Solution
Spectr for software consultancies
A consultancy writes specs for a system it did not build. That is the whole difficulty. The naming, the architecture, the half-finished migration and the reason for it are all things the client knows and you are piecing together. Spectr reads the client's actual repos, docs and designs before it writes, so a spec describes the system that exists instead of a sensible guess at it.
You are estimating a system you have not finished reading
An estimate written without the codebase open is a guess with a number attached, and it is the guess the contract gets written against. The information that would fix it exists - in the repos, in a Notion page, in a decision recorded eight months before you arrived. But nobody has all of it open while writing ticket fourteen, and reading it properly for every ticket is time the engagement does not have.
What it actually looks like
A six-month engagement modernising a client's billing system. Three of your engineers, a codebase none of you wrote.
Week 1
You pin the client's three main repos, their Notion architecture pages and the Figma file to the project in Spectr.
Week 1
Discovery calls get imported. The specs that come back reference their actual service names and the two endpoints that already do half of what was asked for.
Week 2
One estimate comes back higher than the room expected. It is right - the change touches a queue nobody had mentioned, which Spectr found in the repo.
Week 8
A new engineer joins the engagement. Instead of a week of archaeology, they read the project memory Spectr has been building: architecture, glossary, open questions, decisions.
Month 4
A scope disagreement comes up. The tagged requirement version from month one settles it in one open-and-point.
Month 6
Handover includes a written record of what was decided and when, because it was written as you went rather than in the last week.
A week of codebase archaeology for every new joiner has turned into a couple of days - and one wrong estimate got corrected before it was contracted rather than after.
Specs written against the real repos
Pin up to five GitHub repositories per project. Spectr searches for the code a discussion actually concerns and reads those files, rather than skimming the tree. Open issues and recent pull requests come in too, so a story can point at work already in flight instead of duplicating it. That is what makes an estimate reflect what the change would really touch.
Project knowledge that builds up as you go
Spectr keeps a per-project knowledge base - what the product is, how it is built, the glossary, the open questions, who does what - and updates it after each meeting. On a long engagement that is where the stuff you are piecing together accumulates, and it feeds every later run. If it gets messy, one click rebuilds it from the whole project.
A record that survives handover
Versioned requirement documents with a tagged version at each sign-off, plus a recorded decision on every out-of-scope flag. When you hand the system back and someone asks why a thing was built that way, there is an answer written down at the time rather than reconstructed afterwards.
Budget per engagement, read-only
Connect Harvest per project for budget against spend. The projection is based on stories actually completed in Jira, Linear or Shortcut, not on a percentage typed into a status report.
Works with
Frequently asked questions
Does Spectr write to our client's repositories?
No. GitHub access is read-only - no pushes, no pull requests, no comments. The only thing Spectr ever creates is a ticket in your tracker, after someone approves it.
Does it work with private repositories?
Yes, for any repo the GitHub account you connect can already see.
Can one workspace hold several client engagements?
Yes, as separate projects with their own integrations, pinned docs and knowledge base.
What if the client will not let us connect their GitHub?
Everything still works - specs, stories, requirements, scope tracking. You lose the code-grounded estimates, which is the part worth pushing for.
Also worth reading
Digital agencies
Protect the margin on a fixed fee: a scope record, a live budget, and tickets written once
In-house product teams
Requirements that stay current, stories grounded in your own system, evidence-led priorities
Startups
A real backlog without a full-time PM, and numbers instead of the loudest opinion
See it on your own project
Free plan available - five spec sheets a month, two seats, no card required. 30-day free trial on Starter.
Free plan available. 30-day free trial on Starter.