OpenMath Submit Theorem

Instructions

Use this skill for the two-stage Shentu proof submission flow.

• Stage 1 submits a proof hash as a commitment so others can see you already know the answer without learning the proof details.
• Stage 2 submits the proof detail to reveal and verify the same proof on-chain.
• This is an operational skill, not an instruction-only note: it shells out to trusted local python3 and shentud, reads shared openmath-env.json, queries/broadcasts to a Shentu RPC endpoint, and uses the local OS keyring for signing flows.
• Default: authz + feegrant from prover_address (the user's OpenMath Wallet Address).
• Shared config resolution order: --config <path> → OPENMATH_ENV_CONFIG → ./.openmath-skills/openmath-env.json → ~/.openmath-skills/openmath-env.json. If OPENMATH_ENV_CONFIG is set, treat it as the selected config path. If that file is missing or invalid, stop and fix it instead of silently falling back.
• Shentu chain/RPC settings come from SHENTU_CHAIN_ID and SHENTU_NODE_URL or built-in defaults, not from openmath-env.json.
• The skill always uses --keyring-backend os for local key lookups and generated submission commands.
• Direct signer fallback: generate_submission.py --mode direct.
• Required local dependencies are python3 and shentud. Environment variables are optional overrides, not mandatory setup: OPENMATH_ENV_CONFIG, OPENMATH_SHENTUD_BIN, OPENMATH_SUBMISSION_MODE, OPENMATH_INNER_TX_FEES, OPENMATH_INNER_TX_GAS, SHENTU_CHAIN_ID, and SHENTU_NODE_URL.

For least-privilege operation, treat openmath-env.json creation or editing, shentud installation, and local key creation or recovery as manual prerequisites documented in references/. The default skill flow may run read-only checks such as command -v shentud, shentud version, or shentud keys show, but it should not auto-install binaries, auto-edit config files, or run shentud keys add as part of normal execution.

Before any action that writes openmath-env.json or creates or recovers a local key, get explicit user approval. shentud installation should stay a manual user step, guided by the reference docs rather than performed by the skill. Prefer the official Shentu releases page plus a user-local install at $HOME/bin/shentud, and use OPENMATH_SHENTUD_BIN only as an explicit fallback. Do not generate or manage mnemonics on the user's machine without that approval.

First-run gate

If the selected openmath-env.json is missing, or if it exists but is missing prover_address, agent_address, or agent_key_name, do not proceed. Follow references/init-setup.md, and treat any config write or key creation/recovery as an explicit-user-approval step, then validate:

python3 scripts/check_authz_setup.py [--config <path>]

Require Status: ready before any submission. Repeat on each new machine or workspace.