lens.01 product-owner.md
The product owner
Sees what the change does in plain English and a before/after preview of the actual screen. Approves or says "not quite — here's what I meant."
maintain · new capability
The
The control room. Your team ships the small stuff —
without us, without waiting.
Maintain already keeps what we build working. The control room goes further: we deploy a guard-railed build pipeline inside your own infrastructure, on your Claude credits, so your PMs, founders, and engineers can ship copy fixes, config changes, and small bugs — from their phone, in plain language — while the risky work still routes to a human.
~/clients/your-co — control-room.log
$ specshop maintain –control-room –status
illustrative
runs_in
your_vpc // nothing leaves your perimeter
model_credits
your_anthropic_key
operators
pm · founder · engineer // three lenses, one agent
blast_radius
defined_by_spec // the safe zone
last_ship
fix: pricing page typo — amara (pm) — 2m ago ✓
$
// 01 — the math behind the control room
One unblocked week pays for the whole month.
Every trivial change a founder files — a typo, a broken link, a copy tweak, a feature flag — is a context-switch stolen from an engineer and a wait imposed on the business. The control room moves that whole class of work off the engineering queue entirely. Engineers stop being interrupted for the small stuff. The team stops waiting for it.
// 02 — one agent, three lenses
The same change, reviewed at the altitude the operator is comfortable with.
Codex and Cursor show everyone a code diff and ask them to think like an engineer. That's a wall for non-technical operators and overkill for a typo. The control room adapts the review surface to who is holding the phone — that's the part no IDE-shaped tool can copy.
lens.02 founder.md
The founder
Everything the PM sees, plus an optional "show me what changed" — a readable summary of the edit, not raw code. Ships with a glance.
lens.03 engineer.md
The engineer
Full diff, jump into the change, full control. For them the diff is the fastest review — so they get it untouched.
// 03 — what it costs
A one-time deploy, then a monthly retainer. Credits and infra stay yours.
We charge for the engineering — standing up the pipeline and keeping it working. The model credits and cloud bill stay in your accounts, which is exactly what makes 'nothing leaves your perimeter' structurally true rather than a slogan.
solo.md
Solo
One product, a handful of operators. The founder + a PM shipping the small stuff.
$750/mo
// month-to-month · cancel anytime
- ◆ 1 repo under management
- ◆ up to 3 operators
- ◆ pipeline uptime + dependency updates
- ◆ guard-rail config within spec
team.md
Team
A growing product team. More operators, more change volume, more tuning.
$1,500/mo
// month-to-month · cancel anytime
- ◆ up to 2 repos
- ◆ up to 8 operators
- ◆ everything in solo
- ◆ monthly guard-rail review
- ◆ consumption reporting
multi.md
Multi
Multiple products or a larger org running the control room across teams.
$3,000/mo
// month-to-month · cancel anytime
- ◆ 3+ repos
- ◆ unlimited operators
- ◆ everything in team
- ◆ priority response
- ◆ quarterly safe-zone expansion review
// billed monthly · 30 days notice to cancel · no long-term contract
// what you pay your own vendors directly
Credits and infra stay yours.
The moment we hold the key or the server, the sovereignty story breaks and you inherit our vendor risk. Keeping them in your accounts is the design, not an inconvenience.
anthropic credits
~$300/mo*
your key, your account. *budgeted per change-type — we tell you the real number up front.
cloud infra
your VPC
runs where your product already runs. no new vendor.
domain · SSL
your registrar
standard, nominal, fully yours.
// 04 — what 'maintain' means, exactly
The retainer keeps it working. New ground is a new engagement.
// included in the retainer
- pipeline uptime and monitoring inside your infra
- dependency, SDK and security updates
- guard-rail config within the existing safe zone
- operator support across the three lenses
- consumption visibility (team tier and up)
// a new engagement
- expanding the safe zone to new change-types
- new repos brought under management
- new integrations or MCP boundaries
- feature builds beyond the small-change envelope
- re-speccing after a major architecture shift
Thirty minutes. We'll tell you whether the control room is worth it.
If your engineering queue is full of two-minute fixes that take two weeks, this is for you. If it isn't, we'll say so.