<?xml version="1.0" encoding="UTF-8" ?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Discovery - Nirav Joshi</title>
    <link>https://discovery.niravjoshi.dev</link>
    <description>Curated tools, hacks, repos, and rabbit holes from the build.</description>
    <language>en-us</language>
    <lastBuildDate>Sat, 25 Apr 2026 09:22:23 GMT</lastBuildDate>
    <atom:link href="https://discovery.niravjoshi.dev/rss.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Qartez MCP - codebase intelligence for AI agents</title>
      <link>https://discovery.niravjoshi.dev/entry/qartez-mcp</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/qartez-mcp</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Rust MCP server that pre-indexes a codebase with PageRank, blast-radius, git co-change, and complexity signals. Agents query the graph instead of grepping files.

Why I saved this: The modification guard - auto-blocks edits to high-PageRank files until the agent runs an impact query - is the most underrated feature in the repo.</description>
      <category>tool</category>
      <category>mcp</category>
      <category>claude-code</category>
      <category>rust</category>
      <category>code-intelligence</category>
      <category>static-analysis</category>
    </item>
    <item>
      <title>Codeburn - token cost TUI for Claude Code &amp; Codex</title>
      <link>https://discovery.niravjoshi.dev/entry/codeburn</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/codeburn</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Interactive terminal dashboard that breaks down where your AI coding tokens actually go. Surfaces the chat-vs-tool-use split most users get wrong.

Why I saved this: One user&apos;s spend breakdown: 56% conversation, only 20% actual coding. You can&apos;t optimize what you can&apos;t see.</description>
      <category>tool</category>
      <category>claude-code</category>
      <category>codex</category>
      <category>cost-tracking</category>
      <category>observability</category>
      <category>cli</category>
    </item>
    <item>
      <title>agents-md - drop-in AGENTS.md template</title>
      <link>https://discovery.niravjoshi.dev/entry/agents-md</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/agents-md</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Curated AGENTS.md preset that kills sycophancy, blocks drive-by refactors, and forces verification loops. Synthesizes Karpathy&apos;s principles with Cherny&apos;s Claude Code workflow.</description>
      <category>hack</category>
      <category>claude-code</category>
      <category>codex</category>
      <category>gemini-cli</category>
      <category>agents-md</category>
      <category>prompt-engineering</category>
    </item>
    <item>
      <title>abtop - htop for AI coding agents</title>
      <link>https://discovery.niravjoshi.dev/entry/abtop</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/abtop</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Rust TUI that monitors Claude Code and Codex sessions: token spend, context window, rate limits, and ports in real time. Like btop but for agent runtimes.</description>
      <category>tool</category>
      <category>claude-code</category>
      <category>codex</category>
      <category>tui</category>
      <category>rust</category>
      <category>observability</category>
    </item>
    <item>
      <title>oh-my-team - 12-agent Claude Code orchestration</title>
      <link>https://discovery.niravjoshi.dev/entry/oh-my-team</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/oh-my-team</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Multi-agent harness for Claude Code with planning, parallel execution waves, and a five-agent review gate. Hub mode lets you drive sessions from Telegram or Slack.

