Syrtis
Track what your AI coding tools cost
syrtis.nyanako.comVisit sponsor ↗Explore coordination and visibility for multi-agent work.
120 projectsTrack what your AI coding tools cost
syrtis.nyanako.comVisit sponsor ↗Turns your Claude Code chat into a chief of staff: keep talking while he hands work to specialists from your installed helpers and runs them in the background, with a live pixel-art office panel.

Gives Claude a register_assumption tool and shows the turn's assumptions, decisions and considered-but-not-done items above the prompt
Add readable status narration to agent activity so you can follow the work.

Track what your AI coding tools cost
syrtis.nyanako.comVisit sponsor ↗X post promoting Microsoft's open-source dotnet/skills repository — 14 plugins and 99 skills that work in Claude Code, Codex, Copilot and Cursor — with a short walkthrough of five recommended plugins and a link to the author's blog guide on writing your own Claude skills.

Picks the model for each subagent that names none, with OpenAI's Decisions API or Jev, from its short label only. One line per spawn above the prompt; /route shows each pick and what each subagent cost.

BazaarLink is an OpenAI-compatible AI API gateway: one key for 140+ AI models, pay as you go, with budget limits.
bazaarlink.aiVisit sponsor ↗Project Sentinel is an open-source Claude Code mod for long autonomous runs that separately tracks work progress, context usage with automatic handoffs, and prompt-cache health/expiry. It adds autopilot fresh-context handoffs, cache protection, permissions/guardrails, resource limits, activity, profiles, and an optional companion, and can be installed via the plugin marketplace.

Graph Intelligence Engine for RuFlo: single-entry personalized PageRank, streaming delta updates and witness-signed reasoning artifacts (ADR-123), built on sublinear-time-solver. Library: `npm test`, `npm run build`.
A Reddit post introducing Apiarius, an open-source (MIT) Claude Code mod that adds a live "agent deck" showing every agent and subagent, their current activity, parent agent, inter-agent messages and changed files. It works via Claude Code hooks without altering Claude's behavior, and is available in the terminal, a local web page and the desktop app, with vector avatars for agents.
A Claude Code quantitative trading plugin bundle with 18 built-in agents and 82 skills, covering strategy design, backtesting, risk management, portfolio optimization, and cryptocurrency analysis, with a GitHub repository link.
universal-modder is an MIT-licensed Claude Code plugin that can identify game engines, read real code, build mods, generate assets with fal, test in-game, and produce showcase videos.
The author shares Hive, a self-made Claude module that lets you use a panel to inspect the progress of all agents after calling implement-spec; it can also be used without any development workflow.

A Reddit post describing a Claude Code mod that uses a runtime judge model to automatically select between Haiku, Sonnet and Opus and estimate reasoning effort per prompt, plus warnings for cold prompt caches and overloaded chats, and a smart handoff feature. The mod claims to read recent messages before switching to avoid mid-task model flips. It is described as free and open source (MIT) with a website for installation, but no repository link or verifiable source is included.

Top 3 by votes, then GitHub stars
No. Votes and stars measure popularity. Review the source and permission scan before installing.