Audithub Orca Live State Setup

OrCa live-state setup

Overview

Build a proxy-aware live-state deployment configuration that OrCa can fuzz from a selected chain and block.

Required reference

Read ../references/live-state-setup.md. Skip ../references/campaign-state-and-autonomy.md if already loaded by the campaign orchestrator; read it only when running this skill standalone.

For the concrete OrCa configuration and on-chain deployment JSON formatting rules, also read the OrCa documentation under ../../veridise-docs/orca/user_guide/orca_configuration/, especially common_settings.md for fork network/block fields, advanced_settings.md for fuzzing targets, and build_system_and_deployment/debug.md for deployment/proxy caveats. Use ../references/live-state-setup.md#on-chain-deployment-json-checks as the skill-pack validation checklist after applying those docs.

Sub-agent use

Use a sub-agent for proxy/ABI/address discovery or explorer-derived source review when it would add large external or source context. Return chain, block, target names, proxy-to-implementation mappings, ABI/source paths, unresolved inputs, and live-state [V] compatibility notes. Keep current block selection and network freshness explicit with concrete block numbers and dates when applicable.

When delegating, instruct the sub-agent to use explorer JSON API endpoints rather than opening HTML pages — see ../references/live-state-setup.md#explorer-api-patterns. Each HTML page visit adds thousands of tokens to context; a targeted API call adds tens of tokens.

ABI and proxy resolution

Use explorer JSON API endpoints, not HTML page navigation. See ../references/live-state-setup.md#explorer-api-patterns for Etherscan-specific patterns.

When writing or validating on-chain deployment JSON, always include the full ABI for each contract being fuzzed. Do not trim the ABI down to only the functions that seem relevant, even if a smaller ABI would be sufficient for the immediate setup task.

For proxy implementation at a historical block: query getLogs for Upgraded(address) events (topic0 0xbc7cd75a20ee27fd9adebab32041f755214dbc6bffa90cc0225b39da2e5c2d3b) from block 0 to the target block (toBlock=<fork_block>) and take the last result. This is a single API call; do not open multiple Etherscan HTML pages to reconstruct proxy history.

Canonical run_orca_task payload

Use this shape when submitting a live-state task via the ah MCP server. parameters is required; omitting it produces an HTTP 500. specs_override entries are discriminated by type (version | stdlib | orglib | adhoc).

{
  "organization_id": <int>,
  "project_id": <int>,
  "version_id": <int>,
  "task_input": {
    "name": "<campaign>-live-<block>-runN",
    "deployment_info_file": "<project>_live_<block>.deployment.json",
    "specs_override": [
      { "type": "version", "relative_path": "v_specs/<file>.v" }
    ],
    "parameters": {
      "timeout": 120,
      "fork_network": "<NetworkName>",
      "fork_block_number": <int>,
      "fuzz_targets": ["<ContractName>"],
      "language": "solidity"
    }
  }
}

• deployment_info_file must end in .deployment.json to signal on-chain mode.
• specs_override relative_path values are resolved against the uploaded archive's root.
• parameters.fuzz_targets lists contract names, not function selectors.
• parameters.fork_network must be one of the backend-allowed values: mainnet, testnet, arbitrum-mainnet, arbitrum-testnet, base-mainnet, base-testnet, bsc-mainnet, bsc-testnet.

Archive layout

The uploaded archive must be flat-rooted; AuditHub resolves relative_path against the top of the archive. Required top-level entries:

• campaign.json
• run_ledger.json
• <project>_live_<block>.deployment.json
• abis/
• specs/
• v_specs/

Do not nest these under a project-named subfolder.

Workflow

1. Confirm target chain and block. Use the latest block by default only when current chain data has been fetched or supplied.
2. Gather addresses, display names, and source/ABI data for every target.
3. Detect proxies and resolve implementation ABIs while keeping the proxy address as the fuzzed address.
4. Format and validate the on-chain deployment JSON against the OrCa documentation listed above before any smoke run.
5. Record source paths for targets when available, because live-state [V] specs may need Solidity source to map struct fields and enum variants to numeric indexes.
6. Record live-state details in orca_config/campaign.json when writing artifacts.
7. Leave special fuzz users empty unless balances, roles, or approvals justify adding them.

[V] spec compatibility

Live-state OrCa does not have source-level type information from on-chain data. When handing off to audithub-orca-v-spec-writer, include enough source context to replace struct field names with numeric field indexes, enum variant names with numeric variant indexes, and interface-typed vars entries with the known deployed contract names. These replacements are specific to live-state campaigns and are not required merely because a local deployment campaign exists.

Output

Return or write:

• chain and fork block
• target address/name/ABI entries
• proxy and implementation address mapping
• ABI source notes
• validation issues and fixes

If live data must be fetched and network access is unavailable, stop with a precise list of missing chain/block/address/ABI inputs.