Why I saved this: Telegram remote control with per-project topics is the practical hook - fire off a task from your phone and approve permission prompts mid-execution.</description>
      <category>tool</category>
      <category>claude-code</category>
      <category>multi-agent</category>
      <category>orchestration</category>
      <category>slack-bot</category>
      <category>telegram</category>
    </item>
    <item>
      <title>AgentShield - security scanner for AI agents</title>
      <link>https://discovery.niravjoshi.dev/entry/agentshield</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/agentshield</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>CLI, GitHub Action, and GitHub App that scan agent configs, MCP servers, and tool permissions for vulnerabilities. Detects skill poisoning and prompt-injection vectors.</description>
      <category>tool</category>
      <category>security</category>
      <category>mcp</category>
      <category>claude-code</category>
      <category>agent-security</category>
      <category>supply-chain</category>
    </item>
    <item>
      <title>Open Cowork - desktop app for AI coding agents</title>
      <link>https://discovery.niravjoshi.dev/entry/open-cowork</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/open-cowork</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Open-source Windows/macOS desktop that one-click installs Claude Code, MCP tools, and Skills with sandbox isolation, multi-model support, and Slack/Feishu hooks.</description>
      <category>tool</category>
      <category>claude-code</category>
      <category>desktop-app</category>
      <category>mcp</category>
      <category>sandbox</category>
      <category>computer-use</category>
    </item>
    <item>
      <title>sem - semantic version control CLI</title>
      <link>https://discovery.niravjoshi.dev/entry/sem-semantic-vcs</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/sem-semantic-vcs</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Entity-level diff, blame, graph, and impact analysis for code. 23 languages parsed via tree-sitter so reviews work on functions instead of line ranges.</description>
      <category>tool</category>
      <category>git</category>
      <category>code-review</category>
      <category>semantic-diff</category>
      <category>rust</category>
      <category>tree-sitter</category>
    </item>
    <item>
      <title>AIMock - mock everything your AI app talks to</title>
      <link>https://discovery.niravjoshi.dev/entry/aimock</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/aimock</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>One package, one port, zero deps to fake LLM APIs, MCP servers, A2A, AG-UI, vector DBs, and search backends in tests.</description>
      <category>library</category>
      <category>testing</category>
      <category>mcp</category>
      <category>llm</category>
      <category>mock-server</category>
      <category>ai-testing</category>
    </item>
    <item>
      <title>M-flow - relevance-first agent memory</title>
      <link>https://discovery.niravjoshi.dev/entry/m-flow</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/m-flow</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Memory layer for agents that surfaces relevant context instead of just similar embeddings. Episodic plus long-term memory backed by a knowledge graph.</description>
      <category>library</category>
      <category>agent-memory</category>
      <category>knowledge-graph</category>
      <category>rag</category>
      <category>llm</category>
      <category>long-term-memory</category>
    </item>
    <item>
      <title>Cloudflare Agentic Inbox</title>
      <link>https://discovery.niravjoshi.dev/entry/cloudflare-agentic-inbox</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/cloudflare-agentic-inbox</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Self-hosted email client with an embedded AI agent, running entirely on Cloudflare Workers. No backend to manage, edge-distributed by default.</description>
      <category>tool</category>
      <category>cloudflare</category>
      <category>email</category>
      <category>ai-agent</category>
      <category>edge</category>
      <category>self-hosted</category>
    </item>
    <item>
      <title>Obscura - headless browser for AI agents</title>
      <link>https://discovery.niravjoshi.dev/entry/obscura-headless-browser</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/obscura-headless-browser</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Rust-native headless browser purpose-built for agent-driven scraping and web automation. Faster cold starts and lower memory than Chromium-based stacks.</description>
      <category>tool</category>
      <category>scraping</category>
      <category>headless-browser</category>
      <category>automation</category>
      <category>rust</category>
      <category>ai-agent</category>
    </item>
    <item>
      <title>LLM Anonymization - pentest data scrubber</title>
      <link>https://discovery.niravjoshi.dev/entry/llm-anonymization</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/llm-anonymization</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Reverse proxy for Claude Code that strips IPs, hashes, credentials, and PII before requests hit Anthropic. Dual-layer detection: local Ollama LLM plus regex.</description>
      <category>tool</category>
      <category>security</category>
      <category>pentest</category>
      <category>claude-code</category>
      <category>privacy</category>
      <category>proxy</category>
    </item>
    <item>
      <title>Oxideterm - all-in-one terminal workspace</title>
      <link>https://discovery.niravjoshi.dev/entry/oxideterm</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/oxideterm</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Native Rust + Tauri 2 terminal with local shells, SSH, SFTP, remote IDE, AI agent, MCP, RAG, and a file manager - all in one binary. No OpenSSL dependency.</description>
