LLM Mart Basic

Orchestrate end-to-end DevOps/SysOps work - plan the ops chain, delegate to specialized agents (Kubernetes, cloud, GitOps, SRE, sysops), enforce approvals, and verify outcomes.

Design multi-touch outreach sequences - email, LinkedIn, phone - with timing, angles, and stop conditions. Use to systematize prospect follow-up.

Create, audit, install, and manage Navin plugin packs - self-contained bundles of skills and MCP servers (npx/uvx/docker). Use when the user wants to package capabilities, install a pack from git or a folder, or extend Navin with external MCP tooling.

Plan and operate Google Ads, Microsoft Ads, Meta Ads, TikTok Ads, Reddit Ads, and LinkedIn Ads - real analysis of exports / MCP rows with the `ads` engine, targeting, creatives, budgets, and approval-gated optimization loops. Use when paid acquisition is on the table.

Generate PDF documents - reports, one-pagers, certificates, agreements, product sheets, quotes - from HTML laid out as A4 pages (Chromium) or from a finished DOCX/PPTX/XLSX (LibreOffice). Use when the deliverable must be a polished PDF.

Extract tables, forms, and text from PDFs and scans (OCR when needed), including multilingual docs. Use for contracts, invoices, and document intake.

Think like an attacker - chain weaknesses into realistic exploit paths, build proof-of-concept reproductions, and prioritize by real-world impact. Use for /redteam, attack simulation, or "how would someone break in?" analysis. Ethical, authorized, in-scope only.

Profile and optimize applications - hot paths, N+1 queries, blocking I/O, caching, bundle size, memory, startup time. Use for /turbo, "why is it slow?", or pre-launch performance passes.

Enforce least privilege for commands, file paths, domains, and APIs. Use before destructive exec, broad writes, or when the user asks to lock down what the agent may do.

Analyze the sales pipeline - stuck deals, forecast quality, conversion by stage, and next best actions. Use for weekly pipeline reviews and forecasting.

Browser automation - navigate, click, fill forms, multi-tab, upload, extract, infinite scroll, download, screenshot, and deliver web tasks end-to-end. Prefer the built-in `browser` tool (Playwright + optional browser-use). Use with scrape-operator for project delivery.

Create PowerPoint (.pptx) decks - layouts, text, tables, charts, images, and brand templates - using python-pptx. Use whenever the deliverable is a presentation file.

Design the story, structure, and slide-by-slide plan of presentations - pitch decks, client presentations, reports. Use before generating any deck.

Build pricing scenarios - options, margins, discount policies, and negotiation floors - for offers and deals. Use before pricing any proposal or negotiation.

Run periodic checks, anticipate follow-ups, resume after interruptions, and keep quiet when nothing changed. Use with cron, .navin/HEARTBEAT.md, and sustained goals (/goal) for autonomous monitoring.

Generate product imagery - packshots, lifestyle scenes, e-commerce sets, mockups, and short product videos from reference photos. Use for product pages, catalogs, marketplaces, and launches.

Write business letters, reports, memos, meeting notes, and formal documents in French, English, or Arabic. Use for any professional document that must be polished and correctly toned.

Generate hundreds of structured pages from data (locations, integrations, glossaries, comparisons) with unique value per page. Use when a dataset can answer many long-tail queries.

Work the shared project task board (kanban + dependency-aware plan + milestones + timeline) that humans edit in the Dev workbench - decompose objectives, claim tasks, track progress, file findings, assign to subagents or humans, and run the board in a loop. Use for /board, task t

Create and maintain the .navin/metadata project knowledge base - file roles, dependencies, and the metagraph index - so questions map instantly to the right files without grepping. Use at project start and whenever files are added, moved, or repurposed.

Fourteen posts of being wrong in production, compressed to checkboxes

Healthy nodes, a quiet network, 300 restarts in three days, and a latency budget measured in milliseconds

Discovery worked. Ping worked. Every TCP connection timed out, and later the tunnel only worked when someone had a terminal open.

Every VM came back. The cluster did not. Declarative systems converge on config, and the datapath isn't config.

A surprising share of AI-in-the-terminal failures aren't the AI. They're zsh, and a version of bash from 2006.

A Claude Code plugin turns standalone project configuration into a namespaced, installable extension that teams and communities can update as one unit.

None of the safety came from the model. It came from six boring habits.

Skills package instructions and references. Subagents run work in a separate context and return results. They solve different problems and can be composed deliberately.

Six hours in, one step left, everything green, and the incident that didn't happen

CLAUDE.md carries persistent project context. Skills load reusable procedures when relevant. Separating stable facts from task-specific workflows keeps both easier to maintain.

