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/luqueee/kivgraph/agents-mdgit clone --depth 1 https://github.com/Luqueee/kivgraphWhat 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 | $0.09652 | $0.09652 |
| Opus 5 | $0.04826 | $0.04826 |
| Sonnet 5 | $0.01930 | $0.01930 |
| Haiku 4.5 | $0.00965 | $0.00965 |
Grade B, and why
kivgraph AGENTS.md scanned grade B 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 2d 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.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
`~/.claude/settings.json`: comparte fichero con `claude-code` e instalar uno Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
escriben `curl` ni `tar`. How it starts
The opening of the file, as written. The whole thing — 672 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Instrucciones de desarrollo de Kivgraph
Estas reglas aplican a todo el repositorio. Una instrucción más cercana a un archivo puede añadir restricciones, pero no puede relajar los contratos de integridad, compatibilidad o verificación descritos aquí.
Mapa de instrucciones
Este archivo se lee siempre. Los demás cubren un directorio y se leen al trabajar en él; ninguno repite lo que ya está aquí ni lo contradice.
| directorio | archivo | qué cubre |
|---|---|---|
internal/ |
internal/AGENTS.md |
carga de Go, la pasada, caché de hechos, grafo canónico, generaciones, configuración, procesos |
internal/mcp/ |
internal/mcp/AGENTS.md |
superficie de tools, index_project, la skill, coste en tokens |
internal/rustloader/ |
internal/rustloader/AGENTS.md |
rust-analyzer scip, identidad SCIP, sysroot, descubrimiento Cargo |
cmd/kivgraph/ |
cmd/kivgraph/AGENTS.md |
ayuda, registro, index [--full], clean, stop, ui |
ts-worker/ |
ts-worker/AGENTS.md |
worker TypeScript e identidad cross-repository |
web/ |
web/AGENTS.md |
el visor: layout, dibujo, coste por fotograma |
landing/ |
landing/AGENTS.md |
landing y documentación de usuario: capas, paleta, SEO, iconos |
benchmarks/ |
benchmarks/AGENTS.md |
informes, corpus y auditorías |
Un invariante que se infringe desde otro directorio vive aquí, no en el archivo
del directorio que nombra: que landing/ no entre en ningún bundle se rompe
editando scripts/build-bundle.sh.
Identidad del proyecto
- Proyecto:
Kivgraph. - Módulo Go:
github.com/Luqueee/kivgraph. - Ejecutable principal:
cmd/kivgraph. - Worker TypeScript:
ts-worker/, paquete privado@kivgraph/ts-worker. - LadybugDB es el almacenamiento canónico; el HotSnapshot es una proyección derivada y no una fuente alternativa de hechos.
- Los identificadores históricos
LUQUE-####del backlog no se renombran.
Qué pregunta contesta cada tool de Kivgraph
Este bloque es el canal de enrutado portable: CLAUDE.md es un enlace a este
fichero, así que Oh My Pi y Claude Code lo cargan los dos sin que nadie lo pida.
El campo instructions del servidor dice lo mismo, y Zed no lo lee.
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.
- 2d ago First seen · 672 lines · 9,652 tokens per session scan B 8f41d1419615
kivgraph AGENTS.md is an instructions file published in the GitHub repository Luqueee/kivgraph (11 stars, last pushed 2d ago), licensed Apache-2.0. It adds 9,652 tokens to every session, about $0.0483 per session on Opus 5. A static security scan graded it B with 2 findings (reads agent configuration directories, 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
CodeStory AGENTS.md
Instructions for TheGreenCedar/CodeStory, covering codestory agent guide, start here, ownership boundaries, product invariants and identity and configuration.
CodeStory copilot-instructions.md
Instructions for TheGreenCedar/CodeStory: Use CodeStory proactively for repository questions. Do not wait for the user to mention it by name.
EngramGraph AGENTS.md
Instructions for AsiaOstrich/EngramGraph, covering agents.md, build & test, code style, git workflow and testing.
m_flow AGENTS.md
AGENTS.md instructions for FlowElement-xinliuyuansu/m_flow, covering m-flow — developer & agent reference, 1. repository map, extension points, 2. local development and python backend (requires python 3.10 – 3.13).
omnigraph AGENTS.md
Instructions for ModernRelay/omnigraph, covering omnigraph agent guide, required reading, repository snapshot, documentation map and engineering rules.
roam-code AGENTS.md
Instructions for Cranot/roam-code, covering agents.md — roam-code development guide, what this project is, documentation hub, where files go (private vs public) and quality discipline (from internal/dogfood/ + agi-in-md).