// FOR AGENCY OWNERS
You signed the contract.
Now find out what's actually in there.
Ghost Architect™ is the pre-engagement codebase triage tool that agency owners use to bid with confidence, protect margin, and ship without surprises. The audit you would have paid a senior engineer two weeks to do, in an afternoon, for under five dollars. Ghost Open™ is free for one person on your own Anthropic API key, or Ghost Architect™ Local runs on hardware you own, for your team and for the clients whose code cannot leave the building.
// the problem
Most agency engagements lose money in the first two weeks.
It is almost never the team's fault. The team executes. The problem is upstream: somebody scoped a contract on a codebase nobody had ever read. The estimate was a guess wrapped in confident language. By the time engineering opens the repo, the price is locked, the timeline is signed, and the only variable left is margin.
Average time before a poorly scoped engagement starts bleeding budget.
Of agency engagements run over budget on the first sprint. The cost comes out of margin, not the client.
Typical write-off per misscoped Adobe Commerce or WordPress engagement. Sometimes much more.
The math is brutal: a single misscoped engagement can eat the profit from three correctly scoped ones. And the bid-side problem compounds. The senior engineer who would have caught the issue during scoping is on three other engagements. The architect with the relevant expertise is fractional. The codebase is fifty thousand lines of someone else's PHP. The scoping call is Thursday. The proposal is due Friday.
So the proposal goes out the door with a number that came from comparing this engagement to the last engagement and rounding up by twenty percent. Sometimes that works. Forty-two percent of the time it does not, and the margin disappears before the kickoff call.
Ghost Architect™ exists to close that gap. Same scoping engineer, ten times the speed. Repeatable. Auditable. Cheap enough to run on every prospect.
// the eight questions
Eight questions every agency owner
should be able to answer before signing.
Most agencies can answer two of these on a good day. By the time they can answer the rest, they have already signed the contract, scoped the engagement, and lost margin on the first sprint. Ghost Architect™ answers all eight in an afternoon, for the cost of a coffee.
What is actually wrong in here?
What breaks if we change this?
Where is the code arguing with itself?
Should we even bid on this work?
What did we just inherit?
Can I just ask it questions?
Did this get better or worse?
Are we trending up or down?
If you cannot answer these before you bid, you are not bidding. You are gambling. The agencies that win on Adobe Commerce, Magento, WordPress, Drupal, and headless commerce engagements have already figured out that the bid is the deliverable. Ghost Architect™ is how they make that bid honest.
// return on investment
The math is not subtle.
Ghost Open™ costs nothing: one person, every mode, on your own Anthropic API key. Ghost Architect™ Local is eighty dollars a month for one seat, or a thousand dollars a month for fifteen. A single avoided misscope pays for either licence for years. Most agencies see ROI on the first engagement they scope with Ghost.
Bid with confidence, not hope.
Every engagement you bid on costs you scoping time. Senior engineers at one hundred fifty to two hundred dollars an hour. Architects at two-fifty plus. Multiply by the bids you lose. Ghost Architect™ runs a full triage scan in under ten minutes for cents, then produces a PDF you can hand to a sales engineer or read yourself.
Protect margin on every engagement.
Forty-two percent of agency engagements run over on sprint one. The most common cause is upstream: scope was wrong because the codebase was unknown. Ghost surfaces the unknowns before the SOW is signed, when you can still adjust pricing, scope, or walk away.
Win more of the bids you bid on.
Walking into the scoping call with a Ghost Architect™ report changes the conversation. You are not estimating in the abstract. You are pointing at specific findings, specific files, specific risk levels. Prospects close faster on agencies that show their work. Your bid stops being one of three guesses. It becomes the bid that knows.
Ship audits in hours, not weeks.
An Inheritance Audit deliverable that used to take a senior engineer two to four weeks now takes about sixty seconds of compute time. The PDF is client-ready. Stack reality, key-person risk, end-of-life dependencies, ninety-day modernization roadmap. Hand it to the buyer the same day they ask.
Show clients you saved them money.
Run Ghost before the engagement. Run it again at the midpoint. Run it at the end. Use Compare Reports, and the Project Dashboard in Ghost Architect™ Local, to show the client exactly which findings closed, which got worse, and which were introduced. Renewal conversations become data conversations.
White-label as your own service.
Ghost Architect™ Local ships with Ghost Partner™ profiles. Load a profile and every report carries your firm's logo, your firm's brand colors, your firm's methodology in the executive summary. The deliverable looks like an audit your firm did. Because it is.
Ghost Brief™ -- the AI fix loop.
Ghost Brief™ converts every finding into a validated, blast-radius-aware prompt your AI coding agent executes immediately. Ghost finds it. Ghost tells the AI exactly what to do. The AI does it. You deliver faster, with less risk, and a PDF your client can sign off on.
// the math, with real numbers
One engagement. Two outcomes.