Twenty minutes recovering secrets that never existed, and the one sentence from a human that ended it

An API request routing a model's tool call through an approval gate to a remote MCP server

31 config keys, two audits, and why the first one was wrong in both directions

The official MCP Registry stores standardized server metadata rather than package code. Publishers verify a namespace, describe installation or remote access, and submit immutable versions.

Everyone looks at the Dockerfile. The file that actually leaked the key was the project file.

Remote MCP authorization uses established OAuth standards, but secure integration still requires issuer validation, least-privilege scopes, protected token handling, and server-side enforcement.

"Copy it over and switch the reference" is two steps, and the outage lives in the one nobody checks

stdio fits local processes and prototypes. Streamable HTTP fits hosted services and shared integrations. The right choice follows where the capability runs and who must reach it.

The most important rule wasn't about what I could change. It was about what I was allowed to display.

Tools perform operations, resources expose readable context, and prompts provide reusable templates. Choosing the correct primitive makes an MCP server easier to understand and govern.

Close a finished session: sweep for unfinished work, ask once, land, file the follow-ups, hand off, tell the sessions that depend on this one, then archive.

Write the repository handoff file for the next session, and record any durable learning.

Merge an approved pull request, clean up its worktree and branch, then check whether a release is due.

Turn a topic or issue into a plan the reviewer approves in the native plan pane.

Answer a research question with parallel read-only gatherers and one synthesized digest.

Review the branch's diff in two fresh contexts — scope against the spec, then quality — and report findings only.

Audit infra security: secrets, deps, CI/CD, webhooks, AI/skill files

Audit Solana program code for exploitable bugs and write a findings report

Compare per-instruction CU with the stored baseline to catch regressions

Build the web client (Next.js, Vite, React) and check env, types and bundle

Build Solana programs (Anchor, Pinocchio, native), incl. verifiable builds

Build the Unity project in batchmode for WebGL, desktop, Android or PSG1

Turn a solana-ai-kit fork into a project: set up CLAUDE.md, remove kit files

Un-ignore and commit the kit config dir, instruction file, .mcp.json and .gitmodules

Replay a user's failing transaction on forked state and map the error to source

Deploy a program to devnet, or to mainnet after the user's explicit go-ahead

Review the branch diff for Solana security issues, CU waste and AI slop

Read-only check of toolchain and kit config, with one fix-it command per failure

Consolidate MEMORY.md and Project Learnings: dedupe, resolve conflicts, prune

Explain Solana code with a diagram and a step-by-step walkthrough

Your efficient agentic AI coding CLI assistant

AI coding with Aegis governance built into execution: baseline-aware changes, evidence-backed delivery. Free desktop client, your choice of model. 将哲科思维融入 AI 开发…

Android in docker solution with noVNC supported, video recording, mcp server and AI-agent

Your AI assistant, built for the agentic era. Open source and local: talk to it, and it runs your agents, your coding CLIs, your browser and your apps. Windows,…

Verifiable program-level self-evolution for AI-for-Science agents: Skills and typed Operators grow from replay-verified executions.

Let Claude Code and other coding agents drive and debug Chrome from the shell. DOM, network, console and raw CDP as short commands, no screenshots and no MCP ne…

Unofficial, agent-friendly Mercadona shopping CLI (Go) — search products, read prices, build a cart and prepare checkout from the terminal. BYO credentials.

Local-first AI canvas for visual workflows, AI short drama, image/video generation, storyboarding and asset management. ComfyUI, AI agents, MCP & Blender. 本地优先…

Let any AI coding tool — Claude Code, Cursor, Codex — drive your real Chrome. One-prompt setup, muscle memory, local-first.

runs anywhere. uses anything

Open-source dots for the web: an AI agent with its own browser, one that does not get blocked.

Free MIT AI coding agent — no API key needed. Built-in free model, or bring Claude, GPT, Gemini, DeepSeek, Ollama +50 more. Swarm mode, 40+ tools, browser & des…

Open-source AI customer support system. AI-first support, human-ready operations.

A durable, multiplayer, mobile-first web app for Pi agents, built on Pi Durable

Embedded Cypher knowledge graph for Python and Rust. Bundled MCP server, describe() schema, and code-graph parser for LLM agents.

Turn PDFs, books and papers into interactive learning webpages｜将复杂材料转化为可追溯、可测验、可做笔记的学习网页

PhysicsOS 是一个面向初高中物理学习的公益可视化智能体，通过 AI 理解题目并结合物理引擎，将抽象物理过程转化为可交互、可观察、可计算、可验证的真实物理场景。