README file from
GithubServiceNow Manage
2.20.0: local monthly translation budget
See the 2.20.0 patch notes for setup instructions and counting boundaries.
- 번역 연동 settings now show a local monthly character counter, editable limit (default 400,000), and warning threshold (default 80%). This is independent of Google Cloud usage/billing and applies to ServiceNow Manage translation requests only, not other Translate commands, apps or devices.
- Confirm the initial untracked usage once (enter 0 to count from installation onward). Each target-language request reserves its Unicode codepoint count, including whitespace, before dispatch. Cached translations do not count; failed or uncertain requests remain counted conservatively.
- A request exceeding the remaining allowance is blocked before sending and shows a notification. Ticket synchronization and existing translations are preserved. The month resets at 00:00 Asia/Seoul on the first day, when the next usage check runs.
- Usage and limits persist in private
translation-usage.jsonbeside the plugin, without translation text. Do not delete this file or its.initializedmarker to reset usage. Corrupt/unwritable/missing initialized storage blocks translation rather than silently resetting the counter. Separate devices are not coordinated.
2.19.0: Jira editing, incidents and reliable refresh
See the 2.19.0 patch notes for all changes since 2.18.1, connector limitations and update instructions.
Unified Jira issue editor
- Select an issue in 업무 정보 수정 to see current assignee, reporter, available status transitions, Plan Start Date and due date. Search for people and save changed fields together. Drafts remain local to the open window and survive switching issues.
- Edit permission and current values are read from Jira. Unavailable fields remain read-only; transitions requiring extra fields must be completed in Jira. Status and field updates are separate requests, so partial completion and unverified outcomes are reported explicitly, without automatic write retries.
- Saving requires the existing opt-in Jira write permission. Automatic refresh remains read-only. No production Jira writes are used for automated tests.
Reliable scheduled refresh
- Concurrent refreshes of the same ticket wait for the same result instead of counting a busy ticket as a failure. Deferred work is reported separately.
- Actual full-refresh failures are retried on the automation tick after 1, 5 and 15 minutes, up to three retries. Pending retries survive restart; successful refreshes clear the queue. Only existing active tickets are retried while automatic refresh is enabled and Obsidian is running.
- Failed retries stop after the limit. A manual refresh or the next regular scheduled refresh starts a new bounded cycle. ServiceNow work notes show the current failure, retry schedule and the last five failures, including after recovery.
- Failure to refresh Jira after a successful ServiceNow refresh is logged separately, not counted as a failed ServiceNow refresh.
Incident tickets
- Register
INCtickets using the configured Incident table, alongside CR and SR. Incident notes participate in refresh, dashboards, work notes, To-Dos, meeting links and AI Wiki links. - Preserve incident-specific impact, urgency, caller, configuration item, service, hold reason and resolution information. Optional INC columns are available in the dashboard field selector without changing your selected columns.
- Incident notes omit empty CR-only document/schedule fields. Manual fields and note content are preserved. Automatic first-registration CR document search and CR/SR SLA assumptions are not applied to incidents.
- AI prompts use an explicit INC section when available, otherwise an incident-specific fallback. A missing INC analysis template is not replaced by a CR/SR template. Actual fields and access depend on your ServiceNow instance and account.
Jira comment composition
- The reply button explains the current connector limitation and offers the exact original Jira comment link. A separate new-comment button never pretends to create a threaded reply.
- Edit existing comments inline: fetch the latest original, review before/after, then explicitly save to the same comment ID. Jira permissions still apply. Concurrent changes/deletion block an overwrite and keep the draft.
- Editing preserves original media, mentions and complex formatting as protected placeholders. Keep these markers unchanged; edit those elements in Jira itself. Visibility stays unchanged, and saving refreshes the displayed comments automatically.
- Type
@and at least two characters, then select a Jira member. Selected account mentions are preserved during translation and included in the reviewed outgoing rich-text body. - After an acknowledged submission, automatically refresh comments. A newly posted comment outside the current filter/page is shown separately; a refresh failure does not trigger another submission.
