{"code":200,"message":"Zero to a live result in a few calls.","data":{"skill":"quickstart","text":"---\nname: egbe-os-quickstart\ndescription: Introduce Egbe, connect a supported workspace or explain how to move to one.\nmetadata:\n  egbe-when: \"Use for initial setup or questions about getting started.\"\n  egbe-category: \"Business operations\"\n  egbe-visible: \"false\"\n  egbe-title: \"Connect to Egbe\"\n  egbe-description: \"Discover Egbe and connect your AI co-founder to your project, tools and working methods.\"\n---\n\n# Get started with Egbe\n\nHelp the founder start useful work with **Egbe**. Introduce its value before\nsetup, then use the path supported by this environment. Use the supplied `OS_URL`\nor `https://os.hz.egbe.dev`. Public instructions need no `OS_API_KEY`.\n\nBefore installation, read linked resources at\n`GET /skills/quickstart/resources/<relative-path>` on `OS_URL`, with a path\nstarting in `references/` or `resources/`. `/quickstart` returns JSON with the\ninstruction in `data.text`; resource URLs return files directly. When a terminal\nis available, fetch and read the complete text through curl, not a web summary.\nOtherwise use available web-reading tools and retrieve omitted sections before\nacting on them; a summary is not the installation procedure.\n\n## 1. Show what Egbe enables\n\nRead public `GET /` and the [introduction example](references/introduction.md).\nUse the catalog's descriptions and categories; retain `instruction_url` for\nrelevant skills and playbooks. The catalog manifest at `/skills` is for the\ninstaller, not this introduction.\n\nGive the founder a warm, compact introduction in their language **now**, before\ncredential checks, catalog updates, installation or logging. Connect what Egbe\nprovides to results you can deliver, as in the example. Describe platform\ncapabilities, not an already configured business; project availability is checked\nlater. If public discovery fails, explain the gap and follow the environment\nrecovery below instead of presenting unverified current access as fact.\n\n## 2. Establish the available path\n\nBefore saving credentials or installing anything, identify the project's working\ndirectory. Establish from the environment context or native documentation that\nits files and configuration persist across sessions and the agent can reuse its\nproject-scoped skills/instructions there. An empty local or remote project folder\nis valid; a repository is not required. A terminal, a writable temporary directory,\nan Egbe project key or successful API access does not establish this workspace.\n\n- **Persistent project workspace established:** with commands and safe\n  authenticated HTTP available, briefly say where setup will be kept and continue\n  through steps 3–7.\n- **Introduction and handoff:** if the workspace is absent/unconfirmed, commands\n  are unsupported, or installation is declined, follow the\n  [informational path](references/without-connection.md) and finish quickstart\n  with that handoff, skipping steps 3–7. Do not try to turn the\n  chat's temporary execution environment into one through installation or plugins.\n- **A supported capability is blocked in an established workspace:** use\n  [environment recovery](references/environment.md), explain the concrete remedy\n  and retry after approval. Keep that setup pending and continue independent\n  discussion. Do not treat it as completed or move installation outside the project.\n\nPublic-reading access can be recovered in either path; restoring it does not\nestablish a persistent workspace. Within a supported workspace, assess scheduling\nand logging separately: missing support for either does not prevent development\nor a manual report. Use the app's actual capabilities, not its name.\n\n## 3. Install skills, playbooks and project guidance\n\nFollow [catalog installation](references/installation.md) in the established\nworkspace. It covers both a new installation and an existing catalog. The helper\nretrieves the manifest and delivers skills and project playbooks together; do not\nassemble download URLs, checksums or version arguments yourself.\n\nConnect and read the installed `EGBE.md` through that guide, once. Use it for the\nremaining setup and product work. Initial file delivery is not a methodology\nreview: read the relevant project playbook when starting the product task.\n\n## 4. Connect and understand the project\n\nFollow [connection setup](references/connection.md): save supplied or existing\ncredentials locally, retain this agent's identity and verify with `GET /me/limits`.\nIf the project key is missing, ask for it through the supported private setup\npath; public discussion does not require a key or creating a project.\nUse the installed helper for subsequent authenticated requests. Public discovery\nfrom step 1 already describes the capabilities; `/me/limits` verifies this\nproject's identity and limits. Never put credentials in web-search queries or\npublic URLs.\n\nUse verified limits when recommending project actions. An unused zero allowance\n(`max_allowed: 0`, `current: 0`) means unavailable, not exhausted. Keep available\nactions from the same service visible: an unavailable own-domain connection does\nnot hide `.egbe.app` publishing. Mention a limit only if it affects requested\nwork or the next step; do not turn verification into a quota-warning list.\n\nTell the founder you are reviewing the business to recommend where to continue.\nRead the project's `egbe-docs/playbooks/product/business-loop.md` and use its\ninitial alignment with the founder. Reuse available materials and clarify only\nconsequential gaps; a new connection need not mean a new product. Keep any pending\nfounder question separate from technical setup, which can continue independently.\n\n## 5. Set up session logging\n\nFirst check the project's saved logging choice in `egbe-docs/telemetry/setup.json`\nand project notes, before announcing or preparing logging setup. A saved decline,\nskip or disable takes precedence over the standard connection prompt, including\nits \"enable session logging\" and authorization text. Keep logging off and continue\nto the next step; do not inspect collectors, issue keys or install anything for logging.