// GUIDE — MACOS

Kalshi CLI Setup on macOS

May 21, 2026 · Yeti Games, LLC · 8 min read

Getting the Kalshi CLI running on macOS involves three separate problems that aren't documented anywhere official: a corrupted key error, a hardcoded production URL in demo mode, and a silent bug that makes your positions invisible. This guide covers all three.

1. Directory and Credential Setup

Store credentials in a hidden directory with restricted permissions:

mkdir -p ~/.kalshi
touch ~/.kalshi/.env
chmod 700 ~/.kalshi

Add to ~/.zshrc:

export KALSHI_ACCESS_KEY="your-uuid-here"
export KALSHI_PRIVATE_KEY_PATH="/Users/yourname/.kalshi/private_key.pem"
export KALSHI_ENVIRONMENT="demo"

Run source ~/.zshrc after saving.

2. Fixing the InvalidByte Error

If you copy your private key from a browser, hidden characters or smart quotes often corrupt the PEM file. The CLI throws:

ValueError: InvalidData(InvalidByte(4, 95))

Fix by converting to clean PKCS#8 format with OpenSSL:

openssl pkcs8 -topk8 -inform PEM -outform PEM -nocrypt \
  -in ~/.kalshi/private_key.pem \
  -out ~/.kalshi/private_key_fixed.pem

mv ~/.kalshi/private_key_fixed.pem ~/.kalshi/private_key.pem
chmod 600 ~/.kalshi/private_key.pem

This strips formatting artifacts and outputs a standard key the library accepts. If you don't have OpenSSL, install via Homebrew: brew install openssl.

3. Fixing the NOT_FOUND Error (Demo Mode)

If you're using demo credentials and the CLI returns NOT_FOUND on every request, the CLI is hitting the production server. Some builds are hardcoded to production.

Patch cli.py to read your environment variable:

import os

KALSHI_ENV = os.getenv("KALSHI_ENVIRONMENT", "prod").lower()

if KALSHI_ENV == "demo":
    BASE_URL = "https://external-api.demo.kalshi.co"
else:
    BASE_URL = "https://api.elections.kalshi.com"

Then add aliases to ~/.zshrc for easy switching:

alias kalshi-demo='export KALSHI_ENVIRONMENT=demo && echo "→ DEMO"'
alias kalshi-prod='export KALSHI_ENVIRONMENT=prod && echo "→ PRODUCTION"'

Run kalshi-demo before any paper trading session. Run kalshi-prod when trading real money.

4. Fixing the Missing Positions Bug

Symptom: kalshi positionsprints "No open positions" even when you have open trades.

Cause: Kalshi changed their API response format. The position field (integer) was replaced by position_fp (string like "14"). The CLI filters out positions where position == 0 — and since the old integer field is now missing, everything reads as zero and gets dropped.

Find cli.py — usually in ~/.local/share/kalshi-cli/ — and replace the two lines that load positions:

# Before (original two lines):
ps = data.get("market_positions", [])
ps = [p for p in ps if p.get("position", 0) != 0]

# After (replace with this block):
ps = data.get("market_positions", [])
for p in ps:
    if p.get("position", 0) == 0 and p.get("position_fp"):
        try:
            p["position"] = int(float(p["position_fp"]))
        except (ValueError, TypeError):
            pass
ps = [p for p in ps if p.get("position", 0) != 0]

No restart needed — the CLI reads cli.py fresh on each invocation. Run kalshi positions after the edit to verify.

5. Troubleshooting Checklist

Next Steps

Once the CLI is working, the next step is running the 5-step AI trading workflow. The Claude Kalshi Playbook ($17) covers the full process — from scanning for opportunities with /alpha to executing trades with /finalize.