Zoost - workbench for Zoho CRM

Mirror Zoho CRM functions - Deluge, Java, Python and Node - to local files for Git, with search, schema and Deluge call graphs.

Zoost turns your Zoho CRM org into a local, version-controllable, searchable codebase - and then draws you a map of it. If you administer or develop on Zoho CRM you know the gaps: no external editor, no Git, no way to search across all your functions at once, no way to see what calls what before you change something, and no quick answer to "what is the API name of that related list?", and no way to tell which functions still use a given connection. Zoost fills them, with no server and no account: everything it reads lands on your own disk. The single exception is the optional AI assistant, which stays off until you enter your own API key - described in full below. WHAT IT DOES - Judge it before you give it access to anything. "+ Sample" writes a workspace of invented data into your working folder - a couple of hundred Deluge functions with real call chains, modules with lookups, workflows, schedules and connections - so you can open the tree, the graph, the audit and the exports without a Zoho tab and without an account. It is generated, never fetched, everything that would talk to Zoho is disabled for it, and it is deleted like any other workspace. - Local mirror, and Git if you want it. Zoho CRM's own version history covers a function, one at a time. Everything else the pull captures - module schema, layouts, related lists, workflows, schedules, connections, and what Zoho reports as failing - arrives on your disk as plain files, so with Git it gets a history too, and one diff answers what changed across every kind at once. Deluge functions are plain .dg files; Java, Python and Node functions keep every file in a project folder, with a .meta.json sidecar for each function. Git is optional: without it the mirror is still ordinary files you can search and hand over. Functions and project files removed in Zoho are pruned locally on the next pull, so the folder stays a faithful mirror rather than an accumulating pile. - Search across every function at once. Full-text search over all mirrored Deluge, Java, Python and Node files, literal or regular expression - the closest thing to grep for Zoho CRM, and something the platform does not offer. Find every reference to a field, a module, an endpoint or a hardcoded id before you change it. Named patterns - an email address, Zoho's 18-digit record id, and any you add - are one click away and yours to manage. - Auto-sync on save. Save a function in the native Zoho editor and the matching local file updates by itself, so your working copy always mirrors production. - Reference graph. For any function: which functions call it (the impact if you change it) and which it calls (its dependencies) - as a searchable explorer and as a visual node-link diagram. Calls to custom functions are clickable in the code preview: jump to the definition and back. - Reverse usage. Where each function is actually wired across the org - blueprint, button, schedule and so on - read from Zoho's own signal, with no expensive scans. - Module schema, fields and layouts. Every field with type, lookup target, picklist values and mandatory flag. A layout matrix shows, per field, which layouts it belongs to and where it is required - and flags fields that are on no layout at all. - Relations, with the names you actually need. The API name of a related list is not the api name of either module, and it is what zoho.crm.getRelatedRecords() requires. Zoost catalogues every relation with its target module, the lookup or linking module behind it, and the Deluge call ready to copy. - ER diagram that stays readable. Modules as tables, foreign keys as arrows. Focus a module, adjust depth, walk the graph by clicking. A relation-first mode pushes modules into the background and brings relation names forward. Click an arc to isolate one relation and fade everything else. Boxes can be dragged: the automatic layout is a starting point, and what you arrange is what the PDF prints. Live sliders for spacing, spread, label gap and label size, an all-modules scope, and Save PDF for wall-size prints. - Fits the access you actually have. Not every Zoho user may read Deluge or the connections catalogue, and no API says so in advance - so Zoost finds out by asking, once, and records the answer with the date. An area your role is refused is skipped on later pulls, nothing is written for it, and its tab is removed rather than sitting there greyed out; the reason is in Settings, which can also re-check, since roles change. You can hide and reorder the tabs yourself too, and choose which types Pull all should ask for at all. - Every part knows when it was last read. Because a type can be excluded from a pull, the mirror states per area when it last came from Zoho. A report section whose data is behind is unticked in the export dialog, with the date and the reason - you can include it anyway, and the report then says so. Both reports carry the per-area dates whether or not anything is behind. - Automation map. Workflows with triggers, criteria, instant and time-based actions and the functions they invoke; schedules with frequency, status and target function. Rules with actions that run after a delay carry the count and the delay, a filter shows only those, and each rule shows when it last ran. - Connections, cross-referenced. The org's connection catalogue with its connector, status and scopes - and, for each one, how many of your functions use it and exactly which. Filter to the ones no function uses, or the ones configured but not connected. Every function also lists the connections it uses, and who last changed it. - Size and outbound calls. Every function shows its length and how many outbound calls it makes - invokeurl, zoho.crm and the other Zoho service tasks, counted outside comments and string