GlideLens

Chrome & Edge · for ServiceNow developers

Press \
anywhere in ServiceNow.

A command palette over the platform you already work in. Find a record, check an update set before you promote it, search the script the instance's own index can't see, read what actually runs on a catalog item, and replay a form interaction after it has happened.

GLIDELENS
    ↑ ↓ move ↵ jump to it this palette works — try it

    Same commands as the real thing. In ServiceNow they run; here they scroll.

    Forms

    Reads
    sys_documentation
    sys_translated
    sys_choice
    sys_ui_message
    Resolves
    inherited fields

    Which translations are missing, and where they would go

    A read-only report over every field, catalog variable, choice and getMessage key on the surface in front of you, against every active language. A gap links to the record that would hold the translation, prefilled with the exact keys — so closing it is a save, not a search. It runs on a classic form, a catalog item's definition form and a Service Portal catalog item; on a Workspace record the same command offers a link to the record's classic form, where it runs.

    The store depends on the dictionary type, and getting that wrong is silent: translated_field values live in sys_translated keyed by the source string, not in sys_translated_text. Labels resolve through super_class, so a field inherited from task lands on the right dictionary record. Absent data is never counted as coverage: a denied read stays Unavailable and stays out of the denominator.

    per-language chips prefilled gaps coverage per section read-only

    Interactive demo — fictional catalog item. Two rows are open to show the per-language chips. Deselect a language to rescore, or reveal the Minor rows.

    Translation Lens

    Catalog definition form · sc_cat_item · Base language en

    0% 0complete 0partial 0missing / none 0 of 0 languages shown

    Reports where a translation is missing. It never writes one.

    Forms

    Exports
    one JSON file
    Fills
    the page's own boxes
    Publishes
    never — you do

    Translate a catalog item with the AI tool you already use

    On a catalog item's Edit Translations page, one command hands every untranslated, unlocked text — its fields and the messages its scripts show — to your own AI tool as a single JSON file that carries its own instructions. Paste the reply back and it fills the page's boxes for you to read over; the page's own Publish button stays yours. Nothing is saved until you press it, and a reload discards every fill.

    Before writing, every row is checked again against the live page: a source text that changed, a field locked since the draft, a translation too long for its column, or one that shares a stored translation with a field outside the draft is left alone and named. A placeholder mismatch waits for Fill anyway; a box you typed into since the draft waits for Overwrite, shown beside both values. GlideLens never contacts an AI service — the file leaves your browser only when you upload it.

    bring your own model checked before it writes old text kept ⚠ shared translations named

    Demo — fictional catalog item, English to French. Switch between the two steps of the panel.

    Translation Assistant

    Hands this item's untranslated fields to your own AI tool · English → French

    • 12to translate
    • 3already translatedunlock one on this page to redo it
    • 15fields on this item
    Download JSON

    Upload the file to your AI tool and say: “Follow the instructions in this file.”

    The file holds this item's field text. It leaves your browser when you upload it somewhere.

    Copy prompt + JSON instead

    Before you publish4 of these translations are shared: publishing them changes those translations for every catalog item on this instance whose fields use the same source text. 2 of these translations are script messages: publishing them changes those messages for every script on this instance that uses the same message key.

    Never saves or publishes — Translation Assistant leaves Publish to you.

    Catalog

    Covers
    Stored vs live
    record movement
    secret-safe reads

    Variable Values, including the ones you cannot see

    On a classic RITM or record-producer target, compare the raw value stored in the database with the raw value now held by g_form, so an onLoad change is visible without mistaking reference or choice labels for differences. On a verified Workspace record — a Service Operations Workspace request item, or a supplier case or supplier task in Source-to-Pay Workspace — the same command checks a narrower evidence-backed set after confirming the visible record, the Variable Editor, read permission, and each raw value's shape. Each supported surface has its own verified type list rather than inheriting another's. Partial and stored-only results say plainly what was not compared; every other Workspace experience and table remains unsupported. On an ordering form, the same command keeps its current best-effort live-value and hidden-variable view.

    hidden / visible filter differing filter Workspace guarded comparison read-only

    Interactive demo — fictional data, native RITM panel. Switch to Differing, or try Non-empty.

