payments-agent

Solo Founder OS agent #11 — overdue-invoice reminder drafter with HITL gating.

The 7th canonical layer of a one-person-company stack: monetization. Drafts reminder emails for overdue Stripe invoices, picks tone by severity (gentle / firm / final), queues every draft for human approval before send.

Built on solo-founder-os — uses its HitlQueue, AnthropicClient, log_outcome, and record_example primitives.

Why money is special

The agent's system prompt has hard rules other agents don't need:

• NEVER thank a customer for a payment that didn't happen.
• State the EXACT amount and currency from the invoice. No rounding.
• State the EXACT number of days overdue.
• NEVER invent a payment link.

These are asserted in tests so a future prompt rewrite can't drift them away.

Install

pip install payments-agent

Optional Stripe wiring (v0.2):

pip install 'payments-agent[stripe]'
export STRIPE_API_KEY=sk_live_...

Without STRIPE_API_KEY, the agent uses MockProvider with three fixture invoices — useful for dev, demos, dry-runs, and the "I don't have Stripe yet" onboarding path.

Usage

# Dry-run with mock invoices, queue 3 reminders to ~/.payments-agent/queue/pending/
payments-agent triage

# Show queue counts
payments-agent queue-status

# Approve / reject manually:
mv ~/.payments-agent/queue/pending/<file>.md \
   ~/.payments-agent/queue/approved/

# Or use sfos-ui (from solo-founder-os) for a unified dashboard across
# all SFOS agents:
sfos-ui

Severity ladder

Status

v0.1 — mock provider only. Real Stripe wiring lands in v0.2 once the agent has been validated against fixture data.

License

MIT.

🧩 Part of the Solo Founder OS stack

A growing collection of MIT-licensed agents that share solo-founder-os as their base — cron, eval, reflexion, AnthropicClient, HITL queue, notifiers. Each agent is independently useful; together they cover the full one-person company workflow.

🌐 The whole stack is live in production at vibexforge.com.

Each agent's own row is omitted from its README. Install whichever solve real problems for you — pip install <agent-name>.