Decision 0002, 28 October 2026. Selected for discovery only. No FileGuide application release is claimed.
FileGuide is a working-name concept for helping people understand the files on their own computer without risking accidental changes. The initial focus is on Linux desktop users who want clear explanations of storage use and exact duplicates.
The problem we want to understand
There are already capable open-source file tools, including Czkawka and dupeGuru. The question is not whether Comutability can write another scanner. It is whether there is an unmet usability need among people who find technical file-management tools confusing, stressful or inaccessible.
What a first version might do
The proposed journey is intentionally limited:
- Choose the folders you wish to inspect.
- See plain-language totals for space used and file counts.
- Run a cancellable, exact-duplicate scan.
- Review paths and sizes to understand why two files match.
- Export a local CSV report or leave without changing anything.
A possible implementation would group candidates by size before calculating streaming SHA-256 hashes. It would need careful treatment of symbolic links, files changing during scans, unreadable folders and the ability to cancel without locking the interface.
A deliberate safety decision: read-only first
The proposed v0.1 would not delete, rename, move, overwrite or automatically clean files. It would not quietly scan the whole computer, upload personal files or collect analytics. Similar-photo matching, cloud accounts, automatic repair and AI deletion advice are outside the initial scope. This restraint is not a missing feature: it is a way to keep the first project useful, understandable and safe to test.
Accessibility and privacy
Every stage should be keyboard-accessible and usable with a screen reader, with meaningful progress and error messages. The analysis should happen locally, without mandatory accounts or cloud services. We need to test these claims with real people rather than assuming an interface is accessible because it looks tidy.
The research gate before coding
The brief calls for interviews with 10 prospective users and observation of at least 5 people attempting relevant tasks with existing tools. The community should compare building something new with contributing a better interface, guidance or accessibility improvements to an existing project. If the research does not reveal a meaningful difference, narrowing, pivoting or helping upstream would be a valid outcome.
Decisions still open
There is no approved implementation repository, final product name, licence, architecture or release schedule. Python and PySide6 are possibilities to investigate, not settled commitments. A go/no-go decision comes only after discovery.