    Variable Values Best effort

    Stored Table API values compared with raw live g_form values on this RITM. Values are read-only; reference and choice labels are not compared.

    0variables 0hidden 0visible 0differing

    Read-only inspector — does not modify the live form.

    Debugging

    Captures
    g_form
    GlideAjax
    field events, errors

    Debug Timeline

    Record one page's interaction and read back what happened: public g_form calls, native field events, GlideAjax timing, and JavaScript errors on a filterable timeline. Best-effort by design — it shows you what ran, not which script ran it, and says so rather than guessing.

    GlideAjax rows carry what you would otherwise open the Network tab for: the Script Include, the method, every parameter added, and the decoded response, next to how long the call took and in order with the rest of the interaction — rather than pieced back together from a list of XHRs. Parameter and response names that read like secrets are redacted before anything is shown.

    single page params · response reversible patches secrets redacted

    Interactive demo — fictional trace, real panel. Open a GlideAjax row for its params and response.

    Debug Timeline Best effort

    ServiceNow form activity correlated across the frames that were available during this interaction.

    0events 2frames 2.4sduration

    Stacks identify runtime calls, not guaranteed Client Script or UI Policy names.

    Catalog

    Reads
    catalog_script_client
    catalog_ui_policy
    catalog_ui_policy_action
    Where
    Service Portal

    Catalog Logic: what actually affects this item

    Open a catalog item and get every client script and UI policy bound to it or its variable sets, grouped by the variable they watch. Rows that cannot fire while ordering — inactive, or scoped to RITM and Task views only — are flagged, so "why isn't my script running" answers itself before you open a single record.

    group by variable ⚠ won't run here open in platform read-only

    Interactive demo — fictional data, real panel. Try Group by variable, Active only, or search laptop_model.

    Catalog Logic

    Catalog client scripts and catalog UI policies bound to this item or its variable sets. Read-only — nothing here runs or edits the logic.

    0shown 0active 0inactive 0won't run here

    Rows flagged ⚠ are active but scoped away from the catalog form — the usual answer to "why isn't my script running".

    Update sets

    Reads
    sys_update_set
    sys_update_xml
    sys_update_version
    Compares
    another open instance

    Update Set Lens: check the set before you promote it

    Open the palette on an update set's form and every customer update in it is checked against the version this instance runs. Stale means that version came later — from another set, an upgrade or an app install — and the row names it, with who saved it and when. The rest are Can't tell, with the reason, Same content, In this batch or Current. Stale is claimed only when this set's own version was loaded here and later replaced, so an update skipped at commit is never reported as replaced by an older one. A batch parent checks its child sets, nested ones included.

    Check on asks another instance open in your browser, as that tab's own session, whether the set was retrieved there and how far it got, then record by record whether that instance runs the same version — and where it doesn't, when and in which update set that version was saved. Each row links to its update, its record and its version history, and Other sets holding it opens Version Lens on that record: every set that captured it, and whether two open sets hold different versions.

    stale updates batch children check on another instance Version Lens read-only on both
    Update Set Lens on a completed update set of eight updates: three marked Stale, two changed and one deleted later in an in-progress hotfix set, each naming who saved it and when; the rest Current.

    Session

    Finds
    by identity, role or group
    Changes
    your own session

    Impersonate: start from the role, not from a name you already know

    Other tools answer "become this named person". The question a debugging session actually starts from is "who holds catalog_admin, so I can test this ACL?" — so this searches by role as well as by name, user ID, email, title or exact sys_id, and tells you whether each holder has the role directly or through inheritance. Role membership is read from the effective membership table, which includes grants that arrive through a group or through role containment; the roles field on the user record is direct-only, incomplete even at that, and is never used.

