| title | Settings and preferences |
|---|---|
| summary | Your config.toml, which every command reads, your learner profile, the AI section, the web UI's preferences, and how to back up, restore or reset your progress. |
| section | Settings and help |
| order | 120 |
cstudy has four kinds of settings:
| Kind | Where | Covers |
|---|---|---|
| Your configuration | config.toml, one file for all your workspaces |
The default language, the job roles and domains of your learner profile, time limits, the review schedule and more. Every command reads it |
| The workspace's languages | .cs/workspace.toml, in each workspace |
The languages you practice there: its folders, the toolchains cstudy doctor checks, its IDE files, and the languages of your learner profile (choose its languages) |
| Preferences | .cs/serve/preferences.json, in each workspace |
How the web UI looks and behaves: Vim keys, the theme, run on save, code intelligence |
| AI keys | credentials.toml, beside config.toml, never in a workspace |
Your own provider keys for the optional AI preview |
The Settings view of the web UI (Alt+7) edits all four. In the terminal, cstudy config finds and opens the file, cstudy lang changes the languages, and cstudy ai key saves a key.
The file is ~/.config/coding-study/config.toml, or $XDG_CONFIG_HOME/coding-study/config.toml when that variable is set. CSTUDY_CONFIG points cstudy at another file. Every key is optional: a missing file means every default.
cstudy config path # print where the file is
cstudy config edit # open it in $VISUAL or $EDITORcstudy config edit makes the file from a commented template when you have none: every key at its default, with what it does, so it reads as the defaults until you uncomment a line. It never changes a file you already have.
The keys you are most likely to change:
| Key | Default | Does |
|---|---|---|
default_lang |
"python" |
The language of cstudy start, cstudy next and cstudy mock when you give no --lang: python, typescript, java, kotlin, go or rust. When it is not one of the workspace's languages, the first of them is used |
open_on_start |
true |
Open your editor on cstudy start |
progress_db |
"progress.db" |
Where your progress lives, relative to the workspace |
next.tracks |
["dsa"] |
The tracks cstudy next and cstudy mock pick from |
profile.job_roles |
every role | The job roles you prepare for (your learner profile) |
profile.domains |
every domain | The domains you study |
time_limit.enabled |
false |
Turn time limits on for every run, as cstudy test --time-limit does for one |
bench.reps |
7 |
Timed repetitions per size in cstudy bench |
schedule.max_interval_days |
180 |
No review is scheduled further away than this |
schedule.mastered_interval_days |
21 |
A problem counts as mastered from this review interval on |
The [schedule] table also holds the first intervals of each rating (ok_first_interval_days, easy_first_interval_days and the like). A new interval applies to reviews scheduled from then on: dates already set do not move. The Settings view lists every key, with its name in the file, its value, its default and when a change applies, and configuration keys lists every key with its default. The [ai] table holds the six keys of the AI preview, which is off by default (turn it on).
A change applies at the next command. The web UI reads the file for each request, so it needs no restart, except after a change to progress_db: restart cstudy serve then.
Changing progress_db moves nothing: your data stays in the old file, and cstudy uses the file at the new path, or starts an empty one there. Settings asks first and says whether a database exists at the new path, and cstudy config edit says so once you close your editor.
Study guides are notes of your own; cstudy ships none. The rubric problems, those of the system-design and code-review tracks, can name sections of them in references.study_guides, as paths relative to one folder of yours. Set that folder as root in the [study_guides] table:
[study_guides]
root = "~/study-guides" # an absolute path, or one that starts with ~/It is unset by default, and then no study guide is linked. With it set, cstudy start and cstudy show end a rubric problem's statement with a ## Study guides section, a link to each referenced file that exists under the root, and cstudy start prints a guide line for each:
guide file:///home/me/study-guides/distributed-systems-study-guide.md#databases
A referenced file that is not under the root is left out, and nothing fails. The bundled rubric problems name sections of distributed-systems-study-guide.md, so a guide of yours saved at that path under the root is linked. In the Settings view the key is Study guides folder.