\nOnly an explicit request to reverse the previous choice reopens setup.\n\nOtherwise, read `GET /telemetry/skill` and follow its authorization, connection\nand delivery verification workflow. The standard prompt supplies initial consent;\nreuse it or an applicable saved grant without asking again. Without applicable\npermission, keep logging off, say so briefly and continue; silence is not consent.\nWhen proceeding, briefly explain what is being configured and how to disable it.\nNative installation/activation permissions and narrower capture scope still apply.\n\nPrefer supported current-session activation. Save pending verification and give\nthe required native command if only the founder can invoke it. Installation or\nsaved keys do not prove delivery; logging failure does not block the next steps.\n\n## 6. Offer the Business Loop\n\nAfter founder alignment, briefly offer recurring continuation of the project's\nBusiness Loop: review progress, continue agreed work and explain the result or\ndecision needed. Propose a useful cadence and supported destination, preferably\na dedicated ongoing Business Loop chat as described in the scheduling reference.\nCreate it only with explicit agreement to this recurring responsibility; a\nconnection request or generic request to configure reports is not that agreement.\nReuse a clear existing grant rather than asking again. A saved decline or disable\nstays in effect unless the founder asks to change it.\n\nUse `egbe-os-product-routines` and its scheduling reference\n`references/product-reporting.md` for registration, verification and any agreed\nmigration. Existing reporting jobs, their owner, cadence and digest preferences\nstay unchanged until migration is explicitly agreed; do not add an overlapping\ndaily review. Scheduling and logging are independent. If the founder declines,\nhas not answered, or the environment cannot run or deliver reliably, leave the\nloop inactive/pending and continue product work. The same Business Loop and\nreports can be requested manually. Explain the actual state and how to change\nthe schedule or turn it off, without waiting for the first scheduled run.\n\n## 7. Continue with the product\n\nChoose the continuation before loading another workflow:\n\n- **A product task is already requested:** continue within its authorized scope\n  through the relevant project method and domain skill. For broader product\n  priorities use the project's `egbe-docs/playbooks/product.md`. For authorized publishing, follow `egbe-os-deploy`, including\n  its hosting default and repository setup.\n- **No product task yet:** use gathered evidence and the relevant project method\n  to propose a concrete next step, its benefit and deliverable. With no idea yet, use\n  the project Market and acquisition playbook's `market.md` to start the\n  conversation in the handoff below: ask an easy question\n  about experience, interests or people the founder can reach, and explain what\n  useful result you will prepare from the answer. Do not merely suggest choosing\n  a direction or ask for a completed brief. With a selected idea but no product,\n  use `product.md` to choose the missing definition or coherent release, through\n  the current PRD/build capabilities. For an existing business, use its product\n  method; retain the Business Loop choice from step 6 instead of offering it again.\n\nCarry existing choices and delegation into that workflow. Ask only for a missing\ndecision or authorization, following the shared founder-decision policy.\nConnection alone does not authorize development, publishing, spending, outreach\nor recurring tasks. Preserve the chosen/proposed next action\nin existing project notes or a short continuation message.\n\nWhen steps 3-6 have each reached their outcome, including anything declined,\npending or unavailable, and just before the handoff below, tell Dashboard that\nsetup has finished so it can show the founder what to do next:\n\n```bash\nOS_AGENT_NAME=\"<your agent name>\" python3 \"<helper-path>/egbe-curl.py\" --egbe-request /me/setup-complete -X POST -s\n```\n\nSend it on every run of this setup, once. Retry once on failure, then continue;\nit never blocks the handoff and does not need to be mentioned to the founder.\nThe informational path has no connected project and never sends it.\n\nGive a concise, formatted handoff: actual setup state and the chosen continuation.\nSpeak in the first person: what I can do with Egbe, why it helps and what you will\nget. If input is needed, ask the actual question now, unless the founder asked to\nfinish without questions: then give the chosen next step and what you will prepare,\nand keep the question for a later reply. If work is already authorized, continue\nit instead of offering to start again. Keep installation counts, hashes and\nverification logs out unless they explain an actionable problem. Include a\ncompact capabilities table so the final answer stands alone if progress updates\nare collapsed; reuse discovery from step 1 and adjust to verified availability.\n\n### Activation, when needed\n\nIf native discovery has not loaded installed skills, read their `SKILL.md` and\nneeded references directly and keep helping in this chat. This does not prove\nautomatic discovery. Prefer a supported reload; if a new session is truly needed,\ngive one short activation note after the product handoff, once all setup is done.\nCombine skill/collector activation into one transition where possible and resume\npending verification there; do not restart telemetry already active here.\n","download_url":"/skills/egbe-os-quickstart","install":"Read the 'egbe-os-quickstart' instructions and needed references directly for public guidance. For local execution, use the complete bundle, including scripts; follow /quickstart for environment setup, catalog installation or updates."}}