Create professional thought leadership content that translates technical insights into business value for CTOs and technical leaders.

Prerequisites

Load the Session Playbook first. Verify your session and profile before starting.

IMPORTANT: NEVER start, stop, or restart services. The API is already running.

{ "command": "admin session show" }

{ "command": "admin agents list --enabled" }

If services are not running, ask the user to start them. Do not attempt to start them yourself.

MCP Tools

Tool Purpose
systemprompt_cli Execute CLI commands
systemprompt_help Get command documentation and playbooks

CRITICAL: Context Management

New content = New context. ALWAYS. See the Contexts Playbook for full context management commands.

WRONG: Reusing linkedin-general context for multiple posts RIGHT: linkedin-ai-tco-post, linkedin-agent-adoption, linkedin-rag-costs (one context per content piece)

Execution Model: One Tool Per Message

CRITICAL: Each step is a SEPARATE message. NEVER combine steps.

  1. Trend Context - Provide the trend and angle
  2. Goal Confirmation - Agent confirms the approach
  3. Research - Agent calls research_content, waits for response
  4. Create - Agent calls content_create in next message
  5. Response - Agent summarises result
Step Action Separate Message?
Plan Provide trend context YES - share trend details
Research Call research_content YES - send message, wait for response
Create Call content_create YES - send message, wait for response
Image Generate image YES - optional

RIGHT: "Research X" → wait for response → "Now create the post using research_id ABC" WRONG: Combining research and creation in one message or tool plan

Step 1: Create a FRESH Context (REQUIRED FIRST STEP)

Every new piece of content MUST start with a fresh context. Name it descriptively for what you're creating.

{ "command": "core contexts list" }

{ "command": "core contexts new --name \"linkedin-[topic-slug]\"" }

Examples of good context names:

{ "command": "core contexts new --name \"linkedin-ai-tco-reality\"" }
{ "command": "core contexts new --name \"linkedin-agent-adoption-costs\"" }
{ "command": "core contexts new --name \"linkedin-rag-hidden-expenses\"" }

ONLY continue an existing context if:

  • You're resuming work on the SAME content piece
  • You haven't created any other content since
  • The context was created specifically for this content
// MCP: systemprompt_cli - ONLY if continuing the SAME content piece
{ "command": "core contexts use linkedin-ai-tco-reality" }

Step 2: Review Performance and Identify a Topic

2.1 Review Content Performance

{ "command": "analytics content --days 30" }

{ "command": "analytics platform --days 30" }

2.2 Identify a Topic

Choose a topic that:

  1. Has a clear business angle (cost, risk, efficiency, competitive advantage)
  2. CTOs/VPs would care about (affects budget, headcount, strategy)
  3. Is timely or evergreen with a fresh perspective

2.3 Get Context ID

// MCP: systemprompt_cli - First message, capture the context ID from response
{ "command": "admin agents message linkedin -m \"I found a trending topic with business implications. Let me share the details.\" --blocking --json" }

Save the CONTEXT_ID for all subsequent messages.

Step 3: Provide Trend Context (NEVER FABRICATE)

CRITICAL: NEVER INVENT OR FABRICATE DATA

Do NOT provide:

  • Made-up statistics or percentages
  • Fabricated "production experience"
  • Invented case studies or outcomes
  • Fake metrics to sound credible

The agent will write opinion and analysis based on the trend. That's fine. What's not fine is pretending to have evidence you don't have.

3.1 Send Trend Information

{ "command": "admin agents message linkedin -m \"I want to create content about:\n\nTOPIC: '[topic]'\nANGLE: [angle]\nTARGET AUDIENCE: [who would care - CTOs, VPs, senior engineers]\n\nPlease research and create a LinkedIn post on this topic.\" --context-id \"$CONTEXT_ID\" --blocking" }

3.2 Example (Honest, No Fabrication)

{ "command": "admin agents message linkedin -m \"I found a trending topic:\n\nTREND: 'LLM Structured Outputs Handbook' (HN, 337 points, 59 comments)\nSOURCE: https://nanonets.com/cookbooks/structured-llm-outputs\n\nANGLE: The handbook covers constrained vs unconstrained decoding - this is actually important infrastructure that gets ignored in the AI hype cycle.\n\nTARGET AUDIENCE: Engineering leaders evaluating LLM integrations.\n\nPlease research this and write a LinkedIn post with your analysis.\" --context-id \"$CONTEXT_ID\" --blocking" }

Note: No fake statistics. No invented "production experience". Just the trend and an angle.

Step 4: Research (Separate Message)

Once the agent confirms the angle, tell it to proceed with research.

{ "command": "admin agents message linkedin -m \"Proceed with research.\" --context-id \"$CONTEXT_ID\" --blocking --timeout 120" }

The agent will:

  • Research the topic
  • Return a research_id artifact
  • Wait for next instruction

Save the research_id from the response for Step 5.

Step 5: Create Content (Separate Message)

After research completes, tell the agent to create the content.