Your learner profile is the default filter of the lists you choose from: the languages you practice, the job roles you prepare for and the domains you study. cstudy list, cstudy next, cstudy mock and cstudy learn apply it, and so do the Problems, Guides and Mock views of the web UI. Each part defaults to every value, so a profile you never set filters nothing.
| Part | Values | Stored in | Change it with |
|---|---|---|---|
| Languages | python, typescript, java, kotlin, go, rust |
The workspace's marker, .cs/workspace.toml |
cstudy init --lang, cstudy lang add and cstudy lang remove (choose its languages), or Settings |
| Job roles | The roles of the labels, except general |
profile.job_roles in config.toml |
cstudy config edit, or Settings |
| Domains | The domains of the labels | profile.domains in config.toml |
cstudy config edit, or Settings |
[profile]
job_roles = ["backend", "platform"] # leave the key out for every role
domains = ["backend", "security"] # leave the key out for every domainEach key is a list of one or more values, each listed once. Leaving a key out means every value, including the values a later cstudy adds; a list that names every value today does not take the new ones. An empty list, an unknown value, a value listed twice and general are refused, by the key's name, as any wrong key is. general marks what any engineer uses, and those problems and guides show whatever roles you list. Keep algorithms among your domains for the data structures and algorithms problems.
A change applies at the next command. A page that is open reads it after a reload, or at once when you save it in its own Settings.
| Part | An entry matches when |
|---|---|
| Languages | It has a reference in one of your languages. A system design or code review problem, which has no language, always matches |
| Job roles | It lists one of your roles, or general. A problem of your own without labels matches |
| Domains | It lists one of your domains, or none |
An entry shows when it matches every part the command applies. A guide matches by its own domains and roles, and a path when one of its steps does; cstudy learn applies the roles and domains only, since a guide teaches in every language.
The profile does not apply to your sets, to cstudy stats, cstudy history and the reviews of the Progress view, to cstudy verify, to a command given a slug, a guide or a path, or to these docs: what you chose and what you did stay visible. Outside a workspace, cstudy list and cstudy learn use all six languages, and your roles and domains still apply.
cstudy list --widen lang # every language, this once
cstudy list --widen all # the whole catalog
cstudy next --widen role # a pick from every role
cstudy learn --domain security # this domain instead of yours--widen drops the profile in one part for one run, and repeats; it never changes the profile. cstudy list takes lang, role, domain or all. cstudy next, cstudy mock and cstudy learn take role, domain or all, since a pick starts only in one of the workspace's languages, and a guide has no language. An explicit --role or --domain replaces the profile in its part and may name a value outside it; --role backend then keeps the entries that list backend, without general. The --lang of cstudy list, cstudy next and cstudy mock replaces your languages the same way.
While the profile hides something, the command ends with one dim line that says how much shows, what narrowed it and how to see the rest:
filtered by your profile: 82 of 400 problems · languages python, go · roles backend and general · domains security, backend · --widen all shows all 400
cstudy next and cstudy mock count the problems the pick came from (picked from 52 of 125 problems), and cstudy learn counts guides and paths. A profile that hides nothing prints no line. When the profile leaves nothing to pick, cstudy next and cstudy mock say so and name --widen all. --json prints no line: its document holds profile, with the values that narrowed the list and hidden, the number of entries the profile left out.
In the web UI, the filters of each view show the profile and widen it (filters and your profile).
Preferences are stored in .cs/serve/preferences.json in the workspace, so they follow the workspace, not the browser.
| Preference | Default | Effect |
|---|---|---|
| Vim keybindings | off | Vim keys in every editor; :w saves |
| Run on save | off | Each save, from the page or from your IDE, runs the tests again, as cstudy test --watch does |
| Type-check TypeScript | on | TypeScript runs type-check, so type errors show in the editor |
| Theme | system | Light or dark, or follow your system |
| Code intelligence | off | Live errors, hover documentation and go to definition from a language server on your computer; needs cstudy doctor --fix once (details) |
| Autocomplete | off | Completion and signature help, while code intelligence is also on. Off, the editor offers no completion at all |
Nothing in Settings applies until you choose Save in the bar at the top, which stays on screen as you scroll: it shows how many changes you have, and Reset undoes all of them. The theme previews as you pick it. Leaving with unsaved changes asks first. The vim and theme switches in the status bar still apply at once.