      <category>tool</category>
      <category>terminal</category>
      <category>ssh</category>
      <category>tauri</category>
      <category>rust</category>
      <category>mcp</category>
    </item>
    <item>
      <title>Libretto - deterministic AI browser automations</title>
      <link>https://discovery.niravjoshi.dev/entry/libretto</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/libretto</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Skill plus CLI that has coding agents generate inspectable browser-automation scripts instead of running prompts at runtime. You debug a real script, not a prompt.</description>
      <category>tool</category>
      <category>browser-automation</category>
      <category>claude-code</category>
      <category>scripting</category>
      <category>testing</category>
      <category>playwright</category>
    </item>
    <item>
      <title>agentic-stack - portable .agent folder</title>
      <link>https://discovery.niravjoshi.dev/entry/agentic-stack</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/agentic-stack</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>One memory + skills + protocols folder that plugs into Claude Code, Cursor, Windsurf, OpenCode, Hermes, or a DIY Python harness - and survives the switch.</description>
      <category>hack</category>
      <category>claude-code</category>
      <category>cursor</category>
      <category>windsurf</category>
      <category>agent-memory</category>
      <category>portability</category>
    </item>
    <item>
      <title>1mcp - unified MCP server aggregator</title>
      <link>https://discovery.niravjoshi.dev/entry/1mcp-agent</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/1mcp-agent</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Aggregates many MCP servers behind one endpoint. Acts as an MCP gateway/proxy so clients only configure a single server.</description>
      <category>tool</category>
      <category>mcp</category>
      <category>gateway</category>
      <category>proxy</category>
      <category>ai-agent</category>
      <category>aggregator</category>
    </item>
    <item>
      <title>Prometheus MCP server</title>
      <link>https://discovery.niravjoshi.dev/entry/prometheus-mcp-server</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/prometheus-mcp-server</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>MCP server that lets agents query and analyze Prometheus metrics through a standardized tool interface. Useful for ops-aware coding agents.</description>
      <category>tool</category>
      <category>mcp</category>
      <category>prometheus</category>
      <category>observability</category>
      <category>devops</category>
      <category>monitoring</category>
    </item>
    <item>
      <title>mcpc - universal CLI for MCP</title>
      <link>https://discovery.niravjoshi.dev/entry/mcpc-cli</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/mcpc-cli</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>MCP client with persistent sessions, stdio + HTTP transports, OAuth 2.1, JSON output for code mode, and a sandbox proxy. Calls any MCP server from a shell.</description>
      <category>tool</category>
      <category>mcp</category>
      <category>cli</category>
      <category>oauth</category>
      <category>ai-agent</category>
      <category>scripting</category>
    </item>
    <item>
      <title>MCPMark - stress-testing MCP benchmark</title>
      <link>https://discovery.niravjoshi.dev/entry/mcpmark</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/mcpmark</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Benchmark harness that evaluates models and agents on real-world MCP usage. Comparable scores across servers and frontier models.</description>
      <category>tool</category>
      <category>mcp</category>
      <category>benchmark</category>
      <category>evaluation</category>
      <category>agent-eval</category>
      <category>tool-use</category>
    </item>
    <item>
      <title>mcp2cli - turn any MCP/OpenAPI/GraphQL server into a CLI</title>
      <link>https://discovery.niravjoshi.dev/entry/mcp2cli</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/mcp2cli</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Runtime adapter that exposes any MCP, OpenAPI, or GraphQL server as a flat CLI. Zero codegen, zero rebuild - handy for shell scripts and agent toolchains.</description>
      <category>tool</category>
      <category>mcp</category>
      <category>openapi</category>
      <category>graphql</category>
      <category>cli</category>
      <category>ai-agent</category>
    </item>
    <item>
      <title>AgentScript - schema-driven agent orchestration language</title>
      <link>https://discovery.niravjoshi.dev/entry/salesforce-agentscript</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/salesforce-agentscript</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Open language from Salesforce for declaring agent orchestration: tools, handoffs, guardrails, and stop conditions. Runtime-agnostic config.</description>
      <category>library</category>
      <category>agent-orchestration</category>
      <category>dsl</category>
      <category>schema</category>
      <category>multi-agent</category>
      <category>typescript</category>
    </item>
    <item>
      <title>lazyagent - terminal observability for coding agents</title>
      <link>https://discovery.niravjoshi.dev/entry/lazyagent</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/lazyagent</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Watch your Claude Code, Codex, and OpenCode sessions from a tiny Go TUI. Shows what each agent is doing, no SaaS dashboard needed.</description>
      <category>tool</category>
