deployment-and-data-ops
Operate Jarbas and Rosie service setup, data loading, validation, assets, and maintenance workflows safely.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- deployment-and-data-ops
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Operate Jarbas and Rosie service setup, data loading, validation, assets, and maintenance workflows safely.
- disable-model-invocation
- true
02 资源文件 8
- 参考文档references/configuration.md5.1 KB
- 参考文档references/data-loading.md8.7 KB
- 参考文档references/docker-and-services.md7.1 KB
- 参考文档references/frontend-assets.md3.9 KB
- 参考文档references/management-commands.md8.8 KB
- 参考文档references/research-and-maintenance-scripts.md6.2 KB
- 参考文档references/troubleshooting.md7.9 KB
- 脚本scripts/jarbas_manage_check.py6.3 KB
03 SKILL.md 正文
Deployment and Data Operations
Use this sub-skill when the task is to install, start, seed, validate, or troubleshoot the service/data lifecycle around Serenata de Amor's Jarbas Django service and Rosie outputs.
Route elsewhere:
- Jarbas REST endpoint semantics, serializers, filters, and response shapes are owned by
jarbas-data-api. - Rosie classifier behavior, model/cache details, and
suspicions.xzgeneration are owned byrosie-suspicion-pipeline. - Production deployment automation, DigitalOcean updates, cron jobs, and external API publishing are reference-only here; do not execute credential-bound automation unless a human explicitly authorizes the exact target environment.
Fast operating map
-
Identify the environment path: Docker Compose local, local Python service stack, or reference-only production maintenance.
-
Prepare configuration using configuration.md. Never invent or print real credentials.
-
Bring up prerequisites from docker-and-services.md: legacy Python dependencies, PostgreSQL for full Jarbas semantics, optional RabbitMQ/Celery, cache, and optional Node/Elm asset tooling.
-
Run the safe preflight helper before service-changing operations:
$ python sub-skills/deployment-and-data-ops/scripts/jarbas_manage_check.py --repo-root <serenata-checkout>Use
--no-runto inspect the command and default environment keys without invoking Django. -
Seed data using data-loading.md and management-commands.md. The safe sample order is migrations first, then reimbursements, companies, suspicions, search vector, and optional tweets/social-media/receipt commands.
-
Build or collect frontend/static assets only when the legacy Node/Elm stack is available; see frontend-assets.md.
-
For maintenance scripts and one-off research data acquisition, classify the script before running anything using research-and-maintenance-scripts.md.
-
Diagnose known failures with troubleshooting.md.
Safe defaults and validation boundaries
manage.py checkis a configuration/import preflight; it does not replace migrations, sample data loads, API tests, or PostgreSQL-backed search validation.- SQLite can be useful for
manage.py check, but full Jarbas behavior uses PostgreSQL-specific fields and search vectors. tweets,tweet,receipts,receipts_text,socialmedia, andupdatehave side effects or external dependencies; read their command notes before use.- Do not run DigitalOcean, cron, deploy, or research fetch scripts as generic setup. Treat them as historical/maintainer workflows unless the user provides credentials, target environment, rollback plan, and authorization.
Bundled files
- configuration.md — environment variables, secret handling, and check-time defaults.
- docker-and-services.md — Docker Compose topology, local install prerequisites, startup/validation sequences.
- data-loading.md — sample dataset formats, load order, validation counts, receipt text and social-media CSV assumptions.
- management-commands.md — command table, flags, prerequisites, outputs, and safety classification.
- frontend-assets.md — Node 8 / Elm 0.18 asset workflow and pitfalls.
- research-and-maintenance-scripts.md — research, cron, update, and deploy script classifications.
- troubleshooting.md — common failures and recovery steps.
- jarbas_manage_check.py — safe wrapper around
python manage.py check.
04 同仓库更多 Skills
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.
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.
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.
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.
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.
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).