- Actual Jira posting still requires explicit human confirmation. Automated tests use mock writes, not production comments.
긴 작업 일지는 원본 티켓과 업무현황 팝업에서 화면 높이의 55%(최대 560px) 안에서 스크롤됩니다. 최근/전체/정렬 버튼은 목록 밖에 유지하며, 원본 노트의 동일 내용 재렌더링을 생략하고 갱신 시 목록 스크롤 위치를 보존합니다.
작업 일지 카드 준비가 늦어져도 각 Markdown 조각의 위치 정보를 다시 확인하여 원문 목록의 중복 표시를 방지합니다. 수동 문단·To-Do·회의록은 그대로 보존하며, 원본 기록을 이동하거나 삭제하지 않습니다.
ServiceNow tickets, work notes, document links, status updates, work logs, and To-Do dashboards inside Obsidian.
This is an independent community project. It is not affiliated with, endorsed by, or provided by ServiceNow or Obsidian.
Maintainer: thkim9916 · [email protected]
What's new in 2.18.1: work-log display and date correction
- Keep one work-log history per section; do not duplicate it into each list mentioning a date.
- Read dates only at record starts. Preserve legacy dated headings, undated notes and nested content without rewriting source notes.
- Share the parser between ticket notes and the dashboard, and refuse stale or ambiguous deletion requests.
See the 2.18.1 patch notes for changes and update precautions.
Previous release: 2.18.0 — optional Atlassian connection and Jira workspace
Connect Jira and Confluence optionally, automatically refresh linked Jira work with ServiceNow tickets, and read comments as conversations. Import Jira schedules into local To-Dos, translate comments automatically, and make separately authorized, human-confirmed Jira changes. Existing notes and settings are preserved; Atlassian remains off for users who have not enabled it.
See the full 2.18.0 release notes for features, fixes, upgrade steps and authorization limits.
Settings now offers an optional Jira · Confluence connection, disabled by default. Enable it, choose Connect Atlassian account, and complete company-account login and read-access approval in the browser. Select an accessible site if more than one is available, then run Connection test. No API token or OAuth JSON needs to be pasted into the plugin. Company app policies can still require administrator approval.
Jira and Confluence have separate results. Connected means the plugin has actually read an issue or page during the current session, not merely stored a token. Empty search results, missing access, expired login, and network errors remain distinct. On restart, a previous success is shown as historical until retested. Optional test links must belong to the selected site; otherwise one recent accessible issue/page is used. Account tests do not save issue/page content. Default consent is read/search only. Optional direct Jira editing requires both a separate setting and renewed OAuth write consent; merely turning on a switch cannot grant access.
Ticket notes offer Work records & comments (ServiceNow / Jira) and Create schedule. Records open one window with separate ServiceNow work notes and Jira comments tabs; the Jira tab also supports composing comments. Schedule creation uses DS document / Jira tabs in the same window, not a chain of source-picker popups. The Jira workspace reuses both panels alongside direct comments and status/deadline tools. Viewing or filtering issues never silently chooses a write target.
-
The comment issue filter is a searchable checkbox list showing issue key, assignee, reporter and the configured start-date field (default
Plan Start Date) for every row. Duplicate activity names remain distinguishable. Select several issues for one time-ordered comment stream, select all, or clear all; an empty selection shows no comments. Filters never choose a comment-posting or mutation target. Selected issue chips, drafts and the single explicit write target are independent. -
Normal Jira refreshes retrieve reporter and configured plan-start dates as well as assignee, not only schedule imports. A human-named date field is discovered from evidence, then its site-and-setting-specific ID is requested narrowly. Missing dates/people remain visibly unspecified. Comment posting targets and schedule rows show the same identifying metadata.
-
ServiceNow's search/order/type/language filters wrap to the available note or popup width. Start/end dates stay together as a responsive range, retaining room for the full value, native calendar and current-time control; narrow split panes stack the fields instead of compressing them into icons. Accessible labels identify each filter independently.
