OpenAI's Codex CLI runs agentic coding tasks straight from your terminal. Wire SprintFlint in and Codex pulls the next ticket, ships it, and posts the comment back.
Built around how your AI actually works.
export SPRINTFLINT_TOKEN=sf_pk_your_token_here
# smoke test
curl -H "Authorization: Bearer $SPRINTFLINT_TOKEN" \
https://sprintflint.com/api/v1/dashboard
Add a SprintFlint hook in your Codex config so it pulls ticket context before each task.
# ~/.codex/config.toml
[tools.sprintflint]
url = "https://sprintflint.com/api/v1"
auth = "Bearer $SPRINTFLINT_TOKEN"
usage = "Fetch ticket context before coding. Post a comment after committing."
codex "implement next SprintFlint ticket" — Codex fetches your sprint, picks the next undone ticket, codes, tests, and comments.
Same setup, different tools.
Now it tracks the work too.
300 free tickets. Set up before lunch.