Look into and document potential modes of operation

[mds1]

@moodysalem Here's the 4 potential modes of operation we discussed. I think all but the GitHub Action approach are a bit up in the air at this point, pending some searching into how to support the last one and how it impacts the others. Let me know if this looks right / needs changes / etc.

GitHub Actions

This will be a mode of operation

Cron job through GitHub Actions workflow that runs hourly. For each pair of DAO_NAME and GOVERNOR_ADDRESS in governance-checks.yaml, look for any pending proposals (or more generally, unsimulated proposals, to also simulate past proposals at least once?), pull the proposal data from the chain, simulate with tenderly, and write the report to the reports branch of the repo.

To add a DAO to the workflow, fork this repo, add your DAO_NAME and GOVERNOR_ADDRESS, then submit a PR back to main with those changes

TypeScript Simulation Files

TBD whether to support this

This mode would work as follows:

1. Fork the repo
2. In the sims folder, create a configuration file called mySimName.sim.ts (see existing template files)
3. Run SIM_NAME=mySimName yarn start
4. Report is saved in reports/<daoName>/<governorAddress>

@uniswap/governance-seatbelt npm package

TBD whether to support this

Users can install the @uniswap/governance-seatbelt package as normal, which exports a single function with the below interface:

async function simulate(config: SimulationConfig): Promise<SimulationResult>

Users can then generate the configuration object with any tooling they want, and run a simulation with await simulate(config)

npx @uniswap/governance-seatbelt <configFile>

TBD whether to support this

Let users define a TS config file anywhere and run it without needing to explicitly install an npm package using npx @uniswap/governance-seatbelt <path/to/configFile.ts>

[moodysalem]

maybe for the @uniswap/governance-seatbelt npm package we also export code for rendering and implement one for markdown
e.g.

interface Renderer<T> {
  render(result: SimulationResult): T 
}

class MarkdownRenderer implements Renderer<string> {
  render(result: SimulationResult): string {
    // ... whatever we do for the github actions rendering
  }
}

[mds1]

I like that 👌