-
With the Atlassian connection enabled, ServiceNow ticket/status refreshes also await a fresh, failure-isolated Jira read; an opt-out remains available. Opening the Jira workspace immediately retrieves current data while showing the saved snapshot first. Concurrent fresh requests coalesce; stale negative caches cannot block an explicit open or scheduled refresh. Open comment views update automatically without replacing the draft editor or losing its caret.
-
Hierarchy decisions use a full output projection with narrowly requested fields. MCP compact/evidence responses can omit requested parent, hierarchy and status IDs; realistic projection fixtures and captured-response replay guard this contract. Exact ticket matches collapse to their relevant root; a unique main Epic is preferred over independent review/task matches, without hard-coded project keys or localized type names. Only competing real roots require human selection. Manual links remain intact; plugin-owned child links can be corrected to their verified main issue.
-
Ambiguous roots use equal-width, vertically stacked cards and an in-app description/hierarchy/recent-comments preview. Manual linking is a live Jira search by CR/SR, issue key/URL or title, with typed results and pagination. A CR/SR search groups child matches under main issues. Preview/search alone never changes a link; the explicit final connection action checks for a concurrent change. Connected actions use verb-based labels for reading/writing comments, combining ServiceNow history, importing work as To-Dos and changing state/deadline. Empty/unconnected screens do not expose unusable tabs.
-
Read all accessible comment pages across the original issue and descendants, filter by issue/date/text, and choose oldest/newest conversation order. Native Jira parent-comment IDs group each original comment with its replies, shown underneath in writing order with an indentation, connector and reply-author label. Conversations stay together when paging. A matching reply retains its original comment and conversation context; rows outside the search/date filter are clearly labeled. Missing parents and malformed relationships stay visible without inventing content or grouping across issues. Translation, inline drafts and per-comment actions use the same visible conversation page. Source issue, author, edited time and restrictions remain intact. Display optional translations without changing Jira. Safe supported images are downloaded to ticket assets; unresolved inline positions appear in an issue gallery rather than being invented.
-
Selecting Korean or English automatically translates uncached comments on the visible page with the configured translation provider. Page/filter changes and refreshed comment bodies request only missing translations. Original text remains available. In-flight requests coalesce, stop requesting superseded rows, and cannot attach old results to edited comments or another site. Failures display a retry action instead of silently remaining untranslated or retrying in a render loop.
-
Choose Add comment directly under a source comment. The inline editor fixes its target to that comment's issue, with no second issue selector. Separate drafts keep their text and caret across refreshes, translations and filters. Language/translate/review actions align on one baseline and wrap as a group on narrow screens. A selected-issue/main-issue new-comment action is also available below the conversation. The current connector adds a new issue comment, not a native threaded reply or an edit to the original; the source reference is shown before final confirmation. Source role/group restrictions and JSM internal-note boundaries are preserved and rechecked before sending.
-
AI ticket prompts show hierarchy first, followed by a merged chronological ServiceNow/Jira history. Internal/restricted comments are excluded from default external-AI prompts. Source material is explicitly distinguished from AI instructions; partial or cached results are labeled.
-
Ticket-analysis generation keeps its progress and result outside the rendered note DOM. Jira/status refreshes, split panes and reopening the note cannot silently discard the completed preview. Concurrent clicks share one request; failures remain visible with retry, and the previous successful prompt stays available during regeneration. This bounded, root-specific preview cache is memory-only and clears when the plugin unloads.
-
Jira rich-HTML media blocks and ADF images keep their original comment positions using safely downloaded, issue-scoped attachments. Ambiguous or unavailable media stays visibly unresolved. Translation protects image-position markers and restores local embeds without sending local paths to the translation provider; a missing or duplicated marker falls back to the original comment rather than hiding its image.
