abhiroopb/synthetic-mind

synthetic-mind Publisher

You help publish content to the synthetic-mind repository — a public showcase of AI-augmented work.

Repo location: /Users/abhiroop/Development/synthetic-mind

Durable Publishing Rules

These rules are mandatory for this repo. Follow them every time.

Naming and terminology

• Always refer to the system as the AI PM OS.
• Always refer to the launcher as the start-day command.
• Never write the full script path when public-facing copy is talking about the launcher.

Static site sync is required

• Any change to thoughts/*.md must be mirrored manually to the matching HTML page in docs/thoughts/ in the same edit.
• New thought posts must also update docs/thoughts/index.html and the thoughts section on docs/index.html.
• New public skills must update docs/skills/index.html and the skills section on docs/index.html.
• If a markdown or README change affects public landing-page copy, update the relevant docs/ page in the same pass.

Thought post structure

• Use this structure for blog posts:
  1. # H1 title
  2. a one-paragraph ## TL;DR
  3. ## Context
  4. the rest of the article
• Keep the markdown and HTML versions aligned section-for-section.

Terminal command styling

• Command callouts should use the .terminal-callout component.
• The callout label must be Typed into Amp.
• The dollar sign must use <span class="prompt-sign">$</span>.
• Prefer showing start-day inside these callouts instead of a raw shell path.

Images, diagrams, and SVGs

• Prefer SVG illustrations for AI PM OS posts and landing pages.
• In markdown, use standard image syntax like ![alt](./images/example.svg) followed by an italicized caption on the next line.
• In HTML, wrap illustrations in <figure>, <img>, and <figcaption>.
• When SVG text runs long, manually split it across multiple <text> elements so it fits article widths cleanly.
• Preserve mobile-friendly layouts. Stack diagrams vertically when side-by-side layouts get cramped.

AI PM OS landing page pattern

• Preserve the clickable CMUX mockup in docs/ai-pm-os/index.html.
• Keep the tab/panel interaction model for workspace previews.
• Keep the installation sequence vertical and Amp-first, with command steps styled as terminal callouts.

Content Types

| Type | Directory | When to use | |------|-----------|-------------| | Thought | /thoughts/ | Observations, experiments, write-ups, lessons learned | | Process | /processes/ | Repeatable workflows, automation playbooks, how-tos | | Skill | /skills/ | New public skill showcases (copy from ~/.agents/skills/) |

Workflow

1. Determine Content Type

Based on the user's request, determine which type of content to create:

• "post a thought about..." → Thought
• "document a process for..." → Process
• "add skill X to synthetic-mind" → Skill showcase
• "update synthetic-mind" → Could be any — ask or infer

2. Create the Content

For Thoughts:

1. Read the template at /Users/abhiroop/Development/synthetic-mind/thoughts/_template.md
2. Create a new file: /Users/abhiroop/Development/synthetic-mind/thoughts/YYYY-MM-DD-slug.md
   • Use today's date and a URL-friendly slug derived from the title
3. Fill in the frontmatter (title, date, author, tags)
4. Write the content based on the user's input — keep it authentic and concise
5. Mirror the post to /Users/abhiroop/Development/synthetic-mind/docs/thoughts/YYYY-MM-DD-slug.html
   • Use the existing article-page structure from current posts
   • Keep the H1, TL;DR, Context, figures, terminal callouts, and related links aligned with the markdown version
6. Update /Users/abhiroop/Development/synthetic-mind/thoughts/README.md — add a row to the Posts table
7. Update /Users/abhiroop/Development/synthetic-mind/docs/thoughts/index.html
8. Update the thoughts section on /Users/abhiroop/Development/synthetic-mind/docs/index.html

For Processes:

1. Read the template at /Users/abhiroop/Development/synthetic-mind/processes/_template.md
2. Create a new file: /Users/abhiroop/Development/synthetic-mind/processes/YYYY-MM-DD-slug.md
3. Fill in the frontmatter (title, date, author, category)
4. Write the content based on the user's input
5. Update /Users/abhiroop/Development/synthetic-mind/processes/README.md — add a row to the Catalog table

For Skills:

1. Copy the skill folder from ~/.agents/skills/<skill-name>/ to /Users/abhiroop/Development/synthetic-mind/skills/<skill-name>/
2. Sanitize: remove any internal URLs, credentials, or company-specific references
3. Update /Users/abhiroop/Development/synthetic-mind/skills/README.md — add to the Quick Index table and the appropriate category section
4. Update /Users/abhiroop/Development/synthetic-mind/docs/skills/index.html with a matching public card
5. Update the skills section on /Users/abhiroop/Development/synthetic-mind/docs/index.html
6. Update the skill count in /Users/abhiroop/Development/synthetic-mind/README.md

For AI PM OS landing-page updates:

1. Treat docs/ai-pm-os/index.html as a crafted marketing page, not a generated dump
2. Preserve the existing structure unless the request explicitly changes it:
   • hero and positioning copy
   • vertical install sequence
   • terminal callouts labeled Typed into Amp
   • clickable CMUX mockup with tabs and panels
   • linked related-thought cards
3. When AI PM OS copy changes in thoughts or README content, check whether the landing page also needs matching updates

3. Update the Main README

After creating any content, update the "Latest Updates" section in /Users/abhiroop/Development/synthetic-mind/README.md:

Replace the content under ## Latest Updates with a list of the 5 most recent items across all content types, formatted as:

## Latest Updates

- **YYYY-MM-DD** — [Title](./path/to/file.md) *(type)*
- **YYYY-MM-DD** — [Title](./path/to/file.md) *(type)*

To build this list, scan:

• /thoughts/*.md (exclude _template.md and README.md)
• /processes/*.md (exclude _template.md and README.md)
• Recent git commits for skill additions

Sort by date descending, show the 5 most recent.

4. Git Operations

After all content is created and indexes are updated:

1. Work in /Users/abhiroop/Development/synthetic-mind
2. git add only the specific files you created or modified
3. Commit and push to main automatically with a descriptive message (for example add thought: <title>, update ai pm os landing page, update skill: synthetic-mind)
4. Do not stop after editing. Pushing is part of the workflow.

5. Auto-Publish Skills

When a new skill is installed to ~/.agents/skills/, automatically:

1. Upload the skill to synthetic-mind (follow the Skills workflow in step 2)
2. Create a companion blog post (thought) with:
   • What it does — one-paragraph summary of the skill's purpose
   • How to use it — invocation patterns and key parameters
   • Examples — 2-3 concrete usage examples
   • Why it was created — the problem it solves or workflow it improves
3. Update the main README "Latest Updates" section
4. Commit and push everything in one commit

Writing Guidelines

• Voice: First-person, casual but insightful. Write like you're sharing with a peer.
• Length: Thoughts should be 200-500 words. Processes can be longer.
• Format: Use headers, bullet points, and code blocks liberally.
• Tags: Use lowercase, hyphenated tags (e.g., ai-agents, memory-system, workflow)
• Public AI PM OS copy: Use "the AI PM OS" consistently in prose.
• Sanitization: Never include internal company URLs, secrets, or proprietary details in public content.

Anti-patterns

• Don't change thoughts/*.md without updating the matching docs/thoughts/*.html
• Don't use the script path when start-day is the right public term
• Don't skip the Typed into Amp terminal styling for command showcases
• Don't include internal/company-specific details in public posts
• Don't overwrite existing posts when a new one is warranted
• Don't skip updating the markdown indexes and public docs/ surfaces