LLM cost and token optimization

Write less code and say less about it. An agent skill from Green-PT/honey-for-devs.

Green-PT/honey-for-devs

Run the weak-agent adversarial test harness against docx-cli.

kklimuk/docx-cli

Delegate a coding task to Mistral Vibe and supervise the result via git diff.

pcx-wave/vibe-skill

Mines local Claude Code session transcripts with a deterministic Python pipeline to show what the agent is actually used for, how often it fails and what it costs.

amd/gaia

A skill your agent uses for genuinely high-volume Levyra work such as builds, tests, lint, logs, broad searches, dependency output, Git/GitHub or CodeRabbit inspection, CI diagnostics, agent setup…

LUC4N3X/Levyra-deepsound

Surgically select, compress, and recover codebase context using Entroly's MCP tools.

juyterman1000/entroly

Core operating rules for the student knowledge wiki. An agent skill from IssacW228/student-llm-wiki.

IssacW228/student-llm-wiki

Estimates the per-turn token cost of a project's .claude folder and CLAUDE.md, split into always-loaded, path-scoped and invoked-only files, and flags what runs over budget.

poshan0126/dotclaude

Audits a project's Claude Code setup for common token-cost leaks and returns a prioritized fix list, using only static inspection of files and settings.

mergisi/awesome-openclaw-agents

Reviews only the code changed since the last commit, using a code-graph MCP server to find risk-scored changes, test gaps and the blast radius, then writes a short report.

tirth8205/code-review-graph

Add or edit an MCP server entry in mcptoon's config correctly — stdio/streamable-http/sse shapes, command rules, placeholders, and validation.

activeing123/mcptoon

View cost breakdowns, token usage, and call logs from the CLI. Filter by provider, model, or date range. Export usage reports and inspect…

diegosouzapw/OmniRoute

Queries Langfuse traces, prompts, datasets and sessions, and analyzes local LLM gateway logs for requests, context growth, token use and cache hits.

KonghaYao/peri

Personal diagnosis of where your Claude Code + Cowork spend goes.

techwolf-ai/ai-first-toolkit

Benchmarks AMD's GAIA agent against Claude Code and across models on quality, honesty, steps, tokens, time and real cost, using gaia eval tasks.

amd/gaia

Use Entroly for content-blind agent-history audits, explicit baseline/optimized trials, recoverable command or browser evidence, response contracts, and token-efficiency claim verification.

juyterman1000/entroly

Rewrites a memory file such as CLAUDE.md or a todo list in terse caveman-style text to cut input tokens, saving a readable backup outside the project tree.

JuliusBrussee/caveman-code

Analyze prompt-cache effectiveness for Claude Code usage from the Agent Monitor dashboard — cache hit rate (totalcacheread / (totalcacheread + totalinput)), cachewrite vs cacheread reuse, cache-read…

hoangsonww/Claude-Code-Agent-Monitor

Show whether each skill is earning its context-window cost — combined tokens-used view sorted by waste.

khendzel/skills-janitor

Routes a turn or delegated task to the cheapest model and effort lane that will still do it right, using the Jev decision model to classify difficulty and escalate only when needed.

kerpopule/hermes-jev-skills

Split a multi-call LM workflow by cognitive load, not by accuracy: let one strong model make the few reasoning decisions and a cheap model do the many mechanical executions (Aider architect+editor…

agentsope/SkillAlchemy

Always-on context compression for Grok Build. An agent skill from Supercompress/Supercompress.

Supercompress/Supercompress

Connect any AI agent to MCP servers through mcptoon — one config synced everywhere, one stdio gateway, 99.2% fewer tokens on tool discovery.

activeing123/mcptoon

Audit your OpenClaw setup for token waste, context bloat, and cost optimization opportunities

alexgreensh/token-optimizer

Deep-read an arXiv paper and generate structured Deep Note reading notes.

AlphaLab-USTC/ResearchClaw

Minimize token consumption & maximize prompt cache hit rate.

lokikill123/codex-token-skills

Writing and improving agent skills: description as routing signal, progressive disclosure, token budgets, testing before shipping, and improving from repeated mistakes.

damianvtran/local-operator

Describes ClawRouter, a local proxy that forwards each LLM request to the blockrun.ai gateway, which routes to a cheaper capable model, paid by USDC wallet or API key credit.

BlockRunAI/ClawRouter

Speeds up or throttles cognee ingestion: estimate cost with a dry run, tune batching and chunk size, set LLM rate limits and run work in the background.

topoteretes/cognee

Audit and remediate Entroly's MCP marketplace quality with evidence, adversarial validation, and no score gaming.

juyterman1000/entroly

Starts a new Claude Code session from only the last quarter of the current conversation, dropping earlier context to save tokens while keeping recent work.

ykdojo/claude-code-tips

The reference agents' cache-stable request assembly, covering the static system and per-request context split, the fixed tool list, the rolling conversation breakpoint, which config fields are…

anthropics/commerce-agents

Real-time observability for OpenClaw agents — local dashboard + optional encrypted cloud sync.

vivekchand/clawmetry

Shows today's Claude Code spending through the claude-view MCP server, with total cost, running sessions and a per-session breakdown, and other date ranges on request.

tombelieber/claude-view

Diagnose and fix broken MCP servers in mcptoon — doctor, health, per-server probes, and the common failure playbook.

activeing123/mcptoon

Shows a markdown dashboard of basemind activity in the session: tool call counts, top operations and estimated tokens saved against a grep and Read baseline.

Goldziher/basemind

Patches Aider to request Anthropic's one-hour prompt cache lifetime, so long pauses between turns no longer expire the cache or need paid keepalive pings.

OnlyTerp/prompt-cache-skills

Teaches how xc-plugin saves tokens with progressive disclosure, cached responses and consistent configuration, so large device lists and build logs arrive as summaries first.

conorluddy/xclaude-plugin

Trigger when the user asks which model to use, wants to compare model costs, says "what's cheapest for this task", "should I use Opus or Sonnet", "can a smaller model handle this", or…

mergisi/awesome-openclaw-agents

Quick reference for the Breeze RMM AI Agent system architecture, MCP tools, streaming chat, cost tracking, guardrails, and MCP server.

LanternOps/breeze

Ranks a large catalog of installed skills against the current request through the Jev service, and can conclude that no skill applies.

kerpopule/hermes-jev-skills

Chooses which paid frontier model seat should take a task already judged hard, hands it off with proper context, and keeps a watch on the delegated run.

kerpopule/hermes-jev-skills

Analyze and safely optimize OpenSpec main specifications. An agent skill from cdavid817/vanehub-ai.

cdavid817/vanehub-ai

Patch recipe that turns on Aider's --cache-prompts by default for models that support caching, with a verification procedure against the wire request.

OnlyTerp/prompt-cache-skills

Persistent global memory across conversations. An agent skill from lokikill123/codex-token-skills.

lokikill123/codex-token-skills

Lists which models a Weave Router installation may route to and turns individual models or whole providers on or off, mirroring the router dashboard's settings.

weave-os/router

This skill should be used when creating conversation summaries for CodeMemory, discussing summarization quality, optimizing DAG structure, or working with the compaction system.

harrylettering/CodeMemory

Configures OmniRoute's RTK, Caveman and stacked compression modes, manages language packs and rules, and previews token savings of 60 to 90 percent.

diegosouzapw/OmniRoute