AGENT SKILL

frontend-apps

Operate on BiSheng's two React frontends: Platform Vite/Zustand/react-query v3/bs-ui and Client Vite/Recoil/TanStack Query v4/shadcn.

License 许可
Apache-2.0
最近更新
2026-09-03
Stars
★ 351
Forks
27

01 前言 Front Matter

name
frontend-apps
license
Apache 2.0
metadata
{ "disco-role": "operating" }
description
Operate on BiSheng's two React frontends: Platform Vite/Zustand/react-query v3/bs-ui and Client Vite/Recoil/TanStack Query v4/shadcn.
disable-model-invocation
true

02 资源文件 3

03 SKILL.md 正文

frontend-apps

Use this sub-skill when a task touches either BiSheng React frontend: the Platform admin/builder app under src/frontend/platform/ or the Client workspace/chat app under src/frontend/client/.

Start Here

Run bundled helper commands from this sub-skill directory, or adjust the script path to this directory after import.

  1. Decide which app owns the requested change. The two frontends are separate SPAs and must not be mixed.
  2. Inspect the installed stack and available package scripts:
    python scripts/check_frontend_packages.py --repo-root <bisheng-checkout>
    
  3. Read references/workflows.md for app-specific routes, request wrappers, stores, i18n, theme, test commands, and review checklists.
  4. If setup, proxying, i18n, routing, API, or state behavior is failing, read references/troubleshooting.md.

Owned Responsibilities

  • Platform app: src/frontend/platform/, Vite 5, React 18, TypeScript, Zustand + React Context, react-query v3, Radix-based bs-ui, @/ alias.
  • Client app: src/frontend/client/, Vite 6, React 18, TypeScript, Recoil, @tanstack/react-query v4, shadcn/Radix UI, /workspace base path, ~/ or @/ alias.
  • Routes and route guards in each app's src/routes/index.tsx.
  • Wrapped HTTP layers: Platform src/controllers/request.ts plus src/controllers/API/; Client src/api/request.ts plus src/api/ and query hooks.
  • Frontend state boundaries: Platform stores/contexts; Client Recoil atoms/selectors.
  • Frontend i18n files, brand runtime/theme tokens, UI components, icons, unit tests, and Vite dev/build behavior.

Route Sibling Areas Instead of Duplicating Them

  • Use backend-core for FastAPI routers, response envelopes, error-code definitions, backend schemas, and backend tests.
  • Use identity-permissions-tenancy for OpenFGA/ReBAC semantics, tenant hierarchy, menu permission source data, SSO/org sync, and quota rules.
  • Use workflow-engine for backend workflow execution, node semantics, LangGraph/Celery behavior, and workflow persistence.
  • Use knowledge-rag for knowledge ingestion, file parsing, Milvus/Elasticsearch recall, and knowledge worker behavior.
  • Use linsight-mcp for Linsight worker/runtime, MCP integrations, task-mode backend events, and deepagents concerns.
  • Use deployment-maintenance for Docker Compose, Nginx, production rollout, middleware, and environment operations.

Non-Negotiables

  • Never mix Platform and Client stacks, aliases, state libraries, query libraries, or UI component systems.
  • TypeScript only for edited frontend files (.ts / .tsx); functional components only; named component exports.
  • Never import axios directly from business code. Use the app's wrapped request module; stores must not call HTTP directly.
  • Do not introduce new UI, routing, query, or state-management libraries.
  • Let response interceptors handle 403/401 flows. Do not add local 403 branches in business components.
  • Keep comments in English and split any file that grows beyond 600 lines.

04 同仓库更多 Skills

VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-triton-opt

Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-solve

END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-skill

PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-pattern-fusion

Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-orchestrate

Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-feedback

Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).

★ 351