literals - so you can see at a glance where length and API cost concentrate. These are plain counts with no threshold and no verdict: length is verbosity, not complexity, and the interpretation is yours. Computed from the local mirror, with no extra calls to Zoho. - Health audit. Orphan candidates, unresolved and ambiguous calls, broken automations, lookups pointing at modules that are not there. Every check states what it does and does not analyse. Candidates to review - never automatic deletions. - What a rule fires. A workflow almost never does the work itself: it points at an email notification, a field update, a task or a webhook, each of which exists on its own in Zoho and is reused across rules. Zoost lists all four with a kind filter, says how many rules fire each - read from the rules already on your disk - and what each one does: the template and the address a notification goes out as, the field and value an update writes, a task's subject, due date, owner and reminder, a webhook's method and URL. About half of them are attached to nothing in a real org; that is a candidate to review, never a verdict. It does not read the content of an email template, and never who the recipients are. - What is actually failing. Zoho reports which functions have failed, and Zoost lists them beside the rest of the org: the function, what invoked it (a REST call, a workflow, a button, a schedule), the reason with its line number, how often, and how many runs and failures Zoho counted in the last 24 hours. Counts, with no verdict attached. It does not re-run a failed execution - that would make Zoho run code that writes to your records - and it does not read the input of the failed run, which stays in Zoho. - Exports you can hand over. The whole workspace as one self-contained cross-linked HTML file, or as Markdown shaped as context for an external LLM. A dialog decides section by section what the file may contain; source code is opt-in and flagged every time. - Optional AI assistant, bring your own key. Anthropic (Claude) or OpenAI (ChatGPT). With Anthropic it runs as an agent with read-only tools and explores your org itself, showing every tool call; with OpenAI it answers in one pass from the org index and the function you have open. The chat states which engine is active and what it can do. - Your key, your machine, your choice. The API key is stored locally and sent only to the provider you pick. You can protect it with a passphrase you choose - it is then stored encrypted and asked for once per browser session. Off by default, because on a personal machine it buys little and on a shared one a lot, and the trade-off is explained where the choice is made. - Built for multi-org reality. One working folder holds a subfolder per Zoho org, created on demand. Each workspace is bound to its org, host and instance, so a production workspace can never be synced against a sandbox by mistake. SAFE BY DESIGN Zoost reads from the Zoho CRM instance you are already signed in to and writes to a local folder you choose. It has no server of its own: no analytics, no telemetry, no tracking, no remotely hosted code. Two things leave your machine, and only when you make them. The first is a problem report: if something breaks you can press Report this problem, read the whole report in the page, cut anything you want out of it, and send it - it becomes a public issue. Nothing is sent unless you press Send. The second is the optional AI assistant, which is off until you enter your own API key. Once enabled, the content it needs to answer you - including function source code - is sent directly from your browser to the provider you configured (Anthropic or OpenAI) and is processed under their terms. If your organisation restricts sending source code to third-party AI services, leave that feature off; everything else stays local. Zoost calls no endpoint that creates, edits or deletes anything in Zoho CRM. It reads metadata and function source only, through the session you are already signed in to, using the same internal interfaces the Zoho CRM web app uses - those are not documented by Zoho and can change, in which case a feature stops and says so rather than guessing. It does not read, download or export your Zoho CRM records - no contacts, no deals, no customer data. WHO IT'S FOR Zoho CRM administrators, developers and consultants who want proper code management and a way to understand an org: diffs, history, impact analysis, schema documentation, relation lookup and backups. The org does not have to be yours - one working folder holds a subfolder per org, so several clients stay mirrored side by side and each workspace is bound to the org it came from. REQUIREMENTS - Be signed in to Zoho CRM in the same browser. - Choose one local working folder; workspaces are created inside it automatically. DOCUMENTATION AND PRIVACY Source code: https://github.com/ivannot/zoost Guide: https://zoost.it/docs-crm Privacy policy: https://zoost.it/privacy Home: https://zoost.it When something fails, the panel can build a problem report - versions, the error, the last few status lines and counters - open it on a page on zoost.it with the text already in it, where you read it, trim it if you want, and send it - or not. It becomes a public issue only when you press send. It carries no code and no key at all; names, paths and ids are stripped wherever they can be recognised - which is why you are shown the whole text first. Nothing is ever sent automatically. Free and open source, licensed under the Apache License 2.0. Maintained in spare time on a best-effort basis; there is no guaranteed support or response time. NOT AFFILIATED Zoost is an independent, unofficial developer tool. It is not affiliated with, endorsed by, sponsored by or supported by Zoho Corporation. "Zoho", "Zoho CRM" and "Deluge" are trademarks of Zoho Corporation, used here in a nominative sense only, to indicate compatibility.