Demos
WorldFork has two first-class live workflows: the full runtime smoke and the Atlas onboarding demo.
Full Runtime Smoke
worldfork smoke live
This is the end-to-end health check for a configured local backend. It uses real API credits and verifies that audited LLM calls match the effective routes reported by worldfork settings llm. A common route policy is:
openrouter/deepseek/deepseek-v4-flash
openai-codex/gpt-5.4
The smoke validates:
readiness checks
model configuration
settings patch, reread, and restoration
Big Bang pause/resume behavior
root and branch tick execution
manual branch intervention
runtime checkpoints and node attempts
job pause and synchronous run
multiverse reports, final reports, Markdown render, and PDF render
logs and final readiness
Atlas Onboarding
worldfork demo atlas
Atlas reads examples/test-big-bang.md, creates the Atlas Resilience Crisis Big Bang, runs a root timeline, creates a manual transparency branch, allows God-agent-created branches under generous caps, drains every discovered timeline to terminal state, generates per-multiverse reports, generates a final cross-multiverse report, can render a PDF on request, and audits the configured OpenRouter DeepSeek Flash plus OpenAI Codex route split.
While Atlas runs, the same runtime phases described in Runtime Walkthrough apply: Big Bang initialization, checkpointed tick execution, bounded parallel cohort decisions, aggregate event summary, God-agent review with audited JSON tool calls, endpoint-ledger updates, branching, report generation, and cost/timing collection.
Default Atlas sizing:
Setting |
Default |
|---|---|
Tick duration |
720 minutes |
Horizon |
30 simulated days |
Derived terminal tick index |
60 |
Active multiverse cap |
64 |
Branch depth cap |
8 |
Branches per tick cap |
8 |
Completion request cap |
1000 |
Override defaults when you need a shorter or larger demonstration:
worldfork demo atlas \
--tick-duration-minutes 720 \
--horizon-days 7 \
--max-active-multiverses 16 \
--max-branch-depth 4 \
--max-branches-per-tick 3 \
--completion-max-requests 300
Reading Demo Output
At completion, Atlas prints:
big_bang_idroot_multiverse_idchild_multiverse_idterminal multiverse count
final report version ID
audited LLM call count
follow-up
worldfork reports viewcommandfollow-up
worldfork reports render --outputcommandfollow-up
worldfork watchcommand
Useful follow-up inspection commands:
worldfork runs workspace <big-bang-id>
worldfork runs cost <big-bang-id> --include-calls
worldfork runs estimate <big-bang-id>
worldfork ledgers list <big-bang-id>
worldfork ledgers path-mass <big-bang-id>
worldfork reports pack <big-bang-id> --mode summary
worldfork reports adjudication <big-bang-id>
worldfork logs list --status failed
Use the final report command first:
worldfork reports view <report-version-id>
Then render PDF only when you need a file artifact:
worldfork reports render <report-version-id> --format pdf --output report.pdf