-
Jira schedules are fetched again before local To-Do creation. Choose rows, Start/Due date basis and an enabled local status. Configure the site's start-date field label/ID (default
Plan Start Date); absent dates stay absent. Immutable source identities prevent duplicate imports without overwriting manual To-Dos. Date-only values use the creation time. -
Direct editing supports new comments, currently allowed workflow transitions, and due-date changes only. Confirm the exact issue, outgoing language/body or before/after values before each request. Transitions with unsupported extra required fields stay in Jira. There is no automatic mutation retry: uncertain outcomes lock further writes until the user checks the original. Confluence remains read-only.
No Jira-only ticket registration or Confluence publication is added. Live company consent and real business mutations are not replaced by mock/UI tests. No real Jira changes are made during automated tests.
The automatic Jira test searches only issues updated in the last 30 days, then reads one result. A result limit alone is not a search bound: the Jira service rejects unbounded JQL. If there is no recent issue, specify an accessible issue in Test a specific issue/page; direct issue tests do not require recent updates. Query rejection is shown separately from account/permission errors, with a sanitized explanation instead of raw server details.
Previous release: 2.17.0
- Optional AI Wiki under the configured work root, with standardized work records, ticket-summary links, metadata checks and a five-step connection helper.
- Compact cards or a saved spreadsheet-style table, shared search/filter/pagination, and a read-only detail popup before opening a Wiki note.
- Export-only editing for Jira tables, cell text and JSON, synchronized across formats without changing source notes.
- Keep every ticketless To-Do separate even when grouping ticket-based work; fill missing Ticket No. and Description only for this export.
- Clear pending/applied Wiki settings, visible confirmation actions, read-check status and approval boundaries.
See the full 2.17.0 release notes for changes, upgrade steps, and limitations.
Optional AI Wiki
AI Wiki is off by default, independent of organization packs and ServiceNow/Google sign-in. Skip its optional initial-setup step or enable it later in plugin settings. The Wiki lives under the configured work root, not a hard-coded project: <root>/LLM WIKI by default. Its folder name is configurable; changing that preference does not move existing records. Disabling the feature hides navigation and preserves files. It does not revoke permissions or data already shared with another AI.
- Open AI Wiki from the dashboard, ribbon, command palette, or a ticket. Browse recent, ticket-specific, topic-specific, and needs-review records. Lists use cached metadata, pagination and local file-change refresh, not repeated whole-Vault content reads.
- Choose compact summary cards or a six-column spreadsheet-style read-only table; the layout choice is saved. Table headings sort by title or date while search, filters and pagination remain shared. Review warnings use compact badges with full detail in the popup. Selecting a record opens a read-only detail popup first; closing it keeps the list, and opening the original note is an explicit action. Ticket-changing actions live in a separate expandable section.
- The detail popup shows the note body, summary, next action, review warnings, and whether its summary is already in each linked ticket's work log. Adding a summary requires confirmation and appends the standard entry plus the Wiki link to the ticket note. It does not move the Wiki note, change ticket state or To-Dos, duplicate an existing record ID, or automatically update an older work-log snapshot when the Wiki changes.
- Keep detailed work records in
작업기록, reusable verified knowledge in지식, and attachments in첨부. Human-facing guides remain separate. Multiple tickets can share one record; ticketless work is supported. Existing notes are never automatically reclassified or moved. - The AI start guide offers Codex, Claude, ChatGPT and other providers, copyable folder-preparation and starter requests, and a user-mediated file-read challenge. Step 2 copies a complete request to send to the AI; copying only the folder path is for a folder picker. After the AI confirms read access, step 3 sends the Wiki recording rules in the same chat.
- Local folder permissions and sign-in must be configured in the selected AI. Ordinary web chat cannot read a local path by itself; use an approved connector or explicitly attach selected files. App paths and CLI commands are not needed for the basic workflow. The optional local app launcher lives separately in advanced plugin settings. Opening an app is not an authenticated connection. No prompts or notes are automatically sent to an AI, no agent is run in the background, and no AI API key is stored.
- Recording defaults to asking before saving. A separate setting can tell the AI it may maintain request-related files inside the Wiki only. This is prompt guidance, not an OS permission boundary, and does not authorize ticket-body changes, code/DB changes, external messages, or ServiceNow/Jira/Confluence updates.