An agency wins a $120,000 Adobe Commerce migration. Three months of work, two senior developers, scope estimated from a forty-five minute scoping call and a quick repo walkthrough. Pretty standard. Here is what happens with and without a pre-engagement Ghost scan.
Without Ghost Architect™
| Engagement value | $120,000 |
| Estimated cost (engineer-hours) | $78,000 |
| Expected gross margin | $42,000 |
| Scoping miss: undocumented third-party module rewrite | +$22,000 |
| Scoping miss: deprecated PHP functions in 14 custom modules | +$8,500 |
| Scoping miss: missing CSRF protection found late, hot-fix | +$3,200 |
| Final cost | $111,700 |
| Actual margin | $8,300 |
With Ghost Architect™
| Engagement value | $132,000 |
| Scoping cost (Ghost scan) | $0.45 |
| Ghost Open™ (one person) | $0 |
| Estimated cost (engineer-hours) adjusted for findings | $91,000 |
| Expected margin (adjusted) | $41,000 |
| Unknown unknowns at start | Very few |
| Final cost | $93,200 |
| Actual margin | $38,800 |
Margin delta on a single engagement. Enough to run a fifteen-seat Ghost Architect™ Local Agency licence for thirty months.
Notice what changed. The Ghost-scoped engagement bid $12,000 higher because it knew what it was bidding on. The third-party module rewrite was found during scoping, not sprint two. The PHP deprecations were flagged before the kickoff call. The CSRF gap was a line item, not a surprise. And the agency walked into the engagement with a forty-thousand-dollar margin instead of an eight-thousand-dollar one.
This is not theoretical. This is the math that runs every agency engagement, just usually with the second column hidden because nobody ran the analysis upstream.
// when the client code cannot leave the building
Same scans. Same reports.
On hardware you own.
Some engagements come with an NDA that forbids sending source to any third party, a security questionnaire that asks where the model runs, or a client who will not have their code in anyone's cloud. Ghost Architect™ Local is the same product with the model on your machines: a hub in the office, spokes on the engineers' laptops, one queue for the whole firm, and nothing leaving your network. The current release is version 7.4.42.
Zero egress you can prove
Turn zero-egress mode on and every outbound integration is refused at the code level. The product prints its own egress inventory, generated from the code that gates the traffic, so the answer to "where does our code go" is a printout, not a promise.
One monthly licence, no meter
The Agency licence is $1,000 a month for 15 seats. No per-token bill, no per-scan cost, no surprise invoice after a big codebase. A bad answer costs you a re-run, not money.
A ledger for the client report
Every scan that ends is a row: who ran it, which project, when, and how it ended. Exports are CSV or PDF, signed. Hand the client the ledger with the findings.
// who this is for
If any of these describe you,
Ghost is built for your firm.
You sign the bids that go out the door.
You know which engagements are profitable and which ones bled. You have signed contracts you wish you had not. Ghost is the analyst you would hire if you could afford to. Free for one person with Ghost Open™, on every bid, with no margin-of-error issue.
You audit codebases for buyers, PE, and acquirers.
Your billable rate is high. Your time is the bottleneck. Ghost is leverage. Run the scan, review the findings, write the executive summary, deliver the engagement. Charge for the judgment, not the typing. Ten times the throughput, same firm rate.
You run an Adobe Commerce, Magento, or WordPress practice.
Your senior engineers are the bottleneck on every scoping call. Ghost runs the first pass so they only weigh in on the findings that matter. The practice ships proposals faster. The senior engineers stop being a single point of failure on bids.
You have eight to fifty engineers shipping client work.
Margin is the number you watch. Ghost Architect™ Local Small Team is five seats at four hundred dollars a month, and Local Agency is fifteen seats at a thousand, with one hub and one queue for the firm. Every scoping call gets a triage report. Every senior engineer reviews findings instead of generating them. Margin holds.
You sell expertise, not bodies.
Your differentiation is judgment. Ghost makes your judgment look like a service, not an opinion. A Ghost Partner™ profile in Ghost Architect™ Local loads your methodology into every scan. Clients hire you for the framing, not the report. The report comes with the engagement.
You do technical due diligence on acquisitions.
Inheritance Audit mode is built for you. Sixty seconds, two cents, deal-grade PDF. Stack reality, key-person risk, hidden dependencies, ninety-day modernization roadmap. Compare across portfolio companies with the Project Dashboard in Ghost Architect™ Local. Local Enterprise is built for portfolios.
// honest answers
The questions you are actually thinking right now.
Does this replace my senior engineers?
No. It makes them more valuable. Ghost surfaces the findings; your senior engineers decide what matters. The triage report tells your team where to look first. The judgment, the architecture, the actual remediation work -- all human. Ghost is the part of the senior engineer's job that should never have been senior engineer work in the first place: reading fifty thousand lines of someone else's PHP at one in the morning before a scoping call.
