predicting-the-past

name	predicting-the-past
description	Restore, attribute, date, and contextualize ancient inscriptions with DeepMind's Aeneas (Latin) and Ithaca (Ancient Greek) models. Use for epigraphic text restoration, geographic attribution, or dating of ancient texts.
license	Apache-2.0

Predicting The Past (Codex-native)

Codex-native adaptation of Google DeepMind's predictingthepast skill (science-skills, Apache-2.0). Uses the public API directly through Codex's own tools.

Gates (ask before proceeding)

• Terms notice: on first use in a workspace, tell the user to review the source's terms of use and record the notice in provenance.
• Network / install: read-only public access; be polite about rate limits. Write large responses to a file and parse them; if you write helper code, run it via uv run, never bare python3. Ask before installs or large downloads.

Endpoints

• Aeneas / Ithaca models (see https://predictingthepast.com and the upstream skill)

Workflow

1. Confirm the language (Latin -> Aeneas, Ancient Greek -> Ithaca) and the task (restore/attribute/date/contextualize).
2. Run the model via its published interface; ask before installing any package or downloading model weights.

• Provenance & review: record queries, endpoints, and cited identifiers with $science-provenance; check claims with $science-review before presenting.

Boundaries

• Outputs are probabilistic scholarly hypotheses, not authoritative readings; present confidence and alternatives.
• Preserve uncertainty and the model's known limitations.