Daily sub-cap first
Every free tier gets a small per-day allowance. The monthly cap sits roughly 20–30× above it.
Why: A heavy day is isolated to that day. A heavy day under a bare monthly cap burns the whole month.
Pricing Report · For Review
Pricing & Usage Limits
A defensible limits model for Detect & Stage and Context Pack across three phases: early-bird feedback, public free tier, and the Pro fair-use ceiling. Every number is backed by a benchmark or a cost rationale.
Overview
Free tier design in developer tools follows one consistent pattern: a small daily sub-cap combined with a wider monthly cap. Bare monthly caps fail because a user exhausts them in two or three days and then waits the rest of the month.
Throttle, do not block
Pro tiers use a soft cap that slows the queue and lowers priority — not a hard stop.
Why: A block causes churn. A throttle keeps a power user inside Pro while protecting infrastructure.
Numbers stay private on Pro
Pro is marketed as “generous fair-use.” Exact ceilings live in Terms, not on the pricing page.
Why: A named number invites benchmarking and dispute. A fair-use clause is enforceable and invisible to normal users.
Benchmarks
Existing developer tools anchor every number in this report. Free tier is deliberately sub-cap heavy; Pro is either generous or silent.
| Tool | Free Tier | Pro Tier |
|---|---|---|
| ForgeCode | 10–50 requests/day (dynamic) | $20/mo → 1,000 requests/day |
| MCPEmails | 500/month + ~50/day sub-cap | Paid: silent fair-use ceiling |
| Jules | 2 concurrent tasks, 5 tasks/day | — |
| cubic | 10–20 PR reviews/month | — |
| Graphite Diamond | 100 PRs/month free | — |
| LiveReview | Unlimited reviews, up to 3 users | — |
| ContextPacker | 100 total free requests (no card) | — |
Takeaway: free tiers converge on a daily sub-cap. Pro tiers either publish a generous number or stay silent behind a fair-use clause. ZeroCommit follows both patterns.
Early-Bird (Friends & Family)
This phase exists to collect feedback, not revenue. Limits are generous but measurable — generous enough for deep testing, tight enough to prevent abuse and to produce usable usage data.
Detect & Stage
15–20 / dayTemporary cap; tightened after public launch.
Why 15–20: ForgeCode runs 10–50/day, but it is a mature product. Early-stage compute cost on repo parsing, file matching, and staging makes 15–20 the safe ceiling.
Why daily, not lifetime: Friends need 3–5 full loop runs (paste → detect → stage → review) to evaluate the product. A lifetime cap would end the evaluation on day one.
Context Pack
10 / dayTemporary cap; tightened after public launch.
Why 10: ContextPacker offers 100 total requests. A 10/day cap is orders of magnitude more generous than a lifetime cap, while still preventing repeated repacking of the same repo.
Why adjustable: Early-bird is a feedback phase. If a friend needs more, raise the cap manually — rigidity here costs feedback, not money.
Special rule — “feedback raises the limit.” A friend who submits structured feedback (bug report, UX issue, missing response format) gets their daily cap doubled.
Why: Tools like ForgeCode and Jules already prioritise active testers. Rewarding feedback turns a cost center into a signal channel.
Public Free Tier
After early-bird, caps tighten. The aim is a new user who can use the product for two to three weeks without exhausting the limit, but who cannot run bulk automation on a free account.
Detect & Stage
5–8 / dayMonthly cap: 100–150.
Why this combo: Mirrors MCPEmails' “500/month + 50/day” pattern at a smaller scale. The monthly cap is only reached by users who are genuinely serious.
Why not bare monthly: A 100/month cap alone lets a user burn it in two days. The daily sub-cap spreads usage across the month.
Context Pack
3–5 / dayMonthly cap: 30–50.
Why this combo: ContextPacker gives 100 lifetime requests. A 3–5/day cap over a month is already more generous than that, while a monthly cap of 30–50 stops bulk scanning.
Why the ratio: The daily sub-cap is roughly 1/20 to 1/30 of the monthly cap, so a single heavy day never exhausts the month.
Free tier rule. Keep the daily cap small. Keep the monthly cap roughly 20–30× the daily cap.
Why: This is the only shape that isolates a heavy day without punishing a heavy month.
Pro Plan ($10 / month) — Hidden Window
Pro does not say “unlimited.” It uses a silent fair-use ceiling: a soft daily cap that auto-throttles, plus a hard monthly ceiling that exists only to stop abuse.
Detect & Stage
150–200 / day (soft)Hard ceiling: 2,000–3,000 / month.
Why 150–200: ForgeCode Pro at $20/mo gives 1,000/day. ZeroCommit Pro is half the price, so a materially smaller daily cap is the honest positioning.
Why soft, not hard: Past the soft cap, the queue slows and priority drops. A block causes churn; a throttle keeps the power user inside Pro.
Why a monthly ceiling: Abuse protection. A script running 10,000 operations in one day must hit a wall.
Context Pack
50–75 / day (soft)Hard ceiling: 800–1,000 / month.
Why more generous than Detect & Stage: Context Pack token cost falls on the user's own AI API key (Mono is BYOK), so ZeroCommit's infrastructure cost is lower.
Why a ceiling anyway: To prevent automated bulk scanning from turning Pro into a scraping tier.
Positioning. Pricing page: “Generous fair-use limits. Contact us if you need more.” Terms page: “Pro plans include fair-use allowances designed for individual developers. Excessive automated usage may be throttled.”
Why this wording: It is the same approach MCPEmails uses — paid-tier numbers remain silent abuse ceilings, never customer-facing promises. No “unlimited” is claimed, and no normal user ever feels a limit.
Summary
| Early-Bird (Friends) | Public Free | Pro ($10/mo) Hidden | |
|---|---|---|---|
| Detect & Stage | 15–20 / day (temp) | 5–8 / day + 100–150 / mo | 150–200 / day soft → 2,000–3,000 / mo hard |
| Context Pack | 10 / day (temp) | 3–5 / day + 30–50 / mo | 50–75 / day soft → 800–1,000 / mo hard |
| Enforcement | Manual / soft | Hard daily + monthly | Soft throttle + hard fair-use |
| Copy | “Feedback raises the cap” | Clear caps shown | “Generous fair-use”, no numbers |
Important Note
Do not promise lifetime free or “special pricing forever” to early-bird friends. ForgeCode offered unlimited access during early access, then launched pricing tiers — and had to clean up the fallout.
Recommended terms language: “Early-bird feedback period limits are temporary. Standard plans apply after public launch. Feedback contributors will receive launch consideration, but no lifetime-free guarantee is made.”
Why: This preserves feedback collection, validates the limits, and avoids a trust hit when pricing is formalised at launch.