Readigraph
A consultancy-resellable evidence workspace separating official repository metrics, definitions, cost assumptions, baselines, associations, reviewer interpretation and approved recommendations.
Engineering leaders and their advisers want to understand how coding assistants are used across repositories, but tool activity is easy to overinterpret. The supplied research confirms a repository-level usage interface made generally available on July 17, 2026 and reports that it exposes agent-created and merged pull-request activity plus review data. It also confirms established engineering-intelligence competitors, neither described as assistant-adoption-specific or packaged for consultant resale. Current interface fields, permissions and retention must be validated from official documentation.
Readigraph would preserve consultancy, client, organization, repository, team assertion, reporting window, source interface version, metric definition, raw aggregate, missing interval, pull request, agent-attribution assertion, creation event, merge event, review event, comment count, active-user aggregate, seat assignment aggregate, cost assumption, baseline period, comparison period, confounder, calculation, association, confidence note, analyst interpretation, client correction, recommendation, approval, report version, export, retention and deletion as distinct records.
A pull request attributed to a coding agent does not establish that the assistant caused delivery, quality or business value. Merge counts omit difficulty and rejected work; acceptance and active-use metrics do not prove learning, readiness or productivity. Cost assumptions are not invoices, and associations before and after adoption are not causal ROI. Readigraph must not rank individual developers, infer performance from repository activity, disclose person-level telemetry, recommend employment action or claim savings without an approved method and source evidence.
The pilot should use synthetic repositories and permissioned aggregate metrics from a small client portfolio. The likely buyer is an engineering-productivity or development-operations consultancy that wants a repeatable client assessment, with an engineering or technology leader as the report consumer. Interface entitlements, client consent, aggregation thresholds, metric stability, consultancy workflow, attribution method, cost data, budget and incumbent response remain unverified.
An engineering-productivity or development-operations consultancy reselling a repeatable coding-assistant adoption assessment to client technology leaders.
The supplied research confirms general availability of repository-level metrics on July 17, 2026.
Engineering consultancies and client technology leaders are identifiable, while assessment budgets and resale demand need validation.
The new repository-level interface enables a fresher package, but adoption analytics and consultancy reporting existed before it.
The input confirms a newly available repository-level interface, identifies a consultant-as-buyer channel and gives a concrete assessment workflow around official aggregates.
Established engineering-intelligence products compete nearby, the interface provider can ship its own views, causal ROI is difficult and no structural copying cost is evidenced.
Discussion
No comments yet — be the first to weigh in.