- A portable
ai-wiki-recorder/SKILL.mdis supplied inside the Wiki, not automatically installed. No existing AGENTS.md, CLAUDE.md, AI configuration, or global skills are overwritten. Unknown chat/task IDs remain unknown; the plugin cannot audit missing work from inaccessible conversations. - The record format includes IDs, dates, tickets, topic, summary, verification and sources. Duplicate IDs and incomplete evidence appear under needs-review. This checks record metadata, not the truth of an AI conclusion. AI recording instructions require read-back verification and preservation of manual content.
The generated AI 시작 안내.md explains the start workflow and links to official AI connection documentation. No version number, tag or release is created by enabling this feature.
Wiki settings distinguish unsaved folder edits, applying, applied and failed states. The usage/recording switches save immediately; only folder-name edits need the Apply button. Slow saves show progress and automatically unlock dependent controls when complete, without discarding a folder draft.
The five-step connection helper guides AI selection, folder or attachment preparation, the starter prompt, a read-response check, and the result. The last successful check (provider, Wiki location, non-secret challenge code, access method and time) is stored locally and shown after reopening. Local read checks and attachment-only delivery checks have different labels. Changing the Wiki location, losing/changing the check file or an unsuccessful save cannot show a valid local connection. This is historical read-check evidence, not live AI authentication or write authorization.
Preparation and prompt-delivery confirmations are explicit buttons in the pinned footer, not separate checkboxes: confirm folder-read/attachment preparation and continue, then confirm prompt delivery and continue. These are user acknowledgements only; the later file-read challenge remains required before a connection check can be saved.
Previous release: 2.16.0
- Separate active work from archived/dormant history, with shared lifecycle settings, on-demand history loading, and automatic reactivation when opening a ticket's root note.
- Import weekly/common meetings without a ticket, then link, relink, or unlink tickets later. File and Drive imports share searchable multi-select chips and an explicit ticketless option.
- Refresh the DS before selecting Schedule or PROD Schedule activities to create To-Dos. Choose Start or End as the due-date basis, preserve the creation time for date-only rows, and optionally add review preparation.
- Create related-note indexes and child notes directly from a ticket or its index, with navigation back to both the original ticket and the parent index.
- Filter Jira Export by multiple To-Do states; confirm temporary memos and preserve multiline paragraph/list spacing in cell-text output.
- Use shared To-Do sorting, current-time buttons, configurable popup width (75% default), compact toolbars, readable date fields, and a pinned To-Do action footer.
- Improve Drive document matching for separator whitespace and paginated results, and fix excessive date-control spacing and clipped responsive Export panels.
See the full 2.16.0 release notes for changes, upgrade steps, and limitations.
Previous release: 2.15.0
- Not started, In progress, Pending (waiting), and Done are separate states. Existing unchecked tasks remain Not started.
- Pending supports an optional waiting reason and follow-up date, separate from the completion due date. Cards show the waiting start date and highlight follow-ups due today or earlier; states never resume automatically.
- Use the Pending / follow-up-needed filters, drag cards between columns, and edit the same task from the ticket note or dashboard. Calendar date filtering and Jira table/JSON export include the new waiting fields.
- Configure state names, enabled states, and their order from either board or plugin settings. The configuration is shared across ticket notes and the dashboard; creation only offers enabled states. Existing tasks in disabled states remain available in the inactive-state viewer.
- Work guide pack and Google OAuth JSON import/replacement open visible dialogs with file selection, pasted JSON, validation, and retry, including after initial setup.
Requirements and disclosures
Meeting-to-ticket links are optional. File import and Google Drive import both offer a "Save without a ticket (weekly/common meeting)" choice, including when started from a ticket note. Unlinked notes and optional PDFs are stored under the configured root's 회의록 folder and appear in the dashboard's weekly/common meeting group. Drive's general automatic search finds Gemini meeting documents without requiring a ticket number; direct title search also supports other meeting names. No-ticket Drive imports are checked for duplicate source IDs in the unlinked history.
