoff 0.1.0 At 250k tokens, offers Refresh: writes a handoff of the whole chat, shows the steps with progress bars, then copies it or starts a new chat with it. Logs every run to ~/.claude/handoffs/handoff.log and saves a draft first 1
- web-research-saver 0.3.0 Keeps web research out of the main chat: Haiku shortens Claude's own big search results, and /research-router has a cheap worker or subagents read real pages and give Claude a short brief with sources 1
- browser-screenshot-compressor 0.2.0 Cheaper browser tasks: shrinks every screenshot Claude takes while it drives a browser, or has Haiku describe it in words. /efficient-browser full lets one full-size screenshot through 1
More panes mods
- replay-theater 0.1.0 Step through the last turn's edits, one diff at a time. 279.9k
- diff 0.1.0 The diff pane as a plugin: /diff opens the session's uncommitted changes beside the transcript, file by file with their hunks, refreshed as Claude edits and runs commands; the first edit opens it where the terminal is wide enough, and a file's ask button rides its hunks on the next prompt. 155.5k
- ruview-live 0.1.0 RuView live sensing in Claude Code: /ruview opens a pane beside the transcript with three views: an overview of your CSI nodes (ESP32, Realtek) and 60 GHz radar (rates, loss, RSSI, presence, device-reported vitals, alerts), a live CSI amplitude waterfall, and a radar range fan with vitals charts,... 104.2k
- ruflo-swarm 0.3.8 mod + skill Agent teams, swarm coordination, Monitor streams, and worktree isolation — wraps 4 swarm_* + 8 agent_* MCP tools (12 total) plus 6 topologies (hierarchical / mesh / hierarchical-mesh / ring / star / adaptive). With Claude Code function hooks on (early access), a live swarm pane: agent tiles lit b... 81.6k
- headroom-snip 0.44.0 Watch Headroom snip tokens off every Claude Code request: an animated band, a /headroom log and a running total. 79.5k