      <category>claude-code</category>
      <category>codex</category>
      <category>observability</category>
      <category>tui</category>
      <category>go</category>
    </item>
    <item>
      <title>Agentic Video Editor</title>
      <link>https://discovery.niravjoshi.dev/entry/agentic-video-editor</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/agentic-video-editor</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Turns raw footage plus a creative brief into a polished ad using a Gemini + FFmpeg multi-agent pipeline. Useful reference for chaining vision and editing tools.</description>
      <category>tool</category>
      <category>video-editing</category>
      <category>gemini</category>
      <category>ffmpeg</category>
      <category>multi-agent</category>
      <category>python</category>
    </item>
    <item>
      <title>design-extract - pull a design language from any site</title>
      <link>https://discovery.niravjoshi.dev/entry/design-extract</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/design-extract</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>npx CLI plus Claude Code plugin that extracts colors, typography, spacing, and shadows from a live site into a structured design-token report.</description>
      <category>tool</category>
      <category>design-tokens</category>
      <category>claude-code</category>
      <category>css</category>
      <category>scraping</category>
      <category>design-system</category>
    </item>
    <item>
      <title>claude-code-java - embeddable Claude Code engine</title>
      <link>https://discovery.niravjoshi.dev/entry/claude-code-java</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/claude-code-java</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>AI-agent engine for Java apps. CLI plus REST API that wraps the Claude Code execution model so you can drop it into any JVM service.</description>
      <category>library</category>
      <category>claude-code</category>
      <category>java</category>
      <category>jvm</category>
      <category>rest-api</category>
      <category>ai-agent</category>
    </item>
    <item>
      <title>cc-telegram-bridge - Codex &amp; Claude Code on Telegram</title>
      <link>https://discovery.niravjoshi.dev/entry/cc-telegram-bridge</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/cc-telegram-bridge</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Multi-bot, multi-engine Telegram bridge with per-bot personality, budget caps, streaming, session resume, and an Agent Bus for parallel pipelines.</description>
      <category>tool</category>
      <category>claude-code</category>
      <category>codex</category>
      <category>telegram</category>
      <category>automation</category>
      <category>ai-agent</category>
    </item>
    <item>
      <title>agent-flow - real-time Claude Code orchestration view</title>
      <link>https://discovery.niravjoshi.dev/entry/agent-flow</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/agent-flow</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Watch your Claude Code agents think, branch, and coordinate as they run. Live visualization of subagent trees and tool calls in the browser.</description>
      <category>tool</category>
      <category>claude-code</category>
      <category>visualization</category>
      <category>multi-agent</category>
      <category>observability</category>
      <category>developer-tools</category>
    </item>
    <item>
      <title>awesome-harness-engineering</title>
      <link>https://discovery.niravjoshi.dev/entry/awesome-harness-engineering</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/awesome-harness-engineering</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Curated list for AI agent harness engineering: tools, patterns, evals, memory, MCP, permissions, observability, and orchestration.</description>
      <category>article</category>
      <category>awesome-list</category>
      <category>agent-harness</category>
      <category>mcp</category>
      <category>evals</category>
      <category>agent-memory</category>
    </item>
    <item>
      <title>Future AGI - open-source LLM evals &amp; observability</title>
      <link>https://discovery.niravjoshi.dev/entry/future-agi</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/future-agi</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>End-to-end platform for evaluating, observing, and improving LLM and agent apps. Tracing, evals, simulations, datasets, and prompt management in one project.</description>
      <category>tool</category>
      <category>llm-eval</category>
      <category>observability</category>
      <category>tracing</category>
      <category>agent-eval</category>
      <category>self-hosted</category>
    </item>
    <item>
      <title>CrabTrap - LLM-as-a-judge proxy for agent security</title>
      <link>https://discovery.niravjoshi.dev/entry/crabtrap</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/crabtrap</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Brex&apos;s HTTP proxy that uses an LLM judge to vet agent traffic in production. Drop it in front of any agent runtime to block exfiltration and jailbreaks.</description>
      <category>tool</category>
      <category>security</category>
      <category>llm-judge</category>
      <category>proxy</category>
      <category>agent-security</category>
      <category>go</category>
    </item>
    <item>
      <title>ESP-Claw - AI agent framework for IoT</title>
      <link>https://discovery.niravjoshi.dev/entry/esp-claw</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/esp-claw</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Espressif&apos;s chat-coding agent framework for ESP32 devices. Brings tool-calling LLM agents to embedded targets with C-level memory budgets.</description>
      <category>library</category>
      <category>iot</category>
