omni-link-hustlexp

omni-link-hustlexp

HustleXP-tailored multi-repo engineering control plane for Claude Code — powered by the ruflo orchestration engine with swarm coordination, hybrid vector memory, self-learning orbit loops, docs authority ingestion, Swift↔tRPC bridge analysis, phase-drift gating, bounded automation, branch-aware review workflows, and multi-agent orchestration across the HustleXP iOS, backend, and docs repos.

Ecosystem Status (April 2, 2026): Full source-level audit + 7 adversarial stress test loops completed. Backend has 3 CRITICAL, 4 HIGH open issues. 64 TODOs tracked across all repos. See HUSTLEXP-ERRORS-AND-TODOS for the prioritized list and HustleXP-Vault for the full 16-page audit.

What it does

This fork starts from upstream omni-link and narrows it around one operating model:

• HUSTLEXPFINAL1 as the Swift/iOS consumer
• hustlexp-ai-backend as the tRPC/API provider
• HUSTLEXP-DOCS as the authority/governance repo

It still preserves the upstream command surface, but the hustlexp workflow profile adds project-specific truth sources that upstream generic scanning cannot infer reliably.

In max-tier HustleXP mode, the fork behaves like a local-first control plane for this exact topology:

• It keeps warm state in a SQLite-backed daemon instead of rescanning everything on every command.
• It treats the docs repo as an authority source instead of just another Markdown tree.
• It extracts Swift client trpc.call(router: ..., procedure: ...) usage and correlates it with backend procedures under backend/src.
• It detects phase drift between docs, backend, and iOS code before execution is allowed.
• It combines parser coverage with compiler-backed semantic analysis for TypeScript, Go, Python, Java, and Swift, plus AST-backed GraphQL analysis.
• It generates branch-aware review artifacts with risk, owners, policy decisions, and rollback-aware execution plans.
• It publishes or replays provider-native review output for GitHub with live metadata negotiation and idempotent comment updates.
• It keeps automation bounded: branch and PR oriented, policy gated, and auditable.
• It provides structured multi-agent orchestration with handoff protocols, MCP message bus, and Agent Teams bridge.
• It runs swarm-coordinated orbit loops where a queen agent dispatches attack/fix workers, with load balancing and consensus.
• It stores every finding in hybrid vector memory (SQLite + AgentDB) so agents can search semantically across prior rounds.
• It learns from prior orbit rounds — domain risk weights, regression detection, and recommended attack priorities improve automatically.
• It reduces token costs by 30–50% through deduplication, compression, and relevance filtering before the context pruner runs.

Why it is different

• Cross-repo truth, not single-repo guesses — omni-link reasons over the ecosystem graph, not just one working directory.
• Semantic where it matters — compiler-backed analysis is used where available, with structured fallback where necessary.
• Operationally safe — direct protected-branch mutation is blocked, execution is policy-gated, and rollback plans are always generated.
• Provider-aware — GitHub and GitLab publishing is capability-aware, metadata-aware, and proven against live sandbox PR/MR targets.
• Production-hardened — structured error types, partial failure resilience, git operation timeouts, path traversal prevention, and explicit cache invalidation API.
• Multi-agent ready — structured handoff protocol, MCP communication bus, Agent Teams bridge, error recovery triage, and shared memory bulletin board.
• Swarm-powered — ruflo engine provides queen-worker hierarchical swarms, workflow execution with dependency ordering and rollback, hybrid vector memory, and a plugin system with extension points.
• Self-improving — the self-learning loop records findings per orbit round, weights domains by risk, flags regressions, and recommends priorities for the next round.
• Token-efficient — layered optimization (deduplication, compression, relevance filtering) composes with the existing token pruner for 30–50% cost reduction.
• Proven, not hypothetical — 887 tests passing across unit, integration, E2E lifecycle, smoke, stress, ruflo engine, and contract fixture suites.

Features