{
  "edition": {
    "edition": "2026-09-06-draft-01",
    "language": "en",
    "publicationState": "published",
    "sourceRevision": "a7e788eb4b151fd92c1f58153e015773e3a829af",
    "projectCount": 67,
    "documentCount": 94,
    "skillCount": 10,
    "archivePath": "library.zip",
    "archiveBytes": 181812,
    "releasedAt": "2026-09-08T05:03:11.129+00:00"
  },
  "access": {
    "released": true,
    "rehearsal": false
  },
  "project": "connected-systems/site-visit-to-report.md",
  "markdown": "https://legion.tw/api/projects/packet.md/connected-systems/site-visit-to-report.md?edition=2026-09-06-draft-01",
  "documents": [
    {
      "path": "connected-systems/README.md",
      "text": "# Connected systems\n\n[All project types](../README.md) · [Shared launch](../shared/launch.md)\n\nExtend useful applications into a connected workflow, preserving their records and letting each device serve its actual role.\n\n## Projects to build\n\n- [From a site visit to a finished report](site-visit-to-report.md)\n- [Your engineer at home, available from your pocket](engineer-in-your-pocket.md)\n- [A research system that keeps up with your thinking](connected-research.md)\n- [A game night across different screens](multi-screen-game-night.md)\n- [Capture outside and process on the computer](capture-to-processing.md)\n- [A shared lending service with private working tools](lending-service.md)\n\nThese are instructions for projects you can build with your coding agent.\n"
    },
    {
      "path": "connected-systems/site-visit-to-report.md",
      "text": "# From a site visit to a finished report\n\nConnect the [photo-report phone companion](../native-phone/photo-report-companion.md), [desktop report desk and AI colleague](../desktop/ai-photo-report-desk.md), and a suitable delivery-preparation routine into one working system. Build it extremely well. The person should finish a visit with material already organized for the report they owe. Follow the [shared launch](../shared/launch.md).\n\nStart from the applications and records they have. Establish any missing foundation needed for this workflow, retaining the native Tauri desk and the phone's useful capture experience. Use a fictional job when demonstrating work that would otherwise expose private customer information.\n\n## Follow one job through the whole process\n\nLearn what the recipient needs, what the person must observe on site, and how a report becomes ready to deliver. Carry the same job identity and purpose through preparation, capture, editing, and delivery.\n\nBefore the visit, let the phone receive the useful job details and subjects to cover. During the visit, photographs and observations belong to those subjects, with room for something unexpected. Let the person see enough of what they have collected to notice a missing view while another photograph is still possible.\n\nUse Tailscale as the starting choice for connecting their private devices, adapting to a suitable existing connection. Make connectivity and receipt understandable: material can be saved on the phone while the computer is unavailable, then reach the desk with its context intact.\n\n## Let report preparation begin from collected work\n\nAs material arrives, make it available in the desktop's ordinary job records and to the existing coding agent through the report desk's tools. The agent can compare images, group observations, and prepare a useful draft according to the person's direction.\n\nKeep the collected evidence, the person's explanation, and the report's wording distinguishable. Let the person correct a subject or add a late observation without losing edits already made to the report. Make clear which changes have reached the other device and which still await connection.\n\nGive the person a convenient way to indicate that collection is complete enough for the next stage. Several received photographs alone may be only part of the visit. Let them deliberately start preparation early when that helps, with the remaining material still visible as expected work.\n\n## Finish and prepare delivery\n\nKeep the report editable in its established workspace. The person can discuss it with the agent, adjust image choices and wording, and inspect the actual exported document.\n\nConnect a background preparation step when a report is ready: assemble the requested document and supporting files, name the files as the recipient expects, and save the package where the person can find it. The [event-follow-up brief](../background-services/event-follow-up.md) supplies the related automation pattern.\n\nGive each delivered version a recoverable relationship to the report it came from. A later correction should produce an understandable revision while preserving what was previously sent. Let the person choose whether the system prepares delivery or also performs an authorized sending action.\n\n## Deliver the complete workflow\n\nSet up the private connection, receiving operations, agent access, and background preparation as parts of the existing applications. Explain which computer or service must be available for each stage and provide ordinary controls for resuming or retrying unfinished work.\n\nTake a sample job through preparation, several on-site subjects, a connection interruption, report editing, and a usable delivery package. Judge whether the person still has to reconstruct information they already recorded.\n\nTeach how connected software preserves the meaning of work across devices. The phone serves the visit, the desktop serves close inspection and editing, and the agent helps with the same job throughout.\n"
    },
    {
      "path": "native-phone/photo-report-companion.md",