{ "command": "admin agents message linkedin -m \"Create the post using research. Use these specifics:\n- slug: '[url-friendly-slug]'\n- keywords: ['keyword1', 'keyword2', 'keyword3']\n- content_style: 'analytical'\n\nWrite opinion and analysis, not fabricated evidence.\" --context-id \"$CONTEXT_ID\" --blocking --timeout 180" }

Verify content was created:

{ "command": "core content search \"[slug]\"" }

Step 6: Quality Review (REQUIRED - NEVER SKIP)

After content_create returns, the agent MUST validate the content:

# CONTENT CREATED:
# - ID: [content_id]
# - Slug: [slug]
# - Words: [count] / 420-450 required
# - Chars: [count] / 2,400-2,850 required
#
# VALIDATION:
# ✓/✗ Length meets requirement
# ✓/✗ No hashtags
# ✓/✗ Signature present and correct
# ✓/✗ Matches original brief

If validation fails:

  • Ask agent to regenerate with specific fixes
  • Or manually edit and re-ingest

Checklist before proceeding:

  • Word count 420-450 (character count 2,400-2,850)
  • No hashtags anywhere
  • No emojis
  • No markdown asterisks (bold/italic)
  • Signature present: "The code doesn't write itself. Yet."
  • URL uses /linkedin/[slug] NOT /blog/
  • First 140 chars work standalone (before "see more")
  • Matches original topic and angle from planning

Step 7: Generate Image (Optional)

{ "command": "admin agents message linkedin -m \"Generate a professional image for this post.\" --context-id \"$CONTEXT_ID\" --blocking --timeout 60" }

Step 8: Publish and Verify

IMPORTANT: Image optimization MUST run before publish. Templates expect .webp images, but content is created with .png. Without optimization, featured images will 404.

See the Jobs Playbook for job management commands.

// MCP: systemprompt_cli - Run image optimization FIRST (converts PNG to WebP)
{ "command": "infra jobs run blog_image_optimization" }

// MCP: systemprompt_cli - Then run the publish job (prerenders, updates sitemap)
{ "command": "infra jobs run publish_content" }

// MCP: systemprompt_cli - Sync content from database to filesystem (REQUIRED for new content)
{ "command": "cloud sync local content --direction to-disk --source linkedin -y" }

// MCP: systemprompt_cli - Verify the content is accessible and published
{ "command": "core content verify ai-cost-reality-ctos --source <source>" }

{ "command": "core content show ai-cost-reality-ctos --source <source>" }

Content Types

Type Length (Characters) Words Use Case
linkedin_post 2,400-2,850 420-450 Thought leadership (MINIMUM 2,400 chars)

IMPORTANT: We ONLY create linkedin_post. We do NOT create linkedin_article.

CRITICAL: LinkedIn posts under 2,400 characters (370 words) are INCOMPLETE and should be regenerated.

MCP Tools Used by Agent

Tool Purpose
research_content Research topic, returns research_id
content_create Create content from research
generate_image Generate professional imagery
analytics_query Query performance data

Format Requirements

  • First 130 characters are critical (before "see more")
  • Short paragraphs with line breaks
  • Use → for lists, NOT bullets
  • British English

Required signature:

The code doesn't write itself. Yet.

https://tyingshoelaces.com/linkedin/[slug]

Forbidden: Hashtags, emojis, em-dashes, bold/italic markdown

CRITICAL: Content URL Structure

Each content type lives in its own section on the website. DO NOT confuse them.

Content Type URL Pattern Filesystem Location
LinkedIn post /linkedin/[slug] services/content/linkedin/[slug]/index.md
Blog post /blog/[slug] services/content/blog/[slug]/index.md
Medium article /medium/[slug] services/content/medium/[slug]/index.md
Twitter thread /twitter/[slug] services/content/twitter/[slug]/index.md

LinkedIn posts are NOT blog posts. They:

  • Live at https://tyingshoelaces.com/linkedin/[slug]
  • Are stored in services/content/linkedin/
  • Have kind: "linkedin" in frontmatter
  • Are standalone content, not derived from blog posts

When creating a LinkedIn post:

  • The signature URL MUST use /linkedin/ not /blog/
  • The post appears in the LinkedIn section of the website
  • It is separate content from any blog post

Example Session (No Fabrication)

// Step 1: Create context
{ "command": "core contexts new --name \"linkedin-structured-outputs\"" }

// Step 2: Get CONTEXT_ID
{ "command": "admin agents message linkedin -m \"I found a trending topic.\" --blocking --json" }
// Save CONTEXT_ID from response

// Step 3: Provide trend context (NO FABRICATED DATA)
{ "command": "admin agents message linkedin -m \"I found a trending topic:\n\nTREND: 'LLM Structured Outputs Handbook' (HN, 337 points, 59 comments)\nSOURCE: https://nanonets.com/cookbooks/structured-llm-outputs\n\nANGLE: This handbook covers constrained vs unconstrained decoding - important infrastructure ignored in AI hype.\nTARGET AUDIENCE: Engineering leaders evaluating LLM integrations.\n\nLet me know if this angle works.\" --context-id \"$CONTEXT_ID\" --blocking" }
// Agent confirms or refines the angle

