VS
ViralScript AI AIOS
AIOS Runtime: Ready ⚡
skills/2 loaded · config/skills.json

Technical content,
rewritten to land.

ViralScript AI reads an article, a repository, or raw markdown and returns a five-post educational thread. Character caps enforced, claims verified, nothing invented.

Open Workspace View Source

Runtime

~/ .viralscript-ai — boot sequence

aios — runtime
$ aios boot --config ~/.viralscript-ai/config/skills.json
✓ memory/context.json loaded
✓ skills/ scanned 2 manifests found
✓ social_thread_writer ready 5-post pipeline
✓ humanizer ready 26 patterns · v3.1.0
✓ platform adapters x=280ch · linkedin=800ch
✓ accuracy gate armed
 
▸ runtime status: READY
social_thread_writer v1 · active humanizer v3.1.0 · MIT · active x_adapter 280 char cap linkedin_adapter 800 char cap

Workspace

input → generate → 5 posts

0 / 0carriage returns are not counted. Caps apply per post, not per thread.
The live generator runs inside your agent via the social_thread_writer skill. This page previews the output contract and the enforcement.

Thread Preview

sample output cap 280 / post
5 posts · counters recompute on platform change

Registry

config/skills.json

skills.json 
{
  "skills": [
    {
      "name": "social_thread_writer",
      "path": "skills/social_thread_writer.md",
      "platforms": ["X", "LinkedIn"],
      "languages": ["English", "Albanian"],
      "enabled": true
    },
    {
      "name": "humanizer",
      "path": "skills/humanizer.md",
      "license": "MIT",
      "version": "3.1.0",
      "source": "https://github.com/blader/humanizer",
      "enabled": true
    }
  ]
}

What the runtime guarantees

non-negotiables, enforced before output

01No fabrication

If extraction fails, is blocked, or hits a paywall, the run stops and asks for raw text. Unreachable sources never become plausible summaries.

02Verified metrics only

Every number in a post must appear in the source. No rounded figures, no recalled benchmarks, no invented speedups.

03No outcome promises

No guaranteed career results and no fixed learning timelines. Posts 1 to 4 carry no promotion at all.

04Terms kept exact

Protocols and algorithms keep their industry name, followed by a plain-language gloss the reader can use immediately.