    Inactive, locked-out and web-service-only accounts are never listed, because impersonating one can end your own session. You can narrow to the members of a group, alone or together with a role. You can also filter by a field your own instance offers — the list is read from the live dictionary and labelled the way your instance labels it, so a custom country field appears where a hardcoded one would have pointed at the wrong column. Nothing happens on a click: a result opens a confirmation that repeats the identity in full. One confirmation sends at most one request, nothing is ever retried, and ServiceNow records the impersonation in its own audit log. Stop impersonating returns you to the account you started from. The people you impersonated recently are listed when the panel opens. That is ServiceNow's own list, checked like any result, and GlideLens keeps no copy of it.

    role search group members direct vs inherited eligibility guard explicit confirmation never retried

    Interactive demo — invented people, roles and groups, real panel. It opens on the recent list. Pick catalog_admin to see who holds it, pick a group to see who is in it, or type morgan. Pressing Impersonate opens the confirmation with the person's roles, and Start impersonation stops short of sending anything.

    Impersonate

    Find a user by identity, attribute, role or group. Only eligible accounts are listed, and nothing changes until you confirm.

    Type a name, choose a role or group, or pick an attribute to begin.
    GlideLens stores no search term, result, role, group or impersonation history. The recent list is ServiceNow's own, and ServiceNow records impersonation in its audit log.

    Catalog

    Reads
    item_option_new
    catalog_ui_policy_action
    Shortcut
    Alt + double-click

    Per-variable insight icons

    An icon beside every variable on the form. Click one and the panel narrows to that variable alone: the onChange scripts watching it, and what each UI policy does to it — hides, mandatory, read-only, sets value. Attribution is matched on sys_id, not guessed from script text.

    per-variable scope show all to clear

    Install

    One click, then backslash.

    It is on the Chrome Web Store and Edge Add-ons, so your browser keeps it current by itself. Same package either way — nothing about it is browser-specific.

    Google Chrome

    Chrome Web Store

    Add to Chrome

    Microsoft Edge

    Microsoft Edge Add-ons

    Add to Edge

    Chromium browsers, version 111 or newer — from March 2023. The panels colour themselves with color-mix(), so the manifest declares that floor and the browser declines to install rather than leaving you to discover it. There is no Firefox or Safari build.

    1. Pin it

      Click the puzzle-piece icon in the toolbar and pin GlideLens, so the popup is one click from anywhere.

    2. Press \

      Open any ServiceNow tab and hit backslash. Everything else starts from that palette.

    Or run it from source. On the repository, click Code then Download ZIP, extract it somewhere permanent — your browser re-reads that folder on every start, so not Downloads — and load it at chrome://extensions with Developer mode on and Load unpacked. Point it at the folder that directly holds manifest.json; GitHub wraps everything in an inner folder. An unpacked copy will not update itself: re-download over the same folder, hit reload on the card, and refresh your ServiceNow tab.

    Nothing leaves your browser. The extension only activates on *.service-now.com, reads through your own session, and has no server of its own.

    What this isn't

    Deliberately narrow.

    It is not a general ServiceNow toolbelt, and it does not try to become one. Update set switching and environment badges are well served by tools that already exist, so they stay out of scope. This aims squarely at the gaps: catalog and Service Portal debugging, dictionary inheritance, form introspection, and the script no index reaches.

    Impersonation used to be on that list. It came off because the existing tools answer "become this named person", and the question a debugging session actually starts from is "who holds this role, so I can test this ACL?" — which none of them answer. So Impersonate searches by role as well as by identity, and role elevation of your own session stays out.

    Everything it reads goes through the Table API on your own session, with your own permissions — there is no server, no account and no telemetry. It never writes to your records and never submits a form. Three things change anything at all: Prefill, which fills unsaved fields on the catalog form in front of you for you to review; Translation Assistant, which fills unsaved translations on the comparison page in front of you and leaves Publish to you; and Impersonate, which is the only one that changes something on the instance rather than in a page — your own session, after an explicit confirmation, and ServiceNow records it in its own audit log. Code search shows the matching lines and one line either side, and redacts hits whose field names read like secrets. The privacy policy spells all of this out.