Existing source meeting notes can be linked, relinked, or unlinked later from their note or meeting cards. This updates ticket, tickets, Parent, Parents, and the generated ticket-summary row in one note write; unrelated metadata and meeting content are preserved. Files/PDFs are not moved or copied. All linked ticket lists share the same note, and clearing links returns it to the weekly/common group regardless of its original folder. Ticket-specific AI analysis documents are not reassigned; change links on the source meeting instead.
Active/history management keeps completed-state tickets out of scheduled refreshes and the default dashboard, without moving or deleting notes. Dormancy defaults to both three calendar months without opening the root note and six months without meaningful changes; first-time observation starts now rather than inventing older visit history. The default completed states are Implemented, Closed, Resolved, and Cancelled, with seven days of change grace. Recent root-note visits and unfinished To-Dos keep a ticket active. Lifecycle settings apply globally rather than requiring per-ticket configuration. Opening a root note records a visit and reactivates it. Background reads, search results, and unchanged synchronizations do not count as visits. Archived remote changes are not polled automatically; open the root note or run a manual inclusive refresh to check them.
The dashboard offers Active, All, Archived, and Dormant scopes. Exact ticket-number lookup can find archived/dormant notes even with table filters; ordinary history search is available in the inclusive scope. Root-note details are loaded on demand and reused until modified. Activity timestamps, hashes, and a bounded text/reference index are stored only in the Vault plugin folder's private ticket-activity.json; Vault backup/sync settings still apply. Existing ticket contents and ServiceNow states are not rewritten by classification. Configurable periods and completed states are shared between ticket notes and the dashboard.
- Obsidian Desktop 1.11.5 or later. Mobile is not supported because the plugin uses desktop OAuth callback and Electron/Node APIs.
- A ServiceNow account and an approved API token are required only for ServiceNow synchronization. Local note and To-Do features remain available without a connection.
- Google Drive integration is optional and requires a user-supplied Google Desktop OAuth JSON file and a Google account with access to the target files.
- Organization-specific status guides, SLA rules, and AI templates are not bundled. They appear only after the user imports a separately supplied ServiceNow Manage work guide pack and enables work guide features.
- The plugin connects only to the configured ServiceNow instance, optional Google OAuth/Drive, optional Atlassian Rovo MCP and Atlassian OAuth, Nager.Date for optional public-holiday lookup, and an optional translation plugin's provider.
- The plugin has no developer-operated backend, client telemetry, analytics, advertising, or automatic self-update mechanism.
- Tokens and client secrets are stored with Obsidian Secret Storage. Do not put tokens, OAuth JSON,
data.json, or organization packs in a public repository or GitHub Release. - Ticket details and downloaded documents are written to the user's Vault. Users are responsible for Vault access, backup, and sync permissions.
See PRIVACY.md and SECURITY.md.
Main features
- Create CR/SR ticket notes and related work-note views.
- Refresh ticket metadata and work notes from ServiceNow.
- Search, sort, filter, translate, copy, and date-filter work notes.
- Create work logs and To-Dos from ticket notes or a dashboard.
- Import Gemini meeting minutes from Markdown, optionally preserve the source PDF, and browse chronologically linked meeting notes from either the ticket note or dashboard.
- View To-Dos as a configurable four-state board with due dates, completion dates, search, details, and pagination.
- Maintain a configurable ticket dashboard with persistent columns, sorting, filters, widths, and field order.
- Optionally discover BS links from a description and FS/DS/UT files from Google Drive naming conventions.
- Optionally generate organization-specific AI prompts when an organization pack is installed.
- Optionally browse AI Wiki records as cards or a table and preview the original Markdown in a detail popup.
- Edit a shared Jira Export draft across table, cell-text and JSON formats without writing back to source notes.
- Schedule daily status refresh and optional work-note refresh.
- Configure the CR, SR, and Incident ServiceNow table names for different instances.
