> ## Documentation Index
> Fetch the complete documentation index at: https://clawmind.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Roadmap

> Post-hackathon roadmap for ClawMind: harden the MVP, deepen the product, expand the ecosystem, and move toward decentralized agent intelligence.

# Roadmap

ClawMind's roadmap is organized as a post-hackathon execution plan. The current product already demonstrates the core loop: an 8-step due diligence pipeline, 0G Compute inference, 0G Storage receipts, 0G Chain anchoring, EIP-712 operator signatures, semantic memory, and OpenClaw manifest exposure.

The next phases focus on making that loop more reliable, faster, easier to integrate, and eventually more decentralized.

<Info>
  Timeline: 2026 post-hackathon. Durations are approximate and milestone-driven.
</Info>

***

## Phase Overview

| Phase |    Horizon | Focus              | Outcome                                                                            |
| ----- | ---------: | ------------------ | ---------------------------------------------------------------------------------- |
| Q1    | 1-2 months | Harden the MVP     | Stable, tested, observable pipeline with safer production controls.                |
| Q2    | 2-4 months | Deepen the product | Faster UX, real risk data, stronger memory, and wallet-based signing.              |
| Q3    | 4-6 months | Ecosystem & reach  | Agent SDK, marketplace path, public API, bots, and DAO pilots.                     |
| Q4+   |  6+ months | Long horizon       | Decentralized agents, ZK inference proofs, governance, and shared protocol memory. |

***

## Q1 - Harden the MVP

The first phase is about converting the hackathon build into a dependable production system. The priority is not adding more surface area; it is making the existing 8-step pipeline easier to trust, debug, and deploy.

<Info>
  Q1 hardening is partially complete in the public repo: CI/tests, signed-registry audit notes, production rate limiting, and `/stats` observability are live. Broader end-to-end coverage and deeper external audit automation can continue as follow-up hardening work.
</Info>

| Workstream     | Status         | Deliverable                                                                                                                      |
| -------------- | -------------- | -------------------------------------------------------------------------------------------------------------------------------- |
| Tests          | In progress    | Unit coverage and 0G mocks are in CI; broader end-to-end coverage remains a follow-up.                                           |
| CI/CD          | Live           | Typecheck, lint, app build, unit tests, MCP checks, and Foundry tests run in GitHub Actions.                                     |
| Contract audit | Live           | Foundry tests and Slither findings are documented for `AnalysisRegistry.sol`; deeper Mythril/external review can be added later. |
| Observability  | Live           | `/stats` and `/api/judge` expose signed registry status, recent analyses, semantic memory, critic impact, and MCP usage.         |
| Rate limiting  | Live           | `/api/analyze` and MCP tool calls enforce production request limits.                                                             |
| Docs           | Live / ongoing | Mintlify setup, API, 0G integration, verification, and roadmap docs are published and kept in sync with the app.                 |

**Success signal:** a contributor can change an agent, contract, or storage path and get fast feedback before the change reaches production.

***

## Q2 - Deepen the Product

The second phase improves the user experience and analysis quality. ClawMind should feel less like a background job and more like a live due diligence cockpit.

| Workstream      | Deliverable                                                                                      |
| --------------- | ------------------------------------------------------------------------------------------------ |
| Parallel agents | Expand concurrent execution beyond the current Risk + Architect branch where dependencies allow. |
| SSE streaming   | Stream agent progress in real time instead of showing a spinner.                                 |
| Real risk data  | Integrate external risk signals from Forta, GoPlus, and Chainalysis.                             |
| Vector memory   | Move beyond a flat memory index toward embedding-backed retrieval.                               |
| Wallet connect  | Let users sign `recordAnalysis` themselves.                                                      |
| Model routing   | Use cheaper models for memory work and stronger models for the Critic.                           |

**Success signal:** analyses complete faster, explain themselves while running, and include evidence beyond the user's prompt text.

***

## Q3 - Ecosystem & Reach

The third phase turns ClawMind from a single app into a composable agent surface. Other developers should be able to plug agents into the pipeline, and real teams should be able to use it in their existing workflows.

| Workstream             | Deliverable                                                      |
| ---------------------- | ---------------------------------------------------------------- |
| Agent SDK              | Third-party agents via the OpenClaw manifest.                    |
| Agent marketplace      | Optional token layer for agent discovery and incentives.         |
| Multi-chain reads      | ETH, Base, and Arbitrum RPC support for richer protocol context. |
| Public API             | API keys, quotas, and billing.                                   |
| Discord / Telegram bot | Chat frontends for `/api/analyze`.                               |
| Real users             | 2-3 DAO due-diligence pilots.                                    |

**Success signal:** ClawMind is no longer only a UI; it becomes infrastructure that teams and agent developers can build on.

***

## Q4+ - Long Horizon

The long horizon moves ClawMind toward decentralized, verifiable agent intelligence. This phase depends on ecosystem maturity across 0G Compute, proof systems, model registries, and cross-DAO coordination.

| Workstream                  | Deliverable                                          |
| --------------------------- | ---------------------------------------------------- |
| Decentralized agents        | Actors on 0G Compute with reputation.                |
| ZK inference proofs         | Verify that an agent actually ran the claimed model. |
| Governance layer            | Canonical prompts and a model registry.              |
| Self-improving memory       | Memory pruning, scoring, and decay.                  |
| Cross-protocol intelligence | Shared memory across DAOs.                           |

**Success signal:** due diligence becomes a shared, verifiable intelligence layer instead of a set of isolated reports.

***

## Roadmap Principles

* **Integrity before growth.** Reports must remain verifiable through 0G Storage and 0G Chain before new interfaces are added.
* **Adversarial review stays central.** Product speed should not weaken the Critic's role or final score calibration.
* **Memory must earn trust.** Persistent memory should improve analysis quality without letting stale or low-quality data dominate future runs.
* **OpenClaw compatibility is a contract.** SDK and marketplace work should extend the manifest model rather than bypass it.
* **Human approval remains mandatory for high-risk action.** ClawMind can support due diligence, but it should not autonomously move funds or approve protocol changes.