SymbiCorAISecond Brain · alpha.13 pilot

Symbiotic Second Brain

Build a second brain.
Develop as you work.

Bring the sources, decisions and next actions for one responsibility together. Produce useful work while practicing how to direct, inspect and improve AI-supported work.

Start with ChatGPTStart with ClaudeSave the link for later

On your phone? Save this link and continue on a laptop. Opening the page does not install a skill, connect an account or upload information.

Your work and your development, side by side

Illustrative computer mockup showing an AI conversation beside a Second Brain work dashboard with a separate development tab.
Illustrative mockup, not a live product screenshot. Your conversation happens in ChatGPT or Claude. The generated dashboard is a saved, read-only view; its illustrated chat and navigation controls are not features of the static dashboard. Review in your chat, then open the refreshed dashboard. Tap to enlarge.

Setup notes

Both downloads use the same current shared skill and dashboard renderer: 0.4.0-alpha.13 · checked October 15, 2026. Native ChatGPT and Claude setup, Work and Development views, progress review with a refreshed dashboard, and fresh-chat recovery have been tested in Allan’s accounts. This supports a facilitated pilot, not a guarantee for every account or device. Chat-generated PDFs have been tested in ChatGPT; the actual browser Print / save PDF output and Claude-specific PDF export still need checking.

Use your own responsibility and information you are permitted to share. There is no preloaded example brain. Uploading the portable ZIP is not native installation; file and code tools depend on your account. Progress reviews use copy and paste or a direct chat request. The hosted integration remains paused.

Download ChatGPT plugin · alpha.13Download Claude skill · alpha.13Download portable kit

One responsibility. Two separate views.

Work view: sources, decisions and next actions. Development view: formulate, inspect and correct. Human review governs both.

Chat Image illustration. Tap to enlarge.

Work view

Inspect the useful result, dated sources, draft or accepted status, decisions and next action. The saved work record—not the dashboard—is authoritative.

Development view

See optional next-practice guidance. Retain observations only with separate consent. A journal records what you inspected or corrected; it does not score mastery.

The assistant generates the companion from your actual saved records. Open the generated HTML to inspect both views. Manual import into this website is not the normal workflow.

Start in an approved ChatGPT Work conversation.

  1. Download the complete ZIP. Start a new conversation with file and code tools, and attach the ZIP.
  2. Send the request below. Agree one responsibility, its owner, useful outcome, acceptance check and save location.
  3. Inspect and correct the draft. Ask for the saved work JSON and companion generated with the bundled renderer.
  4. Open that HTML. Work opens first. Choose Development for guidance; authorize separate learning retention/display only if you want it.
  5. Download work JSON and companion to a folder you control. For a fresh chat, attach saved work and this kit. Attach learning separately only if you want it included.
Open the ZIP. Read ChatGPT-Startup.md and skills/symbiotic-second-brain/SKILL.md. Create second brain for xyz. Ask me what responsibility xyz should cover, then guide me one step at a time using only information I supply or authorize.

Resume without starting over.

Continue the attached existing brain. Read the saved record, identify draft versus accepted state and the next action, then regenerate the companion. Do not start over.

Generated conversation files may be temporary. They are not automatically saved to your computer or Google Drive. Keep your downloads. If uploads or code tools are unavailable, work through your own responsibility on paper or in an approved chat. You can still work through the responsibility in an approved chat without a dashboard.

Native plugin, local and Claude routes

The optional ChatGPT plugin ZIP contains the same skill and renderer. Native installation requires your workspace’s permission; downloading it grants no access. Once installed, invoke it rather than attach startup instructions again.

The Claude skill archive contains the same generic source. Initial native Claude setup has been checked; that does not establish compatibility across all accounts. Do not assume parity or bypass organization controls.

Full portable/local setup instructions

Claude: laptop setup path

The Claude skill uses the same current shared source as the ChatGPT plugin. Native setup, both views, review-to-dashboard refresh and fresh-chat recovery passed in Allan’s account. Your organization must permit custom skills and code execution; verify the same steps in your own environment.

  1. Download the Claude skill archive · alpha.13. Review it before enabling it.
  2. On your laptop, open Claude in your approved account. Go to Customize → Skills → Add skill → Upload skill. Upload the archive and turn the skill on.
  3. If the uploader rejects the .skill extension, use the same archive with a .zip filename. Do not bypass administrator restrictions.
  4. Start a fresh chat, ask Claude to use Symbiotic Second Brain, and send create second brain for xyz. Replace xyz with the responsibility you want to work on.
  5. Check the generated Work and Development views, download your records, then verify you can resume in a fresh chat. Stop and report any missing step rather than assume success.

Claude skills require code execution and may be restricted by your organization. Official Claude setup instructions. SSB’s complete iPhone setup remains unverified. Use a laptop for setup.

Keep human review part of the work

The human chooses purpose, scope, authority and acceptance. AI works within those limits. The reviewed record supplies the next cycle’s starting point.

Ask AI to explain its evidence and uncertainty. Inspect the draft before accepting it, and save the reviewed state for your next session.

Your conversation remains the place to work.

The companion HTML is read-only. It does not contain a chatbot, write records, synchronize devices or call an evaluation service. Prepare progress review prepares text only. Copy request → paste into your Second Brain chat → review the results. The button does not run AI or save an assessment. You can instead type “Review my development and recommend what to practice next” directly in your existing brain chat. Review is optional, and the skill saves a new report and regenerates the dashboard only when your consent and its tools permit. No hosted service or new account is required; the hosted integration is paused. A new saved checkpoint produces a new companion.

No private connection, credentials, personal brains or client records are included. The older manual-import dashboard remains a legacy fallback, not the normal companion route.

Review your progress when it helps

Progress review is optional. Choose Prepare progress review, then Copy request → paste into your Second Brain chat → review the results. Or type “Review my development and recommend what to practice next” directly in your existing brain chat.

After a consented review, the skill should save a new report and give you a refreshed dashboard with the results and next-practice recommendations. Open the new file; an older HTML page does not refresh itself. If saving or regeneration is unavailable, the assistant must explain what remains incomplete. Useful work does not depend on retaining a development record.

Print or save either view as a PDF

In the companion, choose Work only, Development only, or Work and Development, then Print / save PDF. Select Save as PDF in the browser’s print dialog. If a preview blocks printing, download the HTML and open it in your browser.

Development can include private observations. Use Work only when sharing work without personal learning. The print button does not itself confirm that a PDF was saved. The hosted integration is paused; no additional service account is required.

Keep private work private.

This public website does not receive your work or learning records. Downloaded HTML embeds its source records and has no authentication or encryption. Do not share it with someone who should not see them. Upload only permitted information to your approved AI provider. Work retention is not permission to retain a learning journal.

Use only information you are permitted to share with your chosen AI provider. Your organization controls accounts, connectors and storage.

Learn through the work you already do

Practice cycle: direct one task and inspect the answer; select sources, purpose and limits; maintain decisions and next actions; compose tested skills and review the result.
Practice approach: begin with one task, improve the context and record, then combine methods you have tested. This loop describes practice, not a measurement of competence. Tap to enlarge.