      <category>esp32</category>
      <category>embedded</category>
      <category>ai-agent</category>
      <category>c</category>
    </item>
    <item>
      <title>Agent Governance Toolkit</title>
      <link>https://discovery.niravjoshi.dev/entry/microsoft-agent-governance-toolkit</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/microsoft-agent-governance-toolkit</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Microsoft&apos;s policy engine for autonomous agents: zero-trust identity, execution sandboxing, and reliability checks. Maps to all 10 OWASP Agentic Top 10 categories.</description>
      <category>tool</category>
      <category>security</category>
      <category>governance</category>
      <category>owasp</category>
      <category>agent-security</category>
      <category>sandbox</category>
    </item>
    <item>
      <title>AgentKit - multi-agent networks in TypeScript</title>
      <link>https://discovery.niravjoshi.dev/entry/inngest-agent-kit</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/inngest-agent-kit</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Inngest&apos;s framework for multi-agent networks with deterministic routing and MCP-native tooling. Picks up where the LangGraph/CrewAI camps trail off.</description>
      <category>library</category>
      <category>multi-agent</category>
      <category>typescript</category>
      <category>mcp</category>
      <category>orchestration</category>
      <category>framework</category>
    </item>
    <item>
      <title>any-llm - one interface for every LLM provider</title>
      <link>https://discovery.niravjoshi.dev/entry/any-llm</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/any-llm</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Mozilla AI&apos;s unified Python client for OpenAI, Anthropic, Google, Mistral, and friends. Swap providers without rewriting a single call site.</description>
      <category>library</category>
      <category>llm</category>
      <category>openai</category>
      <category>anthropic</category>
      <category>python</category>
      <category>abstraction</category>
    </item>
    <item>
      <title>ADK-JS - Google&apos;s agent dev kit for TypeScript</title>
      <link>https://discovery.niravjoshi.dev/entry/google-adk-js</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/google-adk-js</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Code-first toolkit for building, evaluating, and deploying agents on Google&apos;s stack. Tool wiring, traces, and eval harness in one package.</description>
      <category>library</category>
      <category>agent-framework</category>
      <category>google</category>
      <category>typescript</category>
      <category>evals</category>
      <category>agentic-ai</category>
    </item>
    <item>
      <title>Amazon Q Developer CLI</title>
      <link>https://discovery.niravjoshi.dev/entry/amazon-q-developer-cli</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/amazon-q-developer-cli</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Agentic chat for the terminal - natural-language workflows wired into shell, MCP servers, and AWS tooling. Linux/macOS native, Rust binary.</description>
      <category>tool</category>
      <category>amazon-q</category>
      <category>cli</category>
      <category>rust</category>
      <category>mcp</category>
      <category>aws</category>
    </item>
    <item>
      <title>codebase-memory-mcp - fast code intelligence MCP</title>
      <link>https://discovery.niravjoshi.dev/entry/codebase-memory-mcp</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/codebase-memory-mcp</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>MCP server that indexes a codebase into a persistent knowledge graph in milliseconds. 66 languages, sub-ms queries, single static binary.</description>
      <category>tool</category>
      <category>mcp</category>
      <category>code-intelligence</category>
      <category>knowledge-graph</category>
      <category>tree-sitter</category>
      <category>performance</category>
    </item>
    <item>
      <title>Golf - production MCP server framework</title>
      <link>https://discovery.niravjoshi.dev/entry/golf-mcp</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/golf-mcp</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Build, deploy, and scale MCP servers with auth, observability, debugging, telemetry, and a runtime baked in. Skips the per-server boilerplate.</description>
      <category>library</category>
      <category>mcp</category>
      <category>framework</category>
      <category>auth</category>
      <category>observability</category>
      <category>python</category>
    </item>
    <item>
      <title>Destructive Command Guard</title>
      <link>https://discovery.niravjoshi.dev/entry/destructive-command-guard</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/destructive-command-guard</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Rust CLI that blocks dangerous git and shell commands before an agent can run them. Pattern-matched safety net for autonomous coding agents.</description>
      <category>tool</category>
      <category>security</category>
      <category>agent-safety</category>
      <category>git</category>
      <category>rust</category>
      <category>cli</category>
    </item>
    <item>
      <title>Zerobox - process sandboxing on the Codex runtime</title>
      <link>https://discovery.niravjoshi.dev/entry/zerobox-sandbox</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/zerobox-sandbox</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Lightweight, cross-platform process sandbox in Rust. Wraps any command with file, network, and credential controls - built on OpenAI Codex&apos;s runtime primitives.</description>
      <category>tool</category>
      <category>sandbox</category>
      <category>security</category>
