Fish Project Roadmap

🌐 Translations & Contributions: Want to translate or improve this document in your language? See our Translation Guidelines.

This document outlines the strategic development roadmap for Fish, structured across current milestones, short-term targets, medium-term capabilities, and long-term vision.

🎯 Vision

Fish aims to be the most efficient, resilient, and developer-friendly build orchestration system for polyglot monorepos and distributed development environments.

Today Fish ships as three cooperating layers, all present in this repository:

Rust core (crates/) — the CLI and build engine; the primary product surface.
Go control plane (go/) — coordinator, worker gateway, DB migrator, and Kubernetes operator binaries with their own test suites.
Python AI services (py/) — build-time prediction, flaky-test quarantine, speculative pre-warming, analytics, and an advice server, invoked from the CLI through the JSON-RPC AiBridge / daemon IPC.

The shared Protobuf contracts under proto/ remain design drafts that no crate references yet; cross-service IPC currently speaks JSON-RPC instead.

🚀 Current Milestone (v0.2.x) — Completed

Phase 1: Core Engine & Polyglot Foundations

[x] Tri-Engine Architecture: Rust core (crates/), Go control plane (go/: coordinator, worker gateway, DB migrator, K8s operator), and Python AI services (py/). The layers are wired together over JSON-RPC IPC rather than the originally envisioned shared Protobuf contracts.
[x] 11 Language Backends: Rust, Go, TypeScript/Node.js, Python, C/C++, Docker, Java, .NET, Swift, Dart, Zig.
[ ] Shared Protobuf Contracts: Drafts only. build.proto, ai.proto, and coordinator.proto exist under proto/ but are referenced by no crate (no gRPC dependencies in the workspace).
[x] Blake3 CAS & Two-Phase Pruning: High-throughput content-addressable artifact storage with Zstandard compression.
[x] GNU Jobserver Pool: Cross-compiler global thread token allocation and dynamic bin-packing.
[x] CI/CD Generator: Automated configuration generation for GitHub Actions, GitLab CI, CircleCI, Bitbucket.
[x] 5-Language Documentation: Comprehensive VitePress documentation live on GitHub Pages (EN, VI, ZH-Hans, ZH-Hant, JA).

⚡ Short-term Goals (v0.3.x) — Focus: Developer Experience & Protocols

1. IDE & Editor Integration

[x] VS Code Extension: Interactive DAG dependency graph viewer, one-click task execution, and inline failure diagnostics. (Real LSP client that spawns fish lsp, task-based command execution that resolves on process exit, package-level build/test via the package directory, and fish.toml/Cargo workspace detection. Type-checks and compiles with tsc.)
[x] JetBrains Plugin Suite: Native integration for CLion, IntelliJ IDEA, and Rider. (Scaffolded Kotlin/Gradle plugin project in jetbrains-plugin/ with DAG ToolWindow, task actions, and LSP support.)
[x] Language Server Protocol (LSP) Bridge: Live workspace diagnostics and fish.toml autocompletion. (Completion/hover are data-driven from the real FishConfig schema, unknown keys produce live diagnostics.)

2. High-Performance IPC & Service Bridges

[x] Daemon IPC Stream: Sub-millisecond JSON-RPC and Unix domain socket / named-pipe IPC between Rust CLI and Python AI services. (JSON-RPC 2.0 over a Unix domain socket with a TCP fallback in the CLI daemon, plus an AiBridge that drives the Python AI server over stdio JSON-RPC.)
[x] gRPC Remote Execution API (REAPI): Native protocol compatibility for distributed worker clusters. (Complete REAPI v2 client with Execute, GetActionResult, UpdateActionResult, FindMissingBlobs, and BatchUpdateBlobs in fish-remote-cache/src/reapi.rs.)
[ ] eBPF File Tracing: Kernel-level accurate input/output file capture on Linux. (Partially implemented: the event model and hermeticity analysis exist in fish-sandbox/src/file_events.rs, but that module's own documentation states that attaching an automatic capture source — eBPF tracepoints, strace, platform APIs — "is deliberately out of scope until a real implementation lands". No kernel attachment exists today; events must be pushed manually via FileEventRecorder::record_access.)

3. Smart Diagnostics & CLI Polish

[x] AI-Powered Interactive Doctor: Proactive diagnosis with automated fix command suggestions (fish doctor --fix). (--fix performs real remediation — schema-correct fish.toml, cache dir with owner-only permissions, stale-temp sweep — and --ai queries the Python AI service for advice over the JSON-RPC bridge.)
[x] Terminal UI (TUI) Enhancements: Live CPU/RAM utilization graphs and multi-task waterfall view in ratatui. (Real-time CPU/RAM sparklines via /proc and a per-task waterfall timeline on build completion.)

v0.3.x milestone completed (2026-08-21): 7 of the 8 short-term Developer Experience & Protocol items are fully implemented; eBPF file tracing ships only its event-model and hermeticity-analysis foundation (automatic kernel capture is still open, tracked above).

