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/vladkesler/initrunner/claude-mdgit clone --depth 1 https://github.com/vladkesler/initrunnerWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/vladkesler/initrunner/claude-md)<a href="https://agentmods.dev/instructions/vladkesler/initrunner/claude-md"><img src="https://agentmods.dev/badge/instructions/vladkesler/initrunner/claude-md.svg" alt="Measured on agentmods" height="20"></a>What 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 | $0.03295 | $0.03295 |
| Opus 5 | $0.01648 | $0.01648 |
| Sonnet 5 | $0.00659 | $0.00659 |
| Haiku 4.5 | $0.00330 | $0.00330 |
Grade C, and why
initrunner CLAUDE.md scanned grade C with 2 findings 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 4d ago.
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.
Downloads and executes remote codehighSupply chain
curl | sh runs whatever the server returns today, which is not necessarily what it returned when this was reviewed.
| Installer script internals (`curl \| sh`) | `docs/getting-started/installer.md` | Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
| Installer script internals (`curl \| sh`) | `docs/getting-started/installer.md` | How it starts
The opening of the file, as written. The whole thing — 199 lines — stays where its author put it; the contents beside it link to each section on GitHub.
InitRunner
YAML-first AI agent platform: define agents as declarative role.yaml files (model, tools, triggers, ingestion, guardrails) and run them as one-shot commands, interactive REPLs, trigger-driven daemons, or OpenAI-compatible API servers. Multi-agent orchestration via flow.yaml.
Python 3.11+ with PydanticAI (agent framework), Pydantic (config/validation), Typer + Rich (CLI), SQLite (audit), LanceDB (vector stores), hatchling (build), uv (package manager). Web dashboard frontend is SvelteKit, managed with pnpm.
Project map
initrunner/ # main package (flat layout)
├── a2a/ # A2A (agent-to-agent) protocol server
├── agent/ # role schema/, loader, executor, capabilities/, runtime_sandbox/
│ └── tools/ # self-registering tool modules
├── audit/ # append-only SQLite audit trail
├── cli/ # Typer CLI entry point
├── credentials/ # credential resolution + encrypted vault
├── dashboard/ # FastAPI backend for the web UI
├── eval/ # eval framework: test roles against prompt suites
├── flow/ # multi-agent orchestration (flow.yaml)
├── ingestion/ # document pipeline: extract, chunk, embed, store
├── mcp/ # MCP server integration
├── packaging/ # OCI bundles: build, push, pull
├── registry/ # community role registry (install, search, manage)
├── runner/ # single-shot, REPL, autonomous, daemon modes
├── server/ # OpenAI-compatible HTTP API server
├── services/ # shared sync business logic (CLI is a thin wrapper over this)
│ └── presets.py # custom-provider preset definitions (OpenRouter etc.)
├── sinks/ # flow output sinks (file, webhook, custom)
├── stores/ # DocumentStore + MemoryStore ABCs and LanceDB impls
├── team/ # team mode: one-shot multi-persona collaboration
├── telemetry/ # opt-in anonymous usage telemetry (PostHog)
├── triggers/ # cron, file watcher, webhook triggers
├── _compat.py # optional dependency helpers
└── _html.py # HTML fetch + markdown conversion
dashboard/ # SvelteKit web UI frontend (pnpm workspace)
tests/ # pytest suite, mirrors source layout
examples/ # example role and flow YAML files
docs/ # detailed documentation by subsystem (index below)
scripts/ # release.sh, build-dashboard.sh, dev scripts
deploy/ # Docker entrypoint, fly.toml
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.
- 4d ago First seen · 199 lines · 3,295 tokens per session scan C 3d0ab429bd61
initrunner CLAUDE.md is an instructions file published in the GitHub repository vladkesler/initrunner (41 stars, last pushed 6d ago), licensed Apache-2.0. It adds 3,295 tokens to every session, about $0.0165 per session on Opus 5. A static security scan graded it C with 2 findings (downloads and executes remote code, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
vizra-adk CLAUDE.md
Instructions for vizra-ai/vizra-adk, covering vizra adk - laravel ai agent development kit, project overview, key concepts, agents and tools.
openacme CLAUDE.md
Instructions for sandydasari/openacme, covering openacme — claude code guide, workspace layout, the agent loop — request path, why sse-only and custom data parts.
Agent-Worlds CLAUDE.md
Instructions for hjl2004-10/Agent-Worlds, covering ai夸父 开发规范文档 (v2.0 - skill化重构), 架构总览：概念坐标系（先看这张图）, 横向：子系统（业界术语 ↔ 本项目目录）, 纵向：作用域分层（每个子系统内部都按此三层） and 正交二维（横×纵 = 网格，不是树）.
rustyhand AGENTS.md
AGENTS.md instructions for ginkida/rustyhand, covering repository guidelines, project structure & module organization, build, test, and development commands, coding style & naming conventions and testing guidelines.
progressive-agent CLAUDE.md
Instructions for Rayan55050/progressive-agent, covering progressive agent, что это, текущая фаза: v1.0.0 — open source release, стек and ключевые архитектурные решения.
Claude-Mind CLAUDE.md
Claude Code instructions for zkysar1/Claude-Mind, covering claude.md, project purpose, architecture, framework vs state split (4-tier architecture) and core design principle: no terminal state.