      "text": "# Capture report photographs while you are there\n\nBuild a native Android companion for the person's [photo-report desk](../desktop/ai-photo-report-desk.md). Photographs and observations should arrive at the desktop already associated with the right job and subject. Build it extremely well for the person taking pictures on site. Follow the [shared launch](../shared/launch.md).\n\nBegin from the desktop application they already use. If it has not been built yet, help them establish that foundation first. Preserve its jobs, photographs, reports, and agent collaboration as this new way of collecting material is added.\n\n## Follow the visit\n\nLearn what the person photographs and how they move through a job. They may document completed work, record damage, or collect images for a property report. Understand what they need to remember about each subject and how that information appears in the finished report.\n\nLet them select the job, identify a subject or location, and take photographs with that context already attached. Keep the current job and subject clear while the camera is in use. Moving to the next subject should be quick and deliberate.\n\nMake it easy to add an observation while it is fresh, using the input methods appropriate to their phone and situation. Several photographs may belong to one observation. Let the person review that grouping, correct a misplaced image, and add a useful note without reconstructing the visit later.\n\n## Make capture dependable away from the desk\n\nKeep photographs and notes saved on the phone through interruptions and unavailable connections. The person should be able to take the next picture, answer a call, return to the app, and continue the same job.\n\nLet them inspect a photograph at a useful size and decide whether another shot is needed while they are still there. Preserve the original material and enough image quality for the report's purpose. Make the relationship between an original, a preview, and any edited version understandable.\n\nGive particular attention to one-handed use, outdoor visibility, and quickly finding the current subject. Choose controls that fit the actual visit.\n\n## Bring organized material home\n\nSet up the connection to the existing desktop application as part of this project. Tailscale is a useful starting choice for private connectivity between the person's phone and computer; adapt to a suitable connection they already have where appropriate.\n\nSend the photographs with their job, subject, and observations. Let the desktop receive them into its ordinary working records so both the person and the existing AI colleague can use them in the report.\n\nShow what is saved on the phone, what is waiting to transfer, and what the desktop has received. Resuming a transfer should preserve the intended records and recognize material already received. Keep the phone's capture available through transfer problems and give the person understandable storage and retention controls.\n\nLet corrections to the captured context reach the report workflow coherently. Preserve work already done at the desktop, and make any decision about changing an existing report understandable.\n\n## Complete a visit and a report\n\nHandle the Android build and installation for the person's phone. Explain the permissions needed for capture and the connection to their computer. Use the shared launch's ownership and recovery guidance for both applications and their material.\n\nCapture a representative job with several subjects, interrupt and resume the phone app, transfer the material, and use it in the desktop report. Judge whether the original observations are still clear and whether evening organization has actually been reduced.\n\nTeach the difference between having a photograph and having a photograph with its purpose preserved. Later connected-system projects can add further report preparation or delivery, building on the same jobs and material captured here.\n"
    },
    {
      "path": "desktop/ai-photo-report-desk.md",
      "text": "# Build a photo-to-report desk with an AI colleague\n\nBuild a native desktop application with Tauri for turning photographs and field notes into reports. Build it extremely well, as a shared workplace where you and the person can organize the material, discuss what it shows, and prepare something useful to send. Tauri is the default; an explicit framework preference from the person takes precedence.\n\nRead this brief and follow the [shared launch](../shared/launch.md), which supplies setup, working skills, ownership, continuity, and learning guidance. Then build this project, carrying forward any setup the person already has.\n\n## Understand the report they actually make\n\nSomeone returns from a visit with photographs and observations. They need to identify the useful images, explain them, and assemble a report for a particular recipient. The work might involve documenting repairs, completed work, a property, or a delivery. Learn which reports the person makes and look at an example they choose to share. Use that understanding to shape the application.\n\nSomeone reading the report may need to understand a problem, confirm completed work, or decide what happens next. Let that purpose determine the arrangement of photographs and explanation. Start with a complete example report whose layout the person can change as they discover what works.\n\n## A place to work on the report\n\nPhotographs and notes belong to a job. The person can browse the images at useful sizes, compare similar shots, choose what to include, and arrange sections and captions in an editable report. Keep the original material available as the report develops. They can save their work, return to it, and export a readable document for the recipient in the format the job calls for.\n\nGive particular care to the movement between examining a photograph and working on its place in the report. The person should be able to see the relevant image while deciding what to say about it. The report's final layout should support the recipient's understanding of the work.\n\n## Work directly in the application\n\nBuild yourself first-class access to the job's photographs, notes, and current report, together with useful tools for operating them. Use your existing coding-agent installation, sign-in, and model access for this collaboration. They can discuss the work with you in their usual conversation and see your changes in the application.