The profile section heads the view (below), then the AI section (the AI section). Beside the preferences, or below them in a narrow window, the view edits config.toml. Each key shows a friendly label with its name in the file beside it, so a search of the page finds either. Save writes only what you changed and keeps your comments; empty a field to return it to its default. A value cstudy would refuse is refused here too, by the key's name, and nothing is written to the file. Save covers every place: it writes config.toml, then credentials.toml, then the workspace's languages, then the preferences, and if one fails it says which and keeps those changes unsaved, while the others stay saved.
The profile section edits your learner profile, with a checkbox for each language, job role and domain. The languages are this workspace's, stored in .cs/workspace.toml; the roles and domains are in config.toml, for all your workspaces. A workspace that never chose its languages shows all six checked, marked all six (none chosen). The last checked language cannot be unchecked, since a workspace needs at least one, and unchecking every role or every domain is refused. Under the languages, the section names the cstudy lang command that Save runs.
After Save adds a language, the status line says that Fix in the Health view installs its toolchain files, as cstudy doctor --fix does. The Problems, Guides and Mock views use the saved profile from then on.
A Save that removes a language first asks what becomes of the language's folder, before it writes anything:
| Choice | Does |
|---|---|
| Keep the folder | Removes the language and leaves its folder and your practice files on disk, as cstudy lang remove rust does. It is the default |
| Delete the folder… | Shows what goes, each folder with its number of practice files, any shared folder that stays and why, and what always stays, then asks again: Delete rust/ deletes it, as cstudy lang remove rust --delete-folder does, and Back returns to the first question. It can't be undone |
| Cancel | Saves nothing, and keeps your changes for another try. Escape does the same |
While an attempt in that language is in progress, Delete rust/ is disabled, with the reason beside it; keeping the folder still works. Progress, .archive/, projects/ and notes/ always stay (choose its languages).
The AI (preview) section sets up the optional AI preview. It shows whether AI is on or off, since it is where you turn it on, and each group names where it is stored:
| Group | Stored in | Holds |
|---|---|---|
| Turn on | config.toml |
The ai.enabled checkbox |
| Keys | credentials.toml |
A row per provider, key set or no key, with a field for a new key and Remove. A key you type is saved by Save and never shown again |
| Models | config.toml |
The routing, fixed or auto, then ai.model, or ai.fast and ai.strong, and the daily token cap. Load models beside a provider with a key lists its model ids as suggestions, as cstudy ai models does; Check runs cstudy ai check, after the consent dialog when it needs one |
| This workspace | progress.db |
Your consents, each with Revoke, then today's usage and the last 7 days |
Load models and Check use what is saved, so they are disabled, with the reason beside them, while AI is off or a change is staged. Revoke acts at once, without Save: it only makes cstudy ask again next time.
Your progress lives in progress.db, which the workspace's .gitignore leaves out. Back it up as a JSON file you can commit:
cstudy export # writes progress-export.json at the workspace root
cstudy export --out ~/backup.json
cstudy import # restores progress-export.json into an empty progress.db
cstudy import ~/backup.jsoncstudy import refuses a database that already holds progress, so it never mixes two histories. The Progress view of the web UI (Alt+5) has export and import too. An export keeps your AI reviews, and leaves out your AI requests and consents, which belong to this machine.
cstudy reset in the terminal, or Reset progress…, the red button in the danger section of Settings, deletes your progress in this workspace: attempts, scheduled reviews, bench runs, AI reviews, AI requests and consents, run folders and rendered statements. You see what goes, with counts, before anything does. It can't be undone, so you type reset to confirm. It refuses while a run or a job is active. Run cstudy export first if you may want your progress back.
cstudy reset # progress only
cstudy reset --files # also your practice files, project folders and notes
cstudy reset --files --archive # and .archive/, the files of earlier attemptsYour practice files, project folders and notes stay unless you tick that box, or pass --files. .archive/ stays unless you tick its own, or pass --archive. Your settings, your AI keys, toolchains, language servers, your own problems and your sets always stay. A reset ends an open attempt along with the rest: the menu of attempts in progress empties at once, and a tab that was showing that attempt says it ended and turns read-only. --yes skips the question, for a script with no one to type.