{"slug": "i-m-building-the-trust-layer-between-humans-and-ai-agents", "title": "I'm building the trust layer between humans and AI agents", "summary": "A developer built a local dashboard that reveals token usage and costs for Claude Code, addressing the lack of visibility into AI agent actions. The tool reads JSONL data from `~/.claude/projects/` without sending anything off the user's machine. The project is part of a broader effort to create trust infrastructure—including audit trails and permission layers—for autonomous AI agents.", "body_md": "I'd been using Claude Code every day for months.\n\nThen one day I asked myself a simple question:\n\n\"How much have I actually spent?\"\n\nI had no idea.\n\nNot approximately. Not roughly.\n\nZero visibility into what was happening.\n\nThat's when I understood the real problem.\n\nIt's not about cost. It's about trust.\n\nClaude Code can read files, write code, run commands,\n\nand make decisions on your behalf.\n\nBut there's no dashboard. No audit trail.\n\nNo way to know what happened, why, or what it cost.\n\nWe're giving AI agents more and more autonomy\n\nwhile building less and less visibility into their actions.\n\nThat's a problem that compounds.\n\nI started with the most obvious gap:\n\ntoken visibility for Claude Code users.\n\nThe data was always there — sitting in\n\n~/.claude/projects/ as JSONL files.\n\nNobody told you it existed.\n\nSo I built a dashboard that reads it locally.\n\nNothing leaves your machine.\n\nWhat I found in my own data:\n\nThe real goal is bigger.\n\nThink about how we built trust around cars:\n\nAI agents need the same stack.\n\nThat's what I'm working toward:\n\nThe dashboard is the audit trail layer.\n\nAgentPass is the identity and permission layer.\n\nTogether, they're the foundation of\n\nwhat I'm calling the AI trust infrastructure.\n\nWe're at the moment right before the problem becomes obvious.\n\nMost companies still say \"ChatGPT is convenient.\"\n\nBut AI agents are already making decisions,\n\nsending emails, placing orders.\n\nThe question isn't whether AI will act autonomously.\n\nIt's whether humans will be able to understand,\n\nverify, and trust those actions.\n\nI'm building the infrastructure for that trust.\n\n```\nnpm install -g @notenkidev/claude-token-dashboard\nclaude-token-dashboard\n```\n\nGitHub: [https://github.com/notenkitoclient-cpu/claude-token-dashboard](https://github.com/notenkitoclient-cpu/claude-token-dashboard)\n\nIf you're building in this space —\n\nidentity, permissions, audit, cost accountability for AI agents —\n\nI'd love to connect.", "url": "https://wpnews.pro/news/i-m-building-the-trust-layer-between-humans-and-ai-agents", "canonical_source": "https://dev.to/notenkitoclientcpu/im-building-the-trust-layer-between-humans-and-ai-agents-4ho9", "published_at": "2026-06-04 06:36:12+00:00", "updated_at": "2026-06-04 06:41:42.784578+00:00", "lang": "en", "topics": ["ai-agents", "ai-infrastructure", "ai-tools", "ai-safety", "ai-startups"], "entities": ["Claude Code", "AgentPass"], "alternates": {"html": "https://wpnews.pro/news/i-m-building-the-trust-layer-between-humans-and-ai-agents", "markdown": "https://wpnews.pro/news/i-m-building-the-trust-layer-between-humans-and-ai-agents.md", "text": "https://wpnews.pro/news/i-m-building-the-trust-layer-between-humans-and-ai-agents.txt", "jsonld": "https://wpnews.pro/news/i-m-building-the-trust-layer-between-humans-and-ai-agents.jsonld"}}