Is my client's code uploaded to your servers?
No. There are no Ghost Architect servers. Ghost is a CLI tool that runs entirely on your local machine, or your engineer's machine, or your CI runner. Your client's source code never leaves your infrastructure. The only network calls are direct from your machine to Anthropic's API, using your own API key, under your own data agreement. Anthropic deletes API inputs and outputs within seven days per their data retention policy. There is no SaaS between you and them.
How does this compare to CodeRabbit, Kodus, or other AI PR review tools?
Different moment, different buyer. AI PR review tools live inside your team's Git workflow. They comment on pull requests. They run continuously, after the engagement has started and the team is in the codebase. Ghost runs before that. During due diligence. During scoping. During pre-engagement audit. When you need to understand a codebase you do not own yet. Many agencies use both: Ghost to win the bid and scope the work, CodeRabbit or Kodus to keep the engagement healthy once it is underway.
What does it actually cost to run a scan?
Anthropic API charges. A typical scan costs twenty to fifty cents in API usage, paid directly to Anthropic on your own key. Recon mode is about five cents. A full Inheritance Audit is around two cents. There are no hidden fees from Ghost. Ghost Open™ is free for one person. The API cost is whatever Anthropic charges for the tokens you actually use. Ghost prints the exact cost of every scan in the terminal so you can verify.
My team will not use a new tool. Adoption will fail.
Ghost is a command-line tool. Run ghost in a terminal, point it at a folder, pick a scan mode. With Ghost Open™, there is nothing to install beyond an npm package. There is no dashboard to log into. There is no workflow to adopt. The deliverable is a PDF you hand to whoever needs to read it. Most of the time, the people who use Ghost are you and one or two senior engineers, not the whole team. That is fine. The output is what matters.
What about white-labeling? My firm's brand matters.
Ghost Architect™ Local ships with Ghost Partner™ profiles. A YAML profile injects your firm's branding into every PDF: logo on the cover, brand colors throughout, your methodology in the executive summary, your billing rates in the cost estimates. The deliverable looks like an audit your firm did, because it is. You did the work. Ghost is the tooling. Ghost Open™ reports always carry Ghost Architect™ branding; white label is in Ghost Architect™ Local.
What if my engagement is on PHP, not just Magento?
Ghost works on any codebase, any language, any platform. PHP, JavaScript, TypeScript, Python, Ruby, Java, Go, Rust, C#, anything Anthropic's API can read. Adobe Commerce and Magento are common use cases because that is what the founder built it for first, but it is not a requirement. Run Ghost on a React app, a Django backend, a Laravel monolith, a Node service, a Spring Boot app -- same scan engine, same triage methodology, same deliverable.
What is the actual catch?
You bring your own Anthropic API key. That is the catch. We do not resell tokens. You pay Anthropic directly for what your scans consume, which is usually a few dollars a month for normal usage. The advantage: you own the data agreement with Anthropic, you set your own retention preferences, you have direct visibility into what your scans cost. There is no markup, no middleman, no surprise bill. Ghost Open™ costs nothing for one person. Anthropic charges you for the tokens, and that is the only bill.
Why hasn't an established player built this already?
Honest answer: because pre-engagement triage was not a software category until large language models got good enough to do it. The major players who could have built this -- Adobe, GitHub, Atlassian -- are all positioned on the wrong side of the engagement. They sell tools that run after you have the codebase. Ghost runs before. We are a small team built specifically for the moment the agency hits the scoping call. The market gap is the product.
// pick the tier that fits
Plans built for how agencies actually buy.
Start with Ghost Open™, free for one person. When more than one architect is running engagements, or client code has to stay in-house, move to Ghost Architect™ Local: your hardware, your model, one monthly licence.
// ghost open™ for one person, ghost architect™ local for the firm
free for one person
Solo engineer or consultant. Every mode, unlimited, on your own Anthropic API key. No license key, no trial clock, no quotas.
Install on npm →5 seats, monthly licence, no per-token bill
A hub in the office, spokes on five engineers' machines, one queue. Client code stays inside your network.
15 seats, monthly licence, no per-token bill
Every architect in the firm on one licence. Zero-egress mode for the engagements that require it, a signed compliance ledger for the ones that ask.
Unlimited seats, monthly licence
Multi-brand portfolio, air-gapped if that is how you work. Convert during the evaluation trial and the first month is half price.
Ghost Open™ is free for one person. Local is one monthly licence. Full pricing breakdown →
Your next bid goes out this week.
It can be a guess, or it can be a report.
Install Ghost Open™ free on npm and run a scan against any prospect codebase in the next ten minutes. If the deliverable looks like something your whole firm should be running on every engagement, Ghost Architect™ Local is the next step.
Ghost Brief™, the AI remediation pipeline that turns your findings into fixes, is included free in Ghost Open™.