Memory for your AI: what you saw on screen, said in calls and worked on, searchable by agents.

Enables agents to store LLM query-response pairs and retrieve them via fuzzy cosine similarity matching, reducing duplicate inference calls; it exposes tools for setting cache entries, semantic lookup, and hit-rate statistics.

Persistent visual cache for LLM-driven software development. Caches screenshots using perceptual hashing, vector search, and AX trees to prevent token overhead and visual hallucination loops.

Enables agents to submit authorized state bundles for free staging dry-run previews that deterministically consolidate memories, plans, tool schemas, and logs over MCP, reporting what stays active, what moves to archive, and why.

DeepSeek Harness plugin: DietrichGebert/ponytail lazy senior mode & 7-rung ladder port, 6 skills with bilingual descriptions & per-skill toggles, zero tools, zero-cache-miss | DeepSeek Harness 插件：DietrichGebert/ponytail 懒人 senior 模式与七阶梯子完美移植，6 个技能描述双语自由切换

DeepSeek Harness plugin: automatically adds a stable per-conversation x-opencode-session header to OpenCode / OpenCode Go provider requests — fixes 400 MissingSessionID and keeps OpenCode's session affinity / prompt-cache routing working.

Restore prior Claude Code AND Codex sessions with zero LLM calls. Cross-tool session handoff, cache-expiry prevention, real cost dashboard. One plugin, both hosts.

Persistent visual cache for LLM-driven software development. Caches screenshots using perceptual hashing, vector search, and AX trees to prevent token overhead and visual hallucination loops.

Markdown and HTML renderer for Svelte 5 — built for rendering streaming AI agent output from Claude Code, ChatGPT, and agentic workflows. XSS-safe defaults, streaming-aware sanitization, token caching, TypeScript types, and Svelte 5 runes.

vMLX - Use MLX models easily - JANGQ (GGUF for MLX) - Not dependant on mlx_vlm

Shared Single-file memory layer for all your agents, sub mili-second RAG over text, photo and video on Apple Silicon.. No Server. No API. One File. Pure Swift

Local MCP server that gives coding agents a verifiable project memory, trims what enters their context, and masks secrets — with honest accounting of what it saves and costs.

Local MCP server that gives coding agents a verifiable project memory, trims what enters their context, and masks secrets — with honest accounting of what it saves and costs.

Static analysis scanner for MCP server code and multi-tenant SaaS applications. 57 deterministic rules covering tenant isolation, tool visibility, cache key scoping, RLS, IDOR, and credential vault isolation. MCP server for AI agent integration.

Build a repo's context graph as a folder of linked markdown files: a local, regenerable cache that every query keeps in sync with your code.

Enables coding agents to audit projects locally for token waste, calculate model arbitrage and pricing alternatives, and make request bodies cache-friendly, with payload optimization sending the supplied request body to the TokenZero service.

Durable agent-created automations for DeepSeek Harness

Provides a shared, repo-local memory that lets coding agents recall relevant decisions, conventions, gotchas, and open threads, save new entries, coordinate file claims, and cache file summaries across tools and sessions.

Enables agents to access locally cached webpage analysis reports sent from a Chrome extension, retrieve the latest or listed reports, and export severity-ordered fix checklists.

Enables AI agents to run free, no-signup Ethereum on-chain lookups over a remote HTTP endpoint: wallet ETH balance, contract/EOA detection, outbound transaction counts and simple risk heuristics, plus NFT metadata (name, description, image URLs, IPFS gateway links and raw tokenURI/uri JSON) for ERC-721 and ERC-1155 tokens. Reads only, with cached results and no key, account, or wallet connection required.

Enables AI coding agents to read GitHub repositories, files, commits, issues, pull requests, releases, Actions, and search results through 18 read-only MCP tools. It reuses existing credentials, routes around unreliable networks, and caches responses on disk to stay within rate limits.

Use GPT, DeepSeek, GLM, Gemini, Kimi and 400+ models in Claude Desktop's own model picker and in Claude Code, while staying signed in to your Claude subscription (1P, no gateway mode). Auto reasoning effort per prompt, subagents routed by difficulty, cache and limit warnings. Claude inside Codex too.

Enables agents to interact with a ServiceNow instance through thirteen consolidated tools covering records, schema, CMDB graph traversal, catalog, attachments, code, operations, knowledge, and administration, with a local store for notes, cache, saved queries, history, and revertible writes.

Enables AI agents to search and invoke 1,939 Atlassian automation tools across Jira, Confluence, Bitbucket, Admin, Assets, and more, with cached reads and JSON-RPC over stdio. It helps agents find tools without calling Atlassian and execute operations while reducing rate-limit pressure.