Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/netdata/ai-viewer/agents-mdgit clone --depth 1 https://github.com/netdata/ai-viewerWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.11604 | $0.11604 |
| Opus 5 | $0.05802 | $0.05802 |
| Sonnet 5 | $0.02321 | $0.02321 |
| Haiku 4.5 | $0.01160 | $0.01160 |
Grade A, and why
ai-viewer AGENTS.md scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
| Realtime transport | **SSE + REST** | trivially debuggable with `curl -N`; browser EventSource built-in; reconnect automatic | How it starts
The opening of the file, as written. The whole thing — 654 lines — stays where its author put it; the contents beside it link to each section on GitHub.
ai-viewer
A read-only, real-time explorer for AI coding-agent session snapshots. Multi-format: ingests ai-agent (v2 and v3), claude-code, codex, and opencode session storage formats, normalizes them into a canonical model, and presents them through a modern dark/light web UI with span-based tracing, topology, timeline, and statistics views.
⚠️ CURRENT PHASE: DEVELOPMENT (started 2026-06-14). The app is unreleased, not installed anywhere, zero users, zero production risk. The PR-per-SOW rule and master branch protection are SUSPENDED for this phase. Work goes directly to
master. CI / Codacy / cubic / CodeQL still run on every push tomasteras defense-in-depth; the CTO reads their findings and addresses the real ones. External reviewers are used as three milestone quality gates on meaningful chunks of work: gap analysis, implementation plan, and implementation review. See the "Phase: Development" section below for the full override list. This phase ends when the operator declares GA; at that point the operator decides which branch/PR protections return.
Goals
- Primary purpose: give the operator a fast, beautiful, low-friction way to see what their AI coding agents have been doing — across time, across formats, across sub-agents.
- Single source of truth: read source-system snapshot files directly; never call live agent systems.
- Real-time: file-watch the source directories; push updates to the browser without polling.
- Multi-format and extensible: an adapter is one Go package implementing one interface; new formats are additive, never schema-breaking.
- Mental model first: anyone (operator or contributor) should be able to read the specs and know exactly what the system does and why.
- Tested, working, production-quality code: no half-built features, no silent failures, no untested code paths.
Phase: Development (active — started 2026-06-14)
The project is in active development: unreleased, not installed anywhere, zero users, zero production risk. Process weight matches risk weight. The rules in this section override the PR-per-SOW and master branch protection rules during this phase. When the operator declares GA, this section is deleted or revised.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 654 lines · 11,604 tokens per session scan A e45135f80055
ai-viewer AGENTS.md is an instructions file published in the GitHub repository netdata/ai-viewer (2 stars, last pushed yesterday), licensed MIT. It adds 11,604 tokens to every session, about $0.0580 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
ai-localbase AGENTS.md
Instructions for veyliss/ai-localbase, covering agent instructions and 回复格式.
token-burn AGENTS.md
Instructions for durandom/token-burn, covering agents.md - token-burn, project intent, language direction, layout and code style.
tuitube AGENTS.md
Instructions for gitcoder89431/tuitube, covering agents.md, project intent, checklist for adding a new feature, key packages and logging.
agentinsight-sdk-python AGENTS.md
Instructions for AgentInsight/agentinsight-sdk-python, covering agentinsight python sdk — 智能体执行规则, 1. 项目基本信息, 1.1 技术栈, 1.2 项目关键特征 and 1.3 项目目录结构.
agentreceipt AGENTS.md
Instructions for ametel01/agentreceipt, covering agentreceipt repository instructions, instruction version, pilot baseline capture, agent skills and issue tracker.
lastsaas CLAUDE.md
Instructions for jonradoff/lastsaas, covering lastsaas development rules, validation, system logging, build verification and dependent project deployment (critical).