vg

A read-only map of a codebase that records files, symbols, and how code elements relate to one another. It helps an agent navigate code through structure and impact information rather than searching through many files.

In plain words
What is it for?
Use it to find symbols, trace relationships, inspect likely change impact, follow paths through the code, and locate central areas of a repository.
Why use it?
It reduces the amount of code an agent must read to understand a change and shows which parts may be affected.

Skill for Claude CodeCodex

Install

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.

agentmods
npx agentmods add skills/vibgrate/cli/vg
Any agent
npx skills add vibgrate/cli --skill vg
Clone the repo
git clone --depth 1 https://github.com/vibgrate/cli

Made for: Claude Code, Codex.

Per session 26 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 826 The whole file, excluding the scripts and references it only reads on demand.
Security scan C 1 finding. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00026 $0.00826
Opus 5 $0.00013 $0.00413
Sonnet 5 $0.00005 $0.00165
Haiku 4.5 $0.00003 $0.00083

Measured yesterday against content hash 0255fb5b326a, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade C, and why

vg scanned grade C 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.

Hidden instructionshighPrompt injection

Directives inside HTML comments, invisible characters or bidirectional overrides are read by the model and not by the person reviewing the file.

<!-- vg:v2 · managed by `vg install` — auto-refreshed when these instructions evolve; remove this line to opt out -->
plugins/claude/vg/skills/vg/SKILL.md · 67 lines

How it starts

The opening of the file, as written. The whole thing — 67 lines — stays where its author put it; the contents beside it link to each section on GitHub.

vg — the code map

This repo has a deterministic code graph built by vg. Prefer it over reading or grepping many files — it is smaller, more relevant context, and free.

Use the MCP tools — not the CLI

When the vg MCP server is registered (it is after vg install), always call its read-only tools instead of shelling out to the CLI. The server keeps the map parsed, the relation index warm, and the embedding model loaded across calls, so an MCP call answers in milliseconds — while every CLI invocation pays Node startup plus a fresh map parse, hundreds of times more. Tools: query_graph, get_node, impact_of, find_path, list_hubs, list_areas, get_graph_summary, search_symbols. They are side-effect-free and auto-approvable, and the server records which client is calling automatically. Reach for the CLI only when the MCP server is genuinely unavailable — never as the first resort.

CLI fallback — only when the MCP server is unavailable

If (and only if) no vg MCP server is available, use the CLI — and always pass --client=claude so your calls are counted (that's how the CLI-vs-MCP split is measured and the tools improved):

  • Understand code: vg "<question>" --client=claude — a budget-bounded, fact-annotated context block.
  • Find a symbol: vg show <name> --client=claude — what it is, what it calls, what calls it.
  • Before changing something: vg impact <name> --client=claude — what breaks if you change it.
  • Navigate: vg path <A> <B> --client=claude, vg tree <name> --client=claude.

Dependencies & library docs

  • Upgrade drift: vg drift lists what is outdated across dependencies (offline; --online for currency). vg scan scores upgrade drift and vg report renders it (text | json | sarif | md).
  • Version-correct docs: vg lib <name> returns drift-annotated, version- specific usage docs for a library — inject these instead of guessing an API.

Read the full file on GitHub · 67 lines

Changes

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.

  1. yesterday First seen · 67 lines · 26 tokens per session scan C 0255fb5b326a

Subscribe to this mod's changes

vg is a skill published in the GitHub repository vibgrate/cli (3 stars, last pushed 2d ago), licensed Apache-2.0. It adds 26 tokens to every session and 826 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it C with 1 finding (hidden instructions). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.