| name | audithub-orca-property-discoverer |
| description | Discover English-language protocol properties for OrCa fuzzing. Use when a user wants real specifications after setup tuning, asks what invariants or temporal properties should hold, needs code/tests/docs reviewed for OrCa properties, or wants a properties.md handoff before [V] translation. |
OrCa property discoverer
Overview
Draft code-grounded English properties before writing [V] specs.
Required reference
Read ../references/specs-and-counterexamples.md.
Sub-agent use
Use sub-agents for source-heavy property discovery. Split by subsystem, target contract group, or property category when possible, and ask each agent to return only English properties, observables, rationale, source references, and [V] feasibility notes. The main campaign context does not need the full code-reading trail once the English properties and caveats are captured in orca_config/specs/properties.md.
Workflow
- Inspect target contracts, tests, docs, deployment scripts, and prior findings.
- Identify protocol assets, accounting units, actors, trust boundaries, and lifecycle states.
- Draft English properties that express intended behavior through observable API effects where possible.
- Group properties by accounting, access control, solvency, standards compliance, oracle assumptions, lifecycle, and emergency behavior.
- Mark properties that may need helper contracts, ghost variables, approximations, or weaker/stronger [V] variants.
- Write or update
orca_config/specs/properties.md when asked to store artifacts.
Output
For each property include title, English statement, contracts/functions involved, observables, rationale, likely [V] feasibility, and caveats.
Do not mirror implementation internals as the property itself; use code review to infer intended behavior, then express the intended protocol rule.