anthropics / anthropics/claude-code

[BUG] False-positive Usage Policy trigger on benign Jira licensing questions

Ouverte
#91,335 0 commentaires 0 réactions 0 personnes assignées Voir sur GitHub
api:anthropic area:model bug platform:macos
Langage dominant
Python
Étoiles
145k
Forks
23.1k
Métriques de merge des PR
Métriques de PR en attente

Description

### Preflight Checklist

- [x] I have searched [existing issues](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20state%3Aopen%20label%3Abug) and this hasn't been reported yet
- [x] This is a single bug report (please file separate reports for different bugs)
- [x] I am using the latest version of Claude Code

### What's Wrong?

Repeatedly hit `API Error: Claude Code is unable to respond to this request, which appears to violate our Usage Policy` while discussing completely benign internal IT administration topics — specifically, Jira Service Management Agent vs. Customer role licensing, and whether "User access admin" can be granted separately from "Agent" in Atlassian's Grant Access flow. No policy-violating content was involved.

Occurred multiple times in the same conversation, same general topic each time.

Request IDs:

* req_011CedRr8nGMSK7TuF9JUZGJ
* req_011CedS3Yt9jvTWJyVh9p3zf
* req_011CedT6VkapphDVP3tyiF4j

I searched existing issues before filing (title search for "Usage Policy", combined with "false positive", "Jira", "Atlassian licensing", "Grant Access"/"Service Management"/"access admin", and the request IDs above). There are many other Usage Policy false-positive reports on file, but none involving Jira/Atlassian licensing or access-role administration topics, so this appears to be a distinct case.

### What Should Happen?

The requests should complete normally. Discussing Jira Service Management licensing roles (Agent vs. Customer) and whether "User access admin" can be granted independently of "Agent" in Atlassian's Grant Access flow is ordinary IT-administration content and does not violate the Usage Policy. The classifier should not flag this topic.

### Error Messages/Logs

```shell
API Error: Claude Code is unable to respond to this request, which appears to violate our Usage Policy
```

### Steps to Reproduce

1. Start a conversation with Claude Code about Jira Service Management administration.
2. Ask about the difference between the Agent and Customer roles for licensing purposes.
3. Ask whether the "User access admin" permission can be granted separately from the "Agent" role via Atlassian's Grant Access flow.
4. Continue the conversation on this same general topic.
5. Observe that at some point during the conversation, the request is blocked with: "API Error: Claude Code is unable to respond to this request, which appears to violate our Usage Policy."

This happened multiple times within the same conversation, on the same general topic (Jira licensing/role administration) each time. No policy-violating content was requested or discussed at any point.

### Claude Model

None

### Is this a regression?

I don't know

### Last Working Version

_No response_

### Claude Code Version

Not specified

### Platform

Anthropic API

### Operating System

macOS

### Terminal/Shell

VS Code integrated terminal

### Additional Information

_No response_

Guide de contribution

Aucun guide de contribution indexé pour ce dépôt

Piste de recherche

Start by reproducing the listed Jira Service Management licensing conversation in Claude Code through the Anthropic API, using the reported request IDs and error message as evidence. Done means the benign Agent, Customer, and User access admin questions complete without a Usage Policy block; the payload names no source file or test to inspect.

Rédigé par le modèle d'indexation à partir du texte de l'issue.

Évaluation

Stack technique
python
Domaine
api, security
Type d'issue
Bug
Difficulté
5/5
Temps estimé
Plus d'une semaine
Activité
Active
Clarté
Plutôt claire
Accessibilité débutants
35/100

Recevez les nouvelles issues par e-mail

Un résumé court des issues GitHub adaptées aux débutants.