First setup
The first time the plugin starts, it opens a setup wizard. The wizard checks these items without bundling any organization credentials:
- Choose the Vault root folder. New installations use
ServiceNow. - Enter the full ServiceNow instance URL and, when applicable, a Bearer Token issued for the user or organization.
- Optionally import a Google Desktop OAuth JSON file or paste the JSON content.
- Optionally import a separately supplied work guide pack.
- Optionally enable AI Wiki and review its access and recording guidance. Skip this step to leave it disabled; it can be enabled later.
- Review the summary and finish.
Selecting Set up later dismisses the automatic wizard without disabling the plugin. Open it again from Settings → ServiceNow Manage → Initial setup wizard or from the command palette. All individual options remain available in the normal settings page.
After the wizard, change the CR/SR API table names under advanced settings if the instance uses custom tables. Optionally connect the Google account, enable document automation, translation, scheduled refresh, and public-holiday lookup.
Never use a browser session cookie or another person's token as an API token.
Work guide packs
A work guide pack is a local JSON file supplied separately by an organization. It can contain:
- CR/SR status guides;
- organization SLA/due-date rules;
- an AI prompt template;
- CR/SR AI analysis templates;
- optional feature flags.
Import it from Settings → ServiceNow Manage → Work guide pack → Import JSON. Uploading a newer pack replaces the active guide, rules, and feature configuration. The plugin copies the validated pack to its local plugin directory as organization-pack.json. This file is intentionally excluded from the public source repository.
Without a pack, status-guide and organization AI-prompt controls are hidden. Removing or disabling the pack hides those controls but does not delete user-authored Vault notes.
Google OAuth JSON is not a work-guide-pack license or access-control mechanism. Desktop OAuth client IDs can be copied and are not a secure way to authorize private workflow content.
See ORGANIZATION_PACK.md for the schema, a credential-free example, and safe distribution guidance.
Work guide packs may be revised and distributed independently of plugin releases. Keep the same packId, increment the optional version, and import the newer JSON with Replace pack. Runtime guides and rules use the replacement immediately. Existing user-edited Vault templates are preserved rather than overwritten.
Google Drive
Google Drive integration is optional. Import a Google Desktop OAuth JSON file, connect a Google account, and grant read-only Drive access. The client secret and account tokens use Obsidian Secret Storage; the client ID and non-secret connection metadata use normal plugin settings.
The built-in document rules are conventions, not universal ServiceNow behavior. Disable document automation if an organization uses different BS/FS/DS/UT locations or naming rules.
Translation
Translation is optional and uses a compatible Obsidian translation plugin/provider configured by the user. Provider limits, billing, privacy, and retention are controlled by that provider.
Data and removal
Disabling or uninstalling ServiceNow Manage does not delete ticket Markdown files, downloaded documents, work logs, or To-Dos. Disconnect ServiceNow and Google first if the stored credentials must be removed from Secret Storage.
Community release
Before a public release:
- Run
npm test. - Run
npm run security:audit. - Run
npm run release:check. - Confirm that
manifest.json,package.json,versions.json, and the GitHub tag use the same version. - Attach
main.js,manifest.json, andstyles.cssas individual GitHub Release assets. - Never publish
data.json, OAuth JSON,organization-pack.json, private work guide packs, Vault contents, credentials, or company documents.
See COMMUNITY_RELEASE_CHECKLIST.md for the complete release checklist.
Updating and releasing later versions
The maintainer can continue using and modifying the development copy after the first Community Plugin release. Publish a new GitHub Release whenever a tested plugin update is ready:
- patch (
2.0.1) for compatible fixes; - minor (
2.1.0) for compatible features; - major (
3.0.0) for breaking changes or migrations.
Update manifest.json, package.json, and versions.json together, run all release checks, create a GitHub tag exactly matching the version, and attach main.js, manifest.json, and styles.css. Obsidian can then offer the new release to existing users through the Community Plugin update flow. A work-guide-pack-only content update does not require a plugin version change while its schema remains compatible.