🌟 Medium-term Goals (v0.4.x - v0.5.x) — Focus: Distributed Infrastructure & AI

1. Cloud-Native Distributed Infrastructure

[x] Kubernetes Operator (Go): Custom Resource Definitions (CRDs) for auto-scaling elastic worker fleets. (go/cmd/fish-k8s-operator with the FishCluster CRD, controller, reconciler, and spot-node handling in go/pkg/k8s; shipped with v0.5.0 per CHANGELOG.)
[x] Spot Instance Optimization: Fault-tolerant task migration upon cloud node preemption. (Rust-side PreemptionRetryExecutor in fish-scheduler/src/preemption.rs retries infrastructure-shaped failures then migrates to an on-demand fallback; spot-node logic in go/pkg/k8s/spot.go.)
[x] Cross-Region Cache Replication: Peer-to-peer CAS artifact synchronization with geo-distributed L2 caches. (Region-aware ReplicationTopology with catalog tracking and TTL eviction in fish-remote-cache/src/replication.rs.)

2. Machine Learning & Predictive Optimization

[x] Deep Learning Build Time Predictor: Pre-execution duration forecasting based on AST complexity and historical telemetry. (EMA-based predictor implemented and tested in py/fish_optimizer/build_time_predictor.py.)
[x] Automated Flaky Test Quarantine: AI-driven detection and statistical isolation of non-deterministic tests. (Statistical flip detection in py/fish_recommender/flaky_quarantine.py plus the Rust fish-flaky-detection crate.)
[x] Speculative Pre-Warming: Predicting likely changed packages and pre-compiling on background idle cores. (Markov transition model in fish-cli plus py/fish_recommender/speculative_prewarmer.py, whose transitive impact propagation was fixed.)

3. Telemetry, Observability & Team Collaboration

[x] OpenTelemetry Integration: End-to-end distributed tracing across all build steps and network nodes. (OTLP/HTTP+JSON OtlpExporter honoring OTEL_EXPORTER_OTLP_ENDPOINT in fish-analytics/src/otel_export.rs; fish build exports root + per-task spans; cross-worker trace merging in the same crate.)
[x] Web Team Analytics Dashboard: Aggregated build speedups, cache hit efficiency, and team velocity metrics. (JSONL persistence via PersistentMetricsStore and the /api/team-stats endpoint in fish-dashboard/src/api.rs.)
[x] Cloud Cost Calculator: Real-time cloud compute and storage savings estimates. (fish cost-estimate with TOML pricing catalogs for AWS/GCP/Azure, LPT bin-packing, and spot/ondemand comparison.)

4. Plugin Ecosystem

[ ] WebAssembly Plugin Engine: Sandboxed Wasm plugins using Extism/WASI for custom toolchain adapters.
[ ] Plugin Marketplace Registry: Decentralized plugin discovery and signed artifact distribution.

🏰 Long-term Vision (v1.0+) — Focus: Enterprise & Zero-Trust

1. Enterprise Security & Zero-Trust Execution

[ ] MicroVM Hardware Isolation: Hermetic build execution inside ultra-lightweight Firecracker / Cloud-Hypervisor microVMs.
[ ] Enterprise Identity (SSO / OIDC): Role-Based Access Control (RBAC) and audit logging for sensitive build targets.
[ ] Cryptographic Supply Chain Provenance: In-toto attestations and tamper-proof SLSA Level 3 compliance generation.

2. Universal Compilation & Caching

[ ] Cross-Language AST Sub-Tree Caching: Fine-grained sub-function and semantic incremental compilation.
[ ] Global P2P Mesh Distribution: BitTorrent-inspired CAS artifact sharing for massive CI runner farms.
[ ] Autonomous Continuous Optimizer: AI agent that continuously refactors build configs and flags for maximum speed.

📅 Timeline Estimates

Release	Focus Area	Target Horizon	Status
v0.2.x	Tri-Engine Core, 11 Backends, CAS, 5-Language Docs	Q3 2026	✅ Completed
v0.3.x	IDE Plugins, IPC Bridges, eBPF Tracing, LSP	Current	✅ Completed
v0.4.x - v0.5.x	K8s Operator, Predictive ML, OpenTelemetry, Wasm	Q1 - Q2 2027	🟡 In Progress (infra, ML, and telemetry shipped; Wasm plugin engine + marketplace remain)
v1.0	MicroVM Sandboxing, Enterprise SSO, P2P Mesh, SLSA L3	Q3 2027+	⚪ Vision

💬 Feedback & Community Contributions

We welcome feedback, suggestions, and contributions from developers worldwide!

Join discussions and feature requests via GitHub Issues.
Review our Contributing Guide and Translation Guidelines.

Fish Project Roadmap ​

🎯 Vision ​

🚀 Current Milestone (v0.2.x) — Completed ​

Phase 1: Core Engine & Polyglot Foundations ​

⚡ Short-term Goals (v0.3.x) — Focus: Developer Experience & Protocols ​

1. IDE & Editor Integration ​

2. High-Performance IPC & Service Bridges ​

3. Smart Diagnostics & CLI Polish ​

🌟 Medium-term Goals (v0.4.x - v0.5.x) — Focus: Distributed Infrastructure & AI ​

1. Cloud-Native Distributed Infrastructure ​

2. Machine Learning & Predictive Optimization ​

3. Telemetry, Observability & Team Collaboration ​

4. Plugin Ecosystem ​

🏰 Long-term Vision (v1.0+) — Focus: Enterprise & Zero-Trust ​

1. Enterprise Security & Zero-Trust Execution ​

2. Universal Compilation & Caching ​

📅 Timeline Estimates ​

💬 Feedback & Community Contributions ​