AGENT SKILL

deployment-maintenance

Operate BiSheng deployment, local setup, Docker Compose, uv/npm commands, backend operational scripts, migrations versus backfills, arch guard, and SDD maintenance workflows.

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

01 前言 Front Matter

name
deployment-maintenance
license
Apache 2.0
metadata
{ "disco-role": "operating" }
description
Operate BiSheng deployment, local setup, Docker Compose, uv/npm commands, backend operational scripts, migrations versus backfills, arch guard, and SDD maintenance workflows.
disable-model-invocation
true

02 资源文件 3

03 SKILL.md 正文

deployment-maintenance

Use this sub-skill when a task asks how to install, run, deploy, configure, migrate, operate, or maintain the BiSheng repository rather than changing one runtime subsystem's business logic.

Start here

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

  1. Inspect the deployment and script layout without starting services:
    python scripts/check_deployment_layout.py --repo-root <bisheng-checkout>
    
  2. Read references/workflows.md for local dev, Docker Compose, worker, config, migrations, backend scripts, SDD, and arch-guard workflows.
  3. Read references/troubleshooting.md for startup, config, storage, script, migration, and deployment symptoms.

Owned responsibilities

  • Docker Compose deployment under docker/, Nginx/entrypoint behavior, optional FT/OnlyOffice/Unstructured services, and production runtime topology.
  • Local backend setup via uv, Platform/Client npm setup, worker commands, and Linsight process startup.
  • config.yaml layering, BS_* env overrides, Redis config cache, encrypted password pitfalls, and storage service prerequisites.
  • Backend operational scripts under src/backend/scripts/, root scripts/, and tools/.
  • Alembic DDL versus manual data backfills, migration sequencing, and script dry-run/apply conventions.
  • AGENTS.md, docs/constitution.md, docs/SDD-Guide.md, scripts/arch-guard.sh, and feature SDD maintenance process.

Route sibling areas instead of duplicating them

  • Use backend-core for API/service/model implementation details.
  • Use knowledge-rag for knowledge worker/parser/vector behavior after the services are running.
  • Use workflow-engine for workflow execution internals after the worker queue is healthy.
  • Use linsight-mcp for Linsight task runtime internals after process startup is healthy.
  • Use identity-permissions-tenancy for permission, tenant, SSO, gateway, and approval semantics.
  • Use frontend-apps for React app code, Vite details beyond setup/proxy, and UI tests.

Non-negotiables

  • Run backend commands from src/backend/ and frontend commands from the owning app directory.
  • Prefer uv sync --frozen for backend dependencies; do not silently mutate user environments.
  • Manual data migrations/backfills default to dry-run and require explicit apply.
  • Alembic migrations contain DDL only; data backfill/cleanup belongs in operational scripts.
  • Do not write plaintext passwords into config YAML. BiSheng encrypts sensitive config values.
  • If scripts/arch-guard.sh reports a violation, fix the violation rather than bypassing the hook.

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