403 errors importing a web transcript, despite auth fine for listing sessions

[riwsky]

Steps to reproduce:

1. uvx claude-code-transcripts web --token $token --org-uuid $my_org_id
2. choose a session from the resulting list

Expected output: the html transcript of the selected session.
Actual output:

Fetching session $session_id...
Error: API request failed: 403 {"type":"error","error":{"type":"permission_error","message":"Permission denied"},"request_id":"req_011CWpedV4quKM89mRk83Vys"}

I'm assuming the credentials are correct, because why else would the remote session listing call succeed—but are there stricter auth requirements for GET-ing an individual session?

Context: the account is a personal claude max sub, then token is from claude setup-token, and I took the org-uuid manually from my ~/.claude.json.

Alternatives I tried, that got the same result:

• Just uvx claude-code-transcripts web and the auto-inferred auth; I tried with the explicit auth to just control for keychain or environment issues.
• explicitly specify a session id, as taken from the web ui, to control for potential issues parsing session picker output

Thanks for the tool and all of your writing, and happy new year!