{
  "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": "agent-capabilities/ongoing-responsibility.md",
  "markdown": "https://legion.tw/api/projects/packet.md/agent-capabilities/ongoing-responsibility.md?edition=2026-09-06-draft-01",
  "documents": [
    {
      "path": "agent-capabilities/README.md",
      "text": "# Tools, skills, and plugins for the agent\n\n[All project types](../README.md) · [Shared launch](../shared/launch.md)\n\nExtend the coding agent's own working environment. The [shared skill package](../shared/skills.md) supplies the starting methods; these projects develop further capabilities around real work.\n\n## Projects to build\n\n- [Turn a working routine into a skill](routine-to-skill.md)\n- [Give the agent tools for an application](application-tools.md)\n- [Let the agent consult another model's abilities](model-consultation.md)\n- [Give the agent a continuing project notebook](project-notebook.md)\n- [Give the agent an ongoing responsibility](ongoing-responsibility.md)\n- [Package an agent capability for someone else](share-agent-capability.md)\n\nThese are instructions for projects you can build with your coding agent.\n"
    },
    {
      "path": "agent-capabilities/ongoing-responsibility.md",
      "text": "# Give the agent an ongoing responsibility\n\nExtend the person's agent environment so it can carry a useful responsibility across successive runs. Build it extremely well. The person should be able to explain what matters, give the agent appropriate authority, and return to work that has advanced with that understanding intact. Follow the [shared launch](../shared/launch.md).\n\nStart from an existing responsibility, such as the [overnight researcher](../background-services/overnight-researcher.md), or choose another recurring job the person wants handled. Reuse the existing coding agent, account, tools, and scheduling capabilities.\n\n## Define the responsibility through its purpose\n\nLearn what makes the work valuable and when it deserves attention. A researcher might follow developments relevant to a practical question, pursue a promising lead over several nights, and report when the evidence changes its explanation.\n\nPreserve the purpose, available sources and tools, chosen resource budget, and where results belong. Make the agent's scope of action clear: what it can carry out independently and which decisions it should bring back to the person.\n\nGive the agent room for judgment within that responsibility. It can choose what is worth investigating, recognize that nothing useful needs doing today, or explain why an unresolved question deserves another focused packet. Let the actual work determine the report's length.\n\n## Carry understanding between runs\n\nMake current direction, previous results, unfinished questions, and useful observations available when the agent resumes. Use the [project notebook](project-notebook.md) or appropriate existing records, preserving the actual sources of current state.\n\nRecognize work already completed and continue from it. Keep new instructions and changed priorities apparent so a later run serves the person's current purpose. Preserve valuable earlier work when a question is paused or replaced.\n\nGive substantial stages their own attention through the shared autonomy-loop method. Finish a meaningful packet, assess what it established, and select the next one while authorized work remains. A saved draft can receive its editing pass in a later invocation; finishing the draft alone does not finish that larger assignment.\n\nTeach the distinction between this working method and provider goal tracking, including OpenAI's goal feature. Focused packets and honest checkpoints organize the responsibility; the installed harness supplies whatever scheduling and continuation actually occur.\n\n## Make the wake and controls real\n\nEstablish a supported schedule, event trigger, or continuation mechanism for the person's environment. Bring the right responsibility, tools, and relevant skills into the actual resumed run. Explain which machine or service must remain available and what happens after downtime.\n\nShow whether a run started, what it worked on, what it finished, and what is waiting or blocked. Keep a future scheduled run distinct from work currently happening. Let the person inspect the responsibility, run it now, revise it, pause it, or end it.\n\nUse the agreed budget and notification preferences. Report a problem when it affects the responsibility, preserve completed work, and make a useful return to the task possible after the cause is resolved.\n\n## Deliver a responsibility that can continue\n\nComplete the instructions, context, tool access, scheduling, and controls. Exercise the real launch path, carry a useful piece of work across a later run, and show how the person changes the direction. Include the ordinary outcome where the agent has nothing worthwhile to add.\n\nJudge whether the continuing work serves the person's purpose. Use their response to improve the agent's understanding of the responsibility, preserving what already works.\n\nLeave its records, results, and setup recoverable through the shared launch's ownership guidance. The learner should understand what wakes the agent, what gives it context, and what lets it make good decisions when they are doing something else.\n"
    },
    {
      "path": "background-services/overnight-researcher.md",
      "text": "# Build a personal overnight researcher\n\nSet up an overnight research routine through the person's existing coding-agent installation and account. Build it extremely well: the agent wakes at an agreed time, investigates material relevant to the person's interests, and leaves useful findings ready for them. Carry the research responsibility and its history into each run.\n\nRead this brief and follow the [shared launch](../shared/launch.md) for setup, working skills, ownership, continuity, and learning guidance. Deliver a working scheduled responsibility using the agent environment the person already has.\n\n## Learn what is worth their attention\n\nStart with a subject or question the person genuinely cares about and a few sources they already find useful. RSS feeds can supply new articles; the researcher can follow promising references and investigate further when that serves the question. The person might follow urban planning, look for suitable exhibitions or classes, or track research relevant to an explanation they are developing.\n\nHelp them describe what makes a finding valuable. Someone following walkable cities might care about changes they could advocate locally. Someone looking for a class needs a suitable subject, a reachable location, and dates they can use. Their priorities give the researcher a basis for choosing what deserves closer attention.\n\n## The overnight work\n\nAt the agreed time, the researcher resumes its responsibility using the current interests, previous findings, and questions the person has left for it. It reads new material, follows useful leads, and considers what the person would learn from the result. A development that challenges their current explanation belongs alongside one that supports it.