      <category>rust</category>
      <category>agent-safety</category>
      <category>cli</category>
    </item>
    <item>
      <title>Superglue - integrations from natural language</title>
      <link>https://discovery.niravjoshi.dev/entry/superglue</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/superglue</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>YC W25 project that builds production-grade tool integrations from a prompt. Aimed at long-tail SaaS connectors that legacy iPaaS doesn&apos;t cover.</description>
      <category>tool</category>
      <category>integrations</category>
      <category>ai-agents</category>
      <category>api-gateway</category>
      <category>typescript</category>
      <category>yc</category>
    </item>
    <item>
      <title>agent-deck - terminal session manager for coding agents</title>
      <link>https://discovery.niravjoshi.dev/entry/agent-deck</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/agent-deck</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>One Bubble Tea TUI to switch between Claude, Gemini, OpenCode, Codex, and Aider sessions. Tabs, tmux feel, no electron.</description>
      <category>tool</category>
      <category>claude-code</category>
      <category>codex</category>
      <category>tui</category>
      <category>go</category>
      <category>session-manager</category>
    </item>
    <item>
      <title>parallel-code - Claude, Codex, Gemini side-by-side</title>
      <link>https://discovery.niravjoshi.dev/entry/parallel-code-worktrees</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/parallel-code-worktrees</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Runs three coding agents in parallel git worktrees so you can compare outputs on the same task without branch swaps.</description>
      <category>hack</category>
      <category>claude-code</category>
      <category>codex</category>
      <category>gemini-cli</category>
      <category>git-worktree</category>
      <category>developer-tools</category>
    </item>
    <item>
      <title>Codesight - universal AI context generator</title>
      <link>https://discovery.niravjoshi.dev/entry/codesight</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/codesight</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Trims and packs codebase context for Claude Code, Cursor, Copilot, and Codex. Saves thousands of tokens per conversation by skipping vendor and generated trees.</description>
      <category>tool</category>
      <category>claude-code</category>
      <category>cursor</category>
      <category>context-engineering</category>
      <category>cli</category>
      <category>codex</category>
    </item>
    <item>
      <title>Paseo - orchestrate coding agents from your phone</title>
      <link>https://discovery.niravjoshi.dev/entry/paseo</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/paseo</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Mobile, desktop, and CLI control plane for Claude Code, Codex, Copilot, and Gemini. Kick off a task in transit and finish at the desk.</description>
      <category>tool</category>
      <category>claude-code</category>
      <category>codex</category>
      <category>mobile</category>
      <category>orchestration</category>
      <category>developer-tools</category>
    </item>
    <item>
      <title>figma-mcp-go - Figma MCP for free users</title>
      <link>https://discovery.niravjoshi.dev/entry/figma-mcp-go</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/figma-mcp-go</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Go-based MCP server that gives agents read/write Figma access without rate limits. Text-to-design and design-to-code in one binary.</description>
      <category>tool</category>
      <category>mcp</category>
      <category>figma</category>
      <category>design</category>
      <category>go</category>
      <category>ai-agent</category>
    </item>
    <item>
      <title>agentfiles - manage agent configs from Obsidian</title>
      <link>https://discovery.niravjoshi.dev/entry/agentfiles</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/agentfiles</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Browse, create, and edit AGENTS.md, .cursor, .codex, and 13+ other agent files from inside Obsidian. Cross-tool config without leaving notes.</description>
      <category>tool</category>
      <category>claude-code</category>
      <category>cursor</category>
      <category>codex</category>
      <category>obsidian</category>
      <category>agents-md</category>
    </item>
    <item>
      <title>SnapAI - icon generation CLI for Expo &amp; React Native</title>
      <link>https://discovery.niravjoshi.dev/entry/snapai</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/snapai</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>AI-powered icon generator wired into the Expo workflow. Generates app icons, splash screens, and adaptive variants from a single prompt.</description>
      <category>tool</category>
      <category>expo</category>
      <category>react-native</category>
      <category>icon-generator</category>
      <category>cli</category>
      <category>openai</category>
    </item>
    <item>
      <title>yt-fts - YouTube full-text search from the CLI</title>
      <link>https://discovery.niravjoshi.dev/entry/yt-fts</link>
      <guid isPermaLink="true">https://discovery.niravjoshi.dev/entry/yt-fts</guid>
      <pubDate>Sat, 25 Apr 2026 00:00:00 GMT</pubDate>
      <description>Pulls transcripts from any YouTube channel into local SQLite + Chroma and exposes full-text and semantic search. Great for building niche knowledge bases.</description>
      <category>tool</category>
      <category>youtube</category>
      <category>search</category>
      <category>rag</category>
      <category>sqlite</category>
      <category>cli</category>
    </item>
  </channel>
</rss>