\n\nProvide an appropriate local integration, such as MCP tools, so you can compare images, group related material, draft captions, choose a useful sequence, and assemble the actual report in the editor. Make that connection part of the delivered application.\n\nMake the person's current job, report, and image selection available through that connection. A reference to “this photograph” should have the same meaning for both of you. Your changes should appear in their working view, and your next action should use the current report, including their direct edits.\n\nFor example, the person might say, “These three show the same damaged hinge. Use the clearest one and include my note about replacing it.” You inspect the images, select one, and make that change in the report. The person can then edit the caption themselves.\n\nTreat the photographs and field observations as the basis for factual descriptions. Keep visible evidence distinguishable from the person's explanation of what happened. Ask about a genuinely unclear image or suggest wording that accurately reflects the information available. Let ordinary discussion, direct editing, and your actions work together naturally, with useful ways to inspect and undo changes.\n\n## Make it theirs\n\nSet up Tauri and the agent connection for the person's computer. Make clear which material reaches the model provider during collaboration and which stays local. The person retains access to their photographs and reports independently of an AI session.\n\nWork in focused steps and carry the project through a real report-making experience. Judge image selection, captions, editing, and the exported report together. Help the person recognize what you can handle and where their knowledge of the visit changes the result.\n\nLeave them able to open the installed application, prepare a report with you, close it, and return to the saved job. Use that experience to explain the difference between discussing an application and giving an agent tools to work in it. Their next improvement can begin from the report they just made.\n\nA later phone companion could capture photographs with their job and notes already attached. Build a foundation that can receive that material while preserving the reports and working habits established here. This desktop application should be useful with the photographs the person already takes.\n"
    },
    {
      "path": "background-services/event-follow-up.md",
      "text": "# Handle the paperwork after the work is done\n\nBuild an automation that performs the recurring administration following a meaningful event in the person's work. Build it extremely well around a routine they already understand. Follow the [shared launch](../shared/launch.md).\n\nChoose one complete follow-up with the person: prepare records after a club session, assemble the supplier order after a [group order](../shared-web/group-order.md) closes, or prepare the delivery material for a completed [report](../desktop/ai-photo-report-desk.md). Use the existing application and its actual records.\n\n## Learn what starts the follow-up\n\nAsk the person to show what they do when this part of the job finishes. Learn what information is ready, what still needs judgment, and what the follow-up produces.\n\nGive the trigger a clear meaning in that workflow. A session's scheduled end, confirmed attendance, and completion of its administration are different events. Start the routine from the state that actually supplies the information it needs.\n\nUse the source application's own record of that event. Keep the follow-up associated with the session, order, or job it belongs to, so the person can see why it ran and find its result from the work itself.\n\n## Complete the usual administration\n\nPrepare the records, lists, or delivery material the person normally creates. Use actual recorded information and preserve the distinctions needed for the next activity. A supplier needs item quantities; the organizer may also need each person's collection details.\n\nKeep reusable instructions or layouts separate from the current event's information. Apply the existing formatting and preparation rules where the person already has a working routine.\n\nAgree which actions run automatically and which prepare material for the person to inspect or send. For an authorized automatic delivery, use the agreed recipients and communication method and show its outcome. For preparation-only work, leave the result clearly ready for the person's next action.\n\n## Let the person understand what happened\n\nShow the originating event, the follow-up's current state, and its completed result. Make a missing input or failed action visible with enough context to resolve it. Let the person retry or invoke the routine manually when useful.\n\nRecognize an event that has already been handled so a repeated signal or retry preserves the completed work and any delivery already made. If the underlying event is corrected afterward, make the revised follow-up an explicit update with the earlier result still understandable.\n\nKeep the person's direct edits to prepared material intact according to the workflow you establish together. Give them a clear choice when refreshing a result would affect those edits.\n\n## Run it from the real application\n\nInstall the automation where the work actually runs and explain its availability, startup, and external service requirements. Give the person a way to inspect, pause, or stop the routine. Keep its setup and useful history under the shared launch's ownership and recovery guidance.\n\nDeliver the complete follow-up and exercise it from a representative event in the source application. Inspect the generated result, handle a correction, and confirm that repeating the event does the right thing with existing output. Where delivery is authorized, verify the chosen delivery result.\n\nJudge whether the person can finish the original job and find the administration ready in the form they need. Explain how a recorded change in work can start another useful action.\n\nLater, add another follow-up routine with its own trigger and authority. Preserve the existing routine as new actions are added.\n"
    }
  ]
}