The context spine that 10x's your Claude Code sessions. Local-first SQLite knowledge graph + MCP server that intercepts file reads and replaces them with structural summaries. Measured 89% token reduction on a real codebase. Apache 2.0, no cloud, no telemetry.
★not rated 1 yesterdayA
tokens not measured
copy · 100%Apache-2.0
Context-shielding, cost-tiered delegation for Claude Code: high-volume work runs in specialized subagents on the cheapest tier a grader can check, returning summaries that keep the main context small. Ships a tiered agent roster, an orchestration skill, three workflows (survey, audit, and a manifest-driven build), and.
★not rated 1 1mo agoA
tokens not measured
originalMIT
Generates structured requirements documents from a project idea or feature description, including functional requirements, non-functional requirements, Gherkin acceptance criteria, edge cases, and out-of-scope items.
★not rated 1 yesterdayA
tokens not measured
originalMIT
Auto-records Claude Code sessions as structured Obsidian notes + Daily Note entries (via the SessionEnd/PreCompact hooks and /done), and ships the obsidian-vault agent — the unified handler for all Obsidian vault work (create, edit, search, organize notes, tasks, canvases, bases). Configure with…
★not rated 1 3mo agoA
tokens not measured
originalMIT
Gentle notification sounds: a rising two-note chime when Claude needs your confirmation, a resolved major arpeggio when a task completes. Sounds are user-replaceable.
★not rated 1 1mo agoA
tokens not measured
originalMIT
Connect Claude to X, LinkedIn, Reddit, and YouTube (browser open, navigate, scroll) via the Socials browser extension. Post, engage, search, and manage social media directly from Claude Code.
Voice + OS notifications when Claude is waiting on you or finished a turn. Cross-platform, focus-aware, with quiet hours and 10s debounce to keep noise low.
★not rated 1 2mo agoA
tokens not measured
originalMIT
An always-on operating protocol for coding agents: build only what's asked, surface ambiguity as Q/A/P/E/X, scale process to task size, and verify before claiming. Injected into every session via a SessionStart hook.
★not rated 1 4mo agoA
tokens not measured
originalMIT
CEO-level strategic review for feature decisions. Challenges assumptions, applies strategy frameworks, and forces rigorous thinking before development begins.
★not rated 1 6mo agoA
tokens not measured
originalMIT
Use QuantCoder from Claude Code to turn quantitative research papers into QuantConnect LEAN algorithm drafts, lint them, and optionally validate/backtest through QuantConnect.
★not rated 1 2mo agoA
tokens not measured
originalApache-2.0
Build production apps on SynapCores AIDB — an AI-native database that unifies SQL, vector search, a Cypher graph, LLM completion, an in-database agentic loop (AGENTRUN), durable agents, native image description, and MCP on one engine. Ships the build-with-synapcores skill, verified against engine v1.14.0-ce.
Scans your Claude Code plugin ecosystem and generates an ARCHITECTURE.md with Mermaid.js dependency graphs, collision reports, and ecosystem health scores.
★not rated 1 6mo agoA
tokens not measured
originalMIT
Auto-indexing for claude-context-mcp. Keeps your codebase index fresh on session start, after git operations, and on session end. Includes /insights skill for promoting memsearch patterns to auto-memory.
★not rated 1 3mo agoA
tokens not measured
originalMIT
End-of-session handoff with bucket-aware dispatch across the canonical 7-bucket docs/ taxonomy (aligned with memory-hygiene v3.3), doc-freshness reverse-lint + skill-freshness audit, and future-to-do emission as GitHub issues. Captures knowledge, routes artifacts to…
★not rated 1 2mo agoA
tokens not measured
originalMIT
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: