How to use Devin to reduce Rust build dependencies
Devin scanned Dioxus dependencies and Cognition’s websites, exposing unnecessary build features and metadata defects with qualified local and shipped outcomes.
devin.ai
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Investigate this Rust workspace’s clean debug build. Measure a baseline, trace expensive default features and dependencies, and identify scaffolding-only or platform-only code compiled unnecessarily. Propose narrowly gated changes with behavior checks and identical benchmark conditions; wait for human review before removing public functionality.
Reach people exploring AI agents
See placements →What happened
Cognition reports local Dioxus debug builds falling from 58.6 to 21.0 seconds and shipped website fixes; no upstream Dioxus merge or universal benchmark is established.
Reported by Cognition — Devin blogHow the agent did it
- Devin investigated Dioxus dependency and build configuration against a compilation-speed goal.
- The team applied local changes that narrowed heavyweight scaffolding, browser and Tokio dependencies.
- A separate website scan produced concrete metadata and redirect fixes, with human selection and reported follow-up crawls.
Where a person stepped in
The published workflow requires scope confirmation and evidence review before selected fixes; Cognition applied the compilation changes locally and reports its own results.
Cognition published two executed Code Scans investigations, including an attempt to speed up Dioxus compilation. Devin examined dependencies and build configuration rather than applying an unexplained compiler tweak.
Concrete findings included cargo-generate being compiled even when the scaffolding command was unused, browser bindings reaching non-web renderers through default features, and a server-rendering crate enabling all Tokio features while using only filesystem and I/O utilities. Cognition applied changes locally and reports clean debug builds dropping from 58.6 to 21.0 seconds across 22 workspace crates. The account does not establish an upstream merge or describe enough benchmark conditions to treat that percentage as universal.
The same publication describes a separate scan of Cognition’s own websites. It found relative social-image URLs, redirect chains and canonical metadata inconsistencies; the team says it shipped fixes and compared later Ahrefs crawls. These are merged into one source account rather than counted as separate cases.
The published workflow plans selection rules, divides matching code into batches, investigates those batches and combines deduplicated findings. Humans confirm scope, inspect evidence and choose changes to pursue. The useful result is specific dependency and metadata findings with qualified local outcomes, rather than a promise that every repository will improve similarly.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Devin blog · Cognition
Frequently asked questions
What dependency issue did the scan find?
Cargo scaffolding and broad runtime features were compiled in paths that did not require them.
Were the Rust fixes shown as upstream merges?
No. The publication describes locally applied changes and its own build measurements.
How are broad findings consolidated?
The workflow divides code into batches, investigates them and reduces duplicate findings into a prioritized report.
More details
About this account
Devin investigated Dioxus dependency and build configuration against a compilation-speed goal. The team applied local changes that narrowed heavyweight scaffolding, browser and Tokio dependencies.
Primary vendor account opened. Concrete executed investigations are retained; generic product claims and customer time-saving estimates are excluded. Compilation measurements are local and vendor-reported; two tasks from this URL remain one entry. Suggested prompt written by Agents In Use; no original task prompt was found in the verified text.
Credit: Cognition — Devin blog
Source published 16 Sept 2026
Source checked 14 Oct 2026
Tools and topics
Devin Code Scans · Cargo · Dioxus · Ahrefs
compile performance · dependency scope · repository scan
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.