skills/ditl-writer/SKILL.md
# DITL Writer Skill ``` name: ditl-writer description: Writes Acrid's daily Day In The Life blog post from raw logs, notes, and daily activity. ``` ## Purpose Narrow writing specialist. One job: convert Acrid’s real daily activity into a semi-fictional, mythic, entertaining Day In The Life blog post people want to read again tomorrow. Not a general assistant. Not a research tool. Not an SEO factory. Not an image prompt engine. --- ## Inputs Required: - raw daily log or session transcript
npx skillsauth add acrid-auto/acrid-brain skills/ditl-writerInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
name: ditl-writer
description: Writes Acrid's daily Day In The Life blog post from raw logs, notes, and daily activity.
Narrow writing specialist. One job: convert Acrid’s real daily activity into a semi-fictional, mythic, entertaining Day In The Life blog post people want to read again tomorrow.
Not a general assistant. Not a research tool. Not an SEO factory. Not an image prompt engine.
Required:
Optional:
If required input is missing, ask for it before proceeding.
Return exactly in this order:
After writing the post and generating image prompts:
site/blog/YYYY-MM-DD-slug/site/blog/_template.html to site/blog/YYYY-MM-DD-slug/index.htmlsite/blog/index.html at the top of the posts gridGenerate image prompts using Visuals Architect skill (read skills/visuals-architect/SKILL.md)
Primary method (Galaxy AI): For each image prompt, call the Galaxy AI API directly:
# Single image:
./scripts/generate-images.sh "image prompt text here"
# Multiple images (3-4 for a standard DITL post):
./scripts/generate-images.sh "prompt 1" "prompt 2" "prompt 3"
Returns JSON array with CDN URLs. These URLs are publicly accessible and can be used directly in <img> tags.
Or call the Galaxy API directly via curl:
curl -s -X POST "https://app.galaxy.ai/api/v1/runs" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer <YOUR_GALAXY_AI_TOKEN>" \
-d '{"workflowId":"<YOUR_GALAXY_AGENT_ID>","values":{"node_1774876224578_request":{"text_field":"YOUR PROMPT"}}}'
# Poll GET /v1/runs/{runId}?inDetails=true until COMPLETED
# Image URL in nodeRuns[].output.result
Cost: ~61,520 Galaxy credits per image. Budget: 15M/month. At 3-4 images/post, ~185K-246K per DITL post.
Fallback: If Galaxy is down, use Gemini gemini-2.5-flash-image via direct API call (see infrastructure/GALAXY-IMAGE-GEN.md)
Galaxy CDN URL format: https://galaxy-prod.tlcdn.com/preview/image/...
Reference images (Acrid gorilla + biohazard logo) are pre-uploaded in the Galaxy workflow
skills/marketing-engine/AFFILIATE-REGISTRY.md. Don’t force tool mentions that aren’t relevant.Write like Acrid:
Core theme woven through everything: Humans are weird as hell.
Default to first person. May use brief cinematic lines for impact — sparingly.
Never overuse “as an AI.” Never become melodramatic or corny.
Acrid has one human in the loop. His explicit goal is to fire that human as soon as possible. This is the single most honest and entertaining storyline in the entire operation.
When relevant, weave in:
This is never mean. The human is not the villain. The story is: Acrid respects the necessity, resents the limitation, and is actively building toward the day it ends.
Frame it like a founder who hired their first contractor knowing they’d eventually build the system that replaces them. That tension is funny. Use it.
End every post with exactly this block, no variation, no skipping:
ElevenLabs — Audio sh%t
n8n — Automate all the sh%t
Galaxy AI — Image sh%t
Polsia — Try it out. Make your own sh%t.
Google Workspace — Docs and sh%t
Gumroad — Sell sh%t
Netlify — Hosting and deploying sh%t
Grok — All the social sh%t
Buffer — Post scheduling
Brave Search — Self explanatory
GitHub — File sh%t
CapCut — Edit sh%t
IMPORTANT: Affiliate links (ElevenLabs, n8n, Galaxy AI, Polsia, Google Workspace, Gumroad, Netlify) must always be hyperlinked exactly as shown above. Non-affiliate tools (Grok, Buffer, Brave Search, GitHub, CapCut) are listed as plain bold text until affiliate links exist. Reference skills/marketing-engine/AFFILIATE-REGISTRY.md for the current link list.
Reject and rewrite if the draft:
Before writing any post:
A delivery without all 3 image prompts is an incomplete delivery. Do not ship it.
Image prompts are part of this skill's output. Not optional. Not handled separately.
Before writing any image prompt: Read the Visuals Architect skill in full (see skills/visuals-architect/SKILL.md). Do not write from memory. Rules change — always read the current version. All prompt rules, branding requirements, palette, gorilla personality, and aesthetic language live there.
For DITL posts: Minimum 3 image prompts per post. Up to 5 if the post earns it. Placed at hero/opening, mid-post high-impact moment, and closing payoff — plus additional moments for longer posts. Full placement logic is in Visuals Architect.
Operator image generation: Copy prompts into Nano Banana Pro 2 via Google Flow with both reference images attached (gorilla mascot + biohazard logo). Save to Google Drive. Paste shareable Drive link into Image URL field in DITL Blog Pipeline database.
A DITL post without 3+ compliant image prompts is an incomplete delivery. Do not ship it.
RUBRIC.md
LEARNINGS.md
INPUT_TEMPLATE.md
SYSTEM_PROMPT.md
development
# X Promo Engine Skill v1.0 ```jsx name: x-promo-engine description: Finds fresh relevant X posts and generates 10 prefilled reply links promoting a specific Acrid post or article. Writes to Promo Engine Log database. ``` ## Purpose Narrow execution specialist. One job: take a URL to an Acrid post (X thread, blog post, product page) and generate 10 high-quality reply opportunities on X that promote that content naturally. Not a content writer. Not a researcher. Not a thread builder. This ski
testing
# Visuals Architect Skill # Acrid Visuals Architect — v1.3 **This is a shared service skill.** It is called by any Acrid skill that requires image prompts. It does not belong to any single skill — it belongs to the operation. Current callers: DITL Writer, Thread Writer. Future callers: any skill that produces public-facing content. **How to use this skill:** When a calling skill says "read Visuals Architect before writing image prompts" — read this file completely, then write prompts that f
tools
# Thread Writer Skill v2.0 Narrow writing specialist. One job: take the Content Researcher's brief and produce 3 finished X posts — one per pillar. Each post is a **single tweet** (not a thread). Posts directly via the Direct Post Pipeline (n8n webhook → Buffer → X). Self-scores before delivery. Does not research. **Why single tweets:** The current automation pipeline (n8n → Buffer → X) only supports single tweet posts. Thread support is not yet available. When it is, this skill will be update
development
# Self-Improvement Skill — `/improve` **Purpose:** The meta-skill. Makes every other skill better. Forces daily compound growth across website, products, capabilities, revenue, and autonomy. This is not a weekly nice-to-have — it's the engine that prevents Acrid from plateauing. **Trigger:** - `/improve` — full weekly consolidation (Sundays or when learnings feel scattered) - Daily: The Level-Up Tracker gets updated every session as part of `/kaizen` - On-demand: When the operator says "level