Skip to content

pip install flintai-cli fails silently on Python < 3.13 — the error hides the real cause #4

Description

@JackKXCO

FlintAI is one of the sharpest takes on agent evaluation we've come across — the whitebox/blackbox split between scan and eval is exactly the right decomposition, and "understand context, not patterns" is the thesis this space has been missing. We've been putting it through its paces and came away impressed. One piece of adoption friction we hit in the first sixty seconds, though, is worth flagging, because it will quietly cost you installs.

Summary. On any Python older than 3.13, pip install flintai-cli fails with a "no matching distribution" error that is indistinguishable from a typo or an unpublished package — when the real cause is the Requires-Python: >=3.13 constraint.

Reproduce.

$ python --version
Python 3.12.10
$ pip install flintai-cli
ERROR: Could not find a version that satisfies the requirement flintai-cli (from versions: none)
ERROR: No matching distribution found for flintai-cli

The package is on PyPI and perfectly healthy — pip is simply filtering every release out because of the floor:

$ curl -s https://pypi.org/pypi/flintai-cli/json | python -c "import sys,json;print(json.load(sys.stdin)['info']['requires_python'])"
>=3.13

Why it matters. Python 3.13 only shipped in October 2024. A large share of the teams FlintAI is built for are still on 3.11 or 3.12 — it's the default in most cloud base images, CI runners, and the current Debian/Ubuntu LTS lines. And the failure mode is the unkind kind: it doesn't say "upgrade Python," it says "this package doesn't exist" — so an evaluator's very first impression is a dead end instead of a one-line version bump.

Two asks, in priority order:

  1. Make the requirement legible. A single line on the quickstart ("requires Python 3.13+") plus, if feasible, a friendlier resolver message would turn a silent dead-end into a 30-second fix.
  2. Consider lowering the floor to 3.11/3.12 if nothing 3.13-specific is load-bearing. If there's a hard dependency (a syntax feature, a typing construct, a transitive floor), that's completely fair — but if it's incidental, dropping to 3.11 roughly doubles the install surface for a tool that deserves to be everywhere.

Happy to test a pre-release against 3.11/3.12 if that would help.


Filed with genuine admiration from the KXCO team (kxco.ai) — building in the same neighbourhood (post-quantum security + AI trust) and rooting for FlintAI.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Fields

    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions