tokens plugins

by Douglas Monsky

Fast, exact, local-first Codex usage facts and evidence.

by Xin He

TokenChronicle 词元日志 bilingual local-first Codex activity archive and memory continuity workflows

by CraniusMaximus LLC

Measure what your Claude Code setup costs on every session before you type anything - opening context, what loads, and how much of it never gets used.

by Green-PT

Three reflexive levers that cut agent token cost — less code (YAGNI/stdlib-first), less prose, and denser agent-to-agent handoffs (ESON/compact-JSON). Plus on-demand satellites (review, eco, gain, compress) and a hive of read-only subagents that return compressed handoffs. Auto-intensity lite/full/ultra; correctness and safety-critical paths stay exact.

by Digital Process Tools

Batched file operations for autonomous Claude Code runs. Collapses N reads/greps/globs into one Bash round-trip. Includes session-start prompt injection and a PreToolUse guard that refuses raw shell commands an op replaces, plus a small shipped rule layer for what `replaces` cannot express (hooks/guard-selftest.py lists it). Blocking the harness's own file tools is the operator's setting, not the plugin's (README, Hard-block native tools).

by Patrice Huetz

Keep noisy command output (tests, builds, package managers, diffs, logs) out of the context window. Ships the lm-resizer skill and the lm-resizer MCP server (compress, tool_output, retrieve, stats). Requires the lm-resizer binary on PATH.

by jinhongan

ContextGuard keeps AI coding-agent context lean, safe, and auditable with project-local setup, scans, large-read guards, output trimming/sanitizing, queryable log artifacts, budgeted context packs, tool/MCP schema pruning, statusline visibility, and usage audits.

by hellohelen-ai

Show 1/8, keep 7/8 below the surface. One-line answers, a four-line ceiling, no summary blocks, and -a for depth.

by Maxwell Volz

slashwork offloader. Interception is on by default: /work init authenticates (a bare /work with no token does the same), and from then on a PreToolUse hook routes self-contained subagent spawns to the offload network and returns the artifact in place, falling back to the local spawn on any miss. /work off pauses routing per project, /work on resumes it; a bare /work shows status and the dashboard.

by Christopher Beaulieu

Claude Code efficiency and hygiene toolkit. Surfaces token spend across the 5h / 7d / Sonnet-7d billing windows with per-model / per-agent / per-skill attribution, and audits your effective Claude Code configuration (custom + plugin agents and skills) for overlap and conflicts. Ships an interactive dashboard, a conversational usage-analysis skill, and a structured config-audit skill.

by Daniel Chen

Local-first Codex usage analytics and observability.

by jabworks

Generate an explorable HTML report of Codex session usage — tokens, cache efficiency, subagent costs, and slash-command frequency — from local transcripts.

by David Muñoz Rey

Price a prompt before it costs money, and read the bill after. Deterministic prompt optimisation (no API key, no network), token budgets and drift gates for CI, usage-log profiling that names what would actually move the bill, and seven MCP tools including a spend guard that judges configured USD ceilings before a call is made.

by Diego Zuluaga

Attribute token spend to the task-tracker id in your git branch name.

by abgonzalez93

Keeps whole-file reads and boilerplate writing out of the expensive model's context by delegating them to a cheap one-shot worker, only in the projects you plug in.

by x96x64

Keeps what matters through Claude Code's own compaction — scores your session's context with Jev right before compaction runs, then hands the highest-scoring entries back as a reminder the moment it finishes. No configuration required; set an explicit goal with /ctxjev:set-goal when you want scoring aimed at something specific.

by yukineko

Local LLMOps telemetry for Claude Code: a Stop hook that reads each session's transcript and records token usage, cache hits, tool calls, latency, and estimated cost to a local store — then `gauge report` rolls it up by project, model, and day. `gauge subagents` attributes cost to each individual sub-agent (Task) live from its transcript, so callers like condukt can record true per-task cost instead of a lumped session total. v0.3.9: `gauge subagents --json` now also emits `tokens_input`/`tokens_output` per sub-agent, so callers can record real token usage alongside cost. Observability for your own agent runs, subscription-native (one hook + bundled Rust binary, no API key, nothing leaves the machine).

by intenex

Compacts an idle Claude Code session shortly before its 1-hour prompt cache expires, so coming back later re-sends a short summary instead of the whole conversation uncached.

by dd106-lang

Keep the expensive lead model on planning and review; route routine work to pinned cheaper workers (Haiku, Sonnet, Opus) and audit which models actually ran.

by Andre Paat

SessionStart hook that runs a fast, read-only environment preflight — auth commands, expected tokens (shell env AND known env files), required CLIs, and a host/user/repo/branch identity line — and surfaces failures at the top of the session so the first tool call is never wasted on a dead environment. Never blocks the session.

by Kasun Peiris

Records token usage for every completed task into a private per-project ledger, and shows it as a table with /token-cost.

by kahero

A local optimization hub that audits token waste and routes work to the best available optimizer.

by Zvi Rahamim

Measure what an AI coding session actually costs and get advice derived from your own transcripts — where the tokens went, which were re-read on every turn, and which fixes are worth making on this machine. Tells one person to install a compressor and the next that a compressor would be worthless, because it reads their numbers first.

by rl0ve

Routes substantial agent work to the leanest reliable parent or subagent configuration on task shape, ambiguity, risk, latency, and token use. Carries a route table for Claude Code and one for Codex. Includes capability discovery and scoped deduplication of installed skills.