What is mcp-grounding CLAUDE.md?

A set of standing instructions for working on the mcp-grounding code project. It points the coding agent to the project plan and notes known problems and rules.

Guiding code changes in the mcp-grounding project, especially work involving its plan, known issues, scoring file, prompts, test data, and protocol errors.

It keeps work aligned with the project's agreed scope, tests, and error behavior. It also helps prevent changes that would make evaluation results incomparable.

How alxnhfr-bit/mcp-grounding configures Claude Code

Reuse

Borrowing it

Nothing to install: this file belongs to alxnhfr-bit/mcp-grounding. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file

curl -O https://raw.githubusercontent.com/alxnhfr-bit/mcp-grounding/main/CLAUDE.md

Clone the repo

git clone --depth 1 https://github.com/alxnhfr-bit/mcp-grounding

Made for: Claude Code.

Wrote 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.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/alxnhfr-bit/mcp-grounding/claude-md/github.svg)](https://agentmods.dev/instructions/alxnhfr-bit/mcp-grounding/claude-md)

Your own site

<a href="https://agentmods.dev/instructions/alxnhfr-bit/mcp-grounding/claude-md"><img src="https://agentmods.dev/badge/instructions/alxnhfr-bit/mcp-grounding/claude-md/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/instructions/alxnhfr-bit/mcp-grounding/claude-md"><img src="https://agentmods.dev/badge/instructions/alxnhfr-bit/mcp-grounding/claude-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 603 This file is loaded in full into every session.

When invoked 603 The same file — it is already loaded in full.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

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.

Model	Per session	Once invoked
Fable 5.1	$0.00603	$0.00603
Opus 5	$0.00302	$0.00302
Sonnet 5	$0.00121	$0.00121
Haiku 4.5	$0.00060	$0.00060

Measured 9d ago against content hash 0c702c651120, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

mcp-grounding CLAUDE.md scanned grade A with 0 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 9d 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.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

CLAUDE.md · 39 lines

What it actually says

Project context for Claude Code

Read PLAN.md for scope and acceptance criteria, FRICTION.md for what has already gone wrong. This file holds only the standing rules.

Hard rules:

The measurement is the deliverable, not the server. A working MCP server proves someone can follow a tutorial. Do not stop when it runs.
Never write a new scorer. eval/score.mjs is copied byte-identical from camunda-trial (sha256 babd9cc5cb4122246ae19b6a151d60f66e8ae5c7b90e102e323d1a5f23a591e7). Reusing exactly that file is the only thing making the two numbers comparable. No model anywhere in the judging loop, ever.
Never soften the error semantics to make something pass. A place that does not exist is a successful result carrying found: false. Only transport, auth and quota failures are protocol-level errors, and they must reject the client call rather than return isError: true — see FRICTION.md 2026-08-06. If a case fails, the tool is wrong or the description is wrong.
Never edit prompts/classifier-system-prompt-v3.md or eval/golden_set_v2_1.csv. They are copies of the camunda-trial artifacts. The rubric's sha256 parity is asserted by the test suite; changing it breaks the comparison.
Tool descriptions are append-only. Every version ever run against the golden set stays in src/server/tool-description.ts with its measured result in docs/TOOL_DESCRIPTIONS.md. A description that scored badly is evidence, not garbage.
No credentials in tracked files. .env is gitignored and stays that way. The API key is server configuration read from the environment, never a tool input — the model must not be able to see it or supply one.
stdout belongs to the JSON-RPC stream. In anything that runs over stdio, all logging goes to stderr. One stray console.log corrupts the transport.
At most two fix attempts, then write the FRICTION.md entry before continuing. The SDK moves faster than the assistant's training data; when something does not behave as expected, read the installed .d.ts and .js in node_modules/@modelcontextprotocol/sdk/dist/esm/ rather than trying a third time from recall.
No scope expansion. One tool, one taxonomy, one golden set. This is a gap-closer with a measurement attached, and its job is to be finished.

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.

9d ago First seen · 39 lines · 603 tokens per session scan A 0c702c651120

Subscribe to this mod's changes

mcp-grounding CLAUDE.md is an instructions file published in the GitHub repository alxnhfr-bit/mcp-grounding (0 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 603 tokens to every session, about $0.0030 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

next.js AGENTS.md

AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.

vercel/next.js · 7,296 tokens

codex AGENTS.md

AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.

openai/codex · 5,153 tokens

vscode buildNext.instructions.md

Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).

microsoft/vscode · 6,785 tokens

mcp-grounding: Instructions file for Claude Code