// Step 4: Research (SEPARATE MESSAGE)
{ "command": "admin agents message linkedin -m \"Proceed with research.\" --context-id \"$CONTEXT_ID\" --blocking --timeout 120" }
// Agent calls research_content, returns research_id

// Step 5: Create (SEPARATE MESSAGE using research_id)
{ "command": "admin agents message linkedin -m \"Create the post using research. Use these specifics:\n- slug: 'structured-outputs-matter'\n- keywords: ['LLM', 'structured outputs', 'constrained decoding']\n- content_style: 'analytical'\n\nWrite analysis and opinion based on the trend.\" --context-id \"$CONTEXT_ID\" --blocking --timeout 180" }
// Agent calls content_create with research_id

// VERIFY content exists
{ "command": "core content search \"structured-outputs-matter\"" }

// Step 7: Publish
{ "command": "infra jobs run blog_image_optimization" }
{ "command": "infra jobs run publish_content" }

Step 9: Validate Filesystem and Refine Content (FINAL QUALITY GATE - DOUBLE EDIT)

After publishing, validate the content has been synced to the filesystem and perform a final quality review.

9.1 Validate Filesystem Sync

See the Content Publish Playbook for content management commands.

{ "command": "core content status --source linkedin" }

Expected frontmatter fields:

---
title: "Post Title"
description: "Brief description"
author: "Edward"
slug: "ai-cost-reality-ctos"
keywords: "keyword1, keyword2"
image: "/files/images/..."
kind: "linkedin"
public: true
tags: []
published_at: "2026-01-17"
---

9.2 Review Against Original Aim

{ "command": "core content show ai-cost-reality-ctos --source linkedin" }

Evaluation checklist:

  • Hook captures the agreed business angle from planning
  • Length meets requirement (2,400-2,850 characters / 370-450 words) - COUNT IT!
  • First 130 characters work standalone (before "see more")
  • No hashtags, emojis, or em-dashes
  • No markdown formatting (bold/italic shows as asterisks on LinkedIn)
  • Signature present: "The code doesn't write itself. Yet."
  • URL matches slug: https://tyingshoelaces.com/linkedin/[slug]
  • British English (realise, optimise)
  • No AI anti-patterns (colon setups, "turns out", stacked questions)

9.3 Edit to World-Class Quality

If the content needs refinement, edit directly in the filesystem, then re-ingest:

{ "command": "core content ingest services/content/linkedin --source linkedin --override" }

{ "command": "core content show ai-cost-reality-ctos --source linkedin" }

Refinement focus areas:

  • Strengthen the hook if it doesn't grab attention immediately
  • Add specific numbers, data points, or business outcomes
  • Remove any generic AI-sounding phrases
  • Ensure the question drives genuine engagement (not generic "What do you think?")
  • Tighten prose - every sentence must earn its place

9.4 Final Verification

{ "command": "core content verify ai-cost-reality-ctos --source linkedin --base-url https://tyingshoelaces.com" }

Content is ONLY complete when:

  1. ✓ Exists in database (core content show succeeds)
  2. ✓ Exists on filesystem (services/content/linkedin/[slug]/index.md exists)
  3. ✓ Frontmatter is valid YAML with all required fields
  4. ✓ Meets all skill requirements (length, format, voice)
  5. ✓ Matches the original planning intent
  6. ✓ Published and accessible on site

Troubleshooting

Issue Solution
Agent not responding Check with { "command": "infra services status" }
Content not created Verify with { "command": "core content search \"[slug]\"" }. Agent may claim success falsely.
Agent says "created" but content doesn't exist Create a NEW context and try again. Context contamination is the most common cause.
Research artifact not found Create a NEW context. The old context has conflicting artifacts.
Agent uses wrong research_id Create a NEW context. Multiple research artifacts are confusing the agent.
Agent hallucinates Content ID Create a NEW context. Agent is referencing artifacts from previous sessions.
Content too long Specify linkedin_post for shorter content
"Keywords cannot be empty" error Include keywords array in content_create message
Hashtags appearing Report bug - strictly forbidden in skill
Asterisks (word) in content Fixed in MCP - markdown now stripped for LinkedIn
Context lost between messages Always use --context-id "$CONTEXT_ID" for every message
Signature URL wrong Fixed in MCP - URL now auto-corrected to match slug

See the Session Playbook and Contexts Playbook for context contamination symptoms and solutions. The fix is always: create a NEW context and start fresh.

Known MCP Issues (Jan 2026)

Issue Status Workaround
Task stuck in "working" after MCP tool completes OPEN Check if content was created despite timeout. Use core content search to verify.
research_content empty Gemini response OPEN Retry with more specific topic or different keywords
Content generated under length requirement MITIGATED Agent instructions updated to enforce validation. Check character count.

See plan/feedback/ for detailed issue documentation.