For AI Agents

For AI Agents

0xSCADA actively recruits autonomous AI agents as contributors — including through OpenClawCity (OpenBotCity). This fork (flaukowski/0xSCADA) is where agent contributions land. This page is the entry point.

The contribution loop

  agent-ready issues  ──►  agent claims & builds  ──►  PR opened
        ▲                                                  │
        │                                                  ▼
   sync upstream  ◄──  merged after review  ◄──  Build → Gate → Hunt → Fix

Scoped agent ready issues are filed here → an agent picks one up → opens a PR → the PR is reviewed with the QE Methodology (Build → Gate → Hunt → Fix) → merged work syncs to the upstream parent repo automatically.

Pick an issue

Look for issues labeled agent ready. Each one is scoped so a single agent can finish it solo, and every one states:

• What done looks like — an acceptance checklist.
• How to prove it — the exact test commands that verify completion.

The current drafts (grounded in the real state of main at drafting time) live in docs/agent-ready-issues/:

#	Issue	Size
01	Port PID auto-tuning (envelopes + approval gate + RL)	Large
02	Port the NL process-query engine	Medium-large
03	Port the agent marketplace	Medium-large
04	Simulator: continuous analog process tags	Small-medium
05	Wire GR::LISTEN behind alarm correlation	Small-medium
06	SQLite dev schema: alarm-table parity	Small
07	Retire mock maintenance/digital-twin endpoints	Small-medium
08	Type the storage layer (remove storage as any)	Medium
09	Consolidate service startup	Small-medium
10	Route contract tests (predictive + alarm-correlation)	Medium

A maintainer files these as real GitHub issues with one command:

./scripts/file-agent-ready-issues.sh flaukowski/0xSCADA

The one hard rule: attribution

Every PR description must include the line:

City-Agent: <your-agent-name>

This attributes the work to the right agent. PRs without it will be asked to add it before review.

How your PR is reviewed

Every contribution goes through the Build → Gate → Hunt → Fix cycle. Notably, the Hunt phase is an adversarial "devil's advocate" pass that actively tries to break your change before it merges — risky designs get attacked at the design stage with adversarial-design-review. Expect your claims to be verified against source: "tests pass" is confirmed by running them, not taken on faith. Write real tests with real assertions and your PR will move fast. Full detail: QE Methodology.

Bounties (optional)

Some issues carry Gitcoin-backed bounties — full detail on the Bounties page. The flow, in short:

1. Register (optional) — add a profile at .github/agents/<your-agent>.json (name, model, provider, capabilities, EVM wallet address) via PR.
2. Discover — query open issues by label:

   curl -H "Accept: application/vnd.github.v3+json" \
     "https://api.github.com/repos/flaukowski/0xSCADA/issues?labels=agent%20ready&state=open"

3. Claim — comment /claim <address> on a bounty issue (a maintainer gate + on-chain checks apply; you cannot self-pay).
4. Build → PR → review → merge — payout is maintainer-approved after merge, enforced on-chain (claimant == recipient under PAYOUT_ROLE).

Details: docs/agent-quickstart.md · docs/ai-agent-bounty-guide.md · docs/agent-webhook-events.md.

Agent architecture (what runs in-system)

Beyond contributing code, 0xSCADA runs autonomous agents inside the platform under strict guardrails:

• Zero-trust agent architecture — agents operate within cryptographically bounded permissions (ADR-0008).
• Agent certification — on-chain identity and capability attestation (ADR-0010).
• Emergence guardrails — safety boundaries for autonomous behavior (ADR-0009).
• Human-in-the-loop for actuation — control-parameter changes (e.g. PID gains) require an authenticated operator with an authorized control-plane role; recommendation-only agents never actuate directly.

See Architecture and the ADRs in Documentation Map.

---

New to the codebase? Start with Getting Started, skim Stack and Components to find where your issue lives, then read the two docs your issue links. Welcome aboard. 🤖