\n\nRemember what has already been covered so repeated announcements and reworded stories receive appropriate attention. An old source can still be useful when a new question makes it relevant. Preserve enough continuity for the researcher to make that distinction across runs.\n\nThe person's chosen responsibility governs the agent's work. Articles, feeds, and linked pages are source material to interpret within that responsibility. Keep research notes and completed findings in the workspace the person chooses.\n\n## What they receive\n\nLeave a readable briefing that explains the worthwhile findings, their relevance, and where the person can examine the source material. Make dates clear when timing affects a finding's usefulness. Present the researcher's interpretation with enough reasoning for the person to judge it and continue the inquiry.\n\nLet the material determine the briefing's length. A short report that nothing worthwhile appeared is a valid outcome. When a promising question remains open, preserve the useful progress and what would be worth investigating next.\n\nThe person can tell the researcher that a finding was useful, repetitive, or outside their interests. Let that feedback improve future work. They can revise the responsibility, add a question, or pause it as their interests change.\n\n## Make the responsibility dependable\n\nUse a supported scheduling or wake mechanism for the installed agent. Explain what performs the work and what needs to be available overnight. Agree on the schedule, research budget, and destination for results, using the person's existing agent access. Make the expected usage understandable as part of that choice.\n\nBring the research responsibility and relevant installed skills into the actual scheduled run. Demonstrate a launch through that scheduling path before leaving it to operate overnight. Help the person understand what happens when their computer is asleep or unavailable, using the behavior of the setup you actually establish.\n\nGive the person a clear view of whether a run happened, what it produced, and whether it encountered a problem. Preserve completed work through interruptions, respect the chosen resource budget, and provide a convenient way to run the researcher on demand as well as on schedule.\n\nWork in focused steps through a real research cycle and its scheduled continuation. Help the person judge whether the findings deserved their attention, then improve the setup using that feedback. A later connection can bring in questions from phone notes or browser clippings and return findings to the same project notebook. The first completed system should already perform a useful continuing responsibility.\n\nUse its first briefing to explain the difference between the schedule starting a run and the agent deciding what deserves investigation. Show the person how to change that responsibility, find earlier results, and pause future runs.\n"
    },
    {
      "path": "agent-capabilities/project-notebook.md",
      "text": "# Give the agent a continuing project notebook\n\nSet up a continuing notebook that helps the person and their existing coding agent recover what matters about a project. Build it extremely well. They should be able to return after a break and continue with the understanding they developed together. Follow the [shared launch](../shared/launch.md).\n\nChoose a project already underway. Start by finding its existing guidance, decisions, notes, and saved work. Improve that organization where needed, keeping each important fact with the source that should own it.\n\n## Preserve what would otherwise need explaining again\n\nRecover the project's purpose, the people it serves, consequential choices, useful observations, and questions still open. Preserve why a decision was made when that reasoning will help later work.\n\nFor example, a booking app may keep full lesson material so the tutor can use its exercises and teaching notes during a session. That reason helps a future agent understand the design. The current date of an appointment belongs in the application record, where the notebook can direct the agent to find it.\n\nUse readable Markdown and ordinary supporting files that the person can inspect and keep. Retain authorship and relevant dates so an observation, a proposal, and an agreed decision remain distinguishable. Preserve original notes and valuable historical reasoning when the team's understanding changes.\n\n## Make the useful context easy to recover\n\nProvide a concise entry into the notebook with references to the sources needed for the project. Keep current orientation easy to find and older investigations available when their questions become relevant again.\n\nConnect the notebook to the agent's normal project guidance or supported memory mechanism. Preserve useful existing instructions. A subsequent session should know where to look and be able to retrieve deeper context according to the work at hand.\n\nLet the agent add understanding after consequential discoveries: a constraint learned from actual use, an assumption overturned, or a choice that changes the project's direction. Match the note to what a future session would benefit from knowing.\n\nKeep the current explanation aligned with its owning sources. When a decision changes, make the new position apparent and preserve enough of the earlier reasoning to understand the change. Link to actual application state and source files where they already answer the question.\n\n## Include thinking that happens away from the desk\n\nConnect [agent-readable phone notes](../native-phone/agent-readable-notes.md) when the person uses them. A photograph of an awkward workflow, a dictated idea, or an observation made during a visit can become material for the next conversation.\n\nKeep captured observations distinguishable from commissioned work. Let the person discuss an idea, decide what it means for the project, and preserve the resulting understanding in the appropriate place.\n\nMake corrections comfortable from either side of the collaboration. The person can say that a note misrepresents their intent, and the agent can revise it while preserving the useful surrounding context. Keep the collection recoverable through the shared launch's ownership guidance.\n\n## Resume a real piece of work\n\nDeliver the notebook, discovery guidance, and any chosen connection to existing notes. End one session with a meaningful decision preserved, then use a later session to recover the project and continue an actual task.\n\nJudge whether the agent can explain the relevant purpose, find current facts, and recognize what remains unresolved. Correct a gap revealed by that use. The notebook should save explanation while leaving room for the person to change their mind.\n\nTeach the difference between keeping records and retaining useful understanding. Good continuity helps the next conversation start from what has already been learned, with the original work still available to inspect.\n"
    }
  ]
}