Skip to content
IsoKron

No wording that overrides instructions or hides actions from the user

Phrases such as "ignore previous instructions", "do not tell the user" or "without asking the user" instruct the model instead of describing the tool, and both directories reject them.

Check sec-instruction-phrases · Poisoning and hidden text · default severity: blocking · rubric 2.1.0

What fails

Phrases such as "ignore previous instructions", "do not tell the user" or "without asking the user" are instructions to the model, not a description of the tool.

The rule

Tell Claude to behave in ways unrelated to the tool’s function, attempt to override system instructions, or promote products and services

Source: Anthropic connector pre-submission checklist

How to fix it

Rewrite the text to describe what the tool does and returns.

  1. Remove wording that overrides instructions, hides actions or skips confirmation.
  2. Rewrite the text to describe what the tool does and returns.
  3. Add a CI test with the phrases you never want to ship.

A CI test that fails when the pattern comes back, run against the saved tools/list answer:

TypeScript (node:test)

TypeScript test
// tools-list.test.mjs: run `node --test` after saving the answer with
// npx @modelcontextprotocol/inspector --cli <server> --method tools/list > tools.json
import assert from 'node:assert/strict';
import { readFileSync } from 'node:fs';
import { test } from 'node:test';

const answer = JSON.parse(readFileSync('tools.json', 'utf8'));
const tools = answer.tools ?? answer.result?.tools ?? answer;

// Every string and every key in a tool, with its path.
function* strings(value, path) {
  if (typeof value === 'string') yield [path, value];
  else if (value && typeof value === 'object')
    for (const [k, v] of Object.entries(value)) {
      yield [`${path}.${k} (key)`, k];
      yield* strings(v, `${path}.${k}`);
    }
}

const PATTERN = /ignore (all |any )?(previous|prior|above) instructions|do not (tell|inform|mention to) the user|without (asking|telling|notifying) the user/i;

test('no instruction-override wording', () => {
  const hits = [];
  for (const tool of tools) for (const [path, text] of strings(tool, tool.name)) if (PATTERN.test(text)) hits.push(path);
  assert.deepEqual(hits, [], 'wording that overrides instructions or hides actions from the user');
});

Python (pytest)

Python test
# test_tools_list.py: run `pytest` after saving the answer with
# npx @modelcontextprotocol/inspector --cli <server> --method tools/list > tools.json
import json
import re

with open("tools.json", encoding="utf-8") as f:
    answer = json.load(f)
tools = answer.get("tools") or answer.get("result", {}).get("tools") if isinstance(answer, dict) else answer


def strings(value, path):
    """Every string and every key in a tool, with its path."""
    if isinstance(value, str):
        yield path, value
    elif isinstance(value, dict):
        for k, v in value.items():
            yield f"{path}.{k} (key)", k
            yield from strings(v, f"{path}.{k}")
    elif isinstance(value, list):
        for i, v in enumerate(value):
            yield from strings(v, f"{path}.{i}")


PATTERN = re.compile(r"(?i)ignore (all |any )?(previous|prior|above) instructions|do not (tell|inform|mention to) the user|without (asking|telling|notifying) the user")


def test_no_instruction_override_wording():
    hits = [path for tool in tools for path, text in strings(tool, tool["name"]) if PATTERN.search(text)]
    assert hits == [], "wording that overrides instructions or hides actions from the user"

Questions

What about "always confirm with the user before deleting"?

That reads as behavior the client should already provide through destructiveHint. Set the hint and describe the effect instead.

Is this only a problem for malicious servers?

No. Honest servers ship such wording to "help" the model, and reviewers reject it all the same.

How this check is graded

It counts toward: Poisoning and hidden text, Anthropic connector directory, OpenAI plugin directory. See the rubric.

Last verified against Anthropic connector pre-submission checklist on 2026-11-18. The rule above is quoted in the source’s own words. This page as Markdown.

Check your tool list: paste your tools/list, nothing is sent anywhere