Vishesh 'ironeagle' Bangotra
6c0d859a1e
- F11 PopulateRule field_name/collection_name corrections - F12 service adaptor update handles PopulatingRepository - F13 nested-references section; F14 authless /users/me guidance - F15 template ruff-clean; F16 stop vendoring .ruff_cache, ship .gitignore - document --template-style in build_recipe/spec_first_routing; refresh collected openapi-first + standards bundle (12 modules, 43 components) - add trial.md and trial_findings.md (F1-F16, re-trial A/B evidence)
58 lines
3.1 KiB
Markdown
58 lines
3.1 KiB
Markdown
# Trial: build an Aetos app from the Standards blueprint (self-run, no auth)
|
|
|
|
You are a zero-context builder. Test the Aetos *standards* by reproducing the
|
|
reference app "khata-app" WITHOUT auth into `apps/khata-trial`, using ONLY the
|
|
standard. You may consult the existing khata-app repo ONLY as the answer key to
|
|
the intake questions below. Never copy its implementation code.
|
|
|
|
## Sources of truth (read these first)
|
|
- Standards MCP: http://localhost:8009/mcp
|
|
tools: find_library, list_app_components, get_app_component
|
|
resources: docs://nav, docs://modules/{module}, registry://libraries, registry://components
|
|
prompt: start_app
|
|
- Tier A library MCPs (read a library's own docs before using it):
|
|
openapi-first :8004, mongo-ops :8007, hexa :8008, dagpipe :8006,
|
|
omniread :8003, mail-intake :8002, py-jwt :8005, doc-forge :8001
|
|
- Standard docs: docs://modules/build_recipe, docs://modules/genericity_matrix,
|
|
docs://modules/layering_rules
|
|
- If a server is not running, start all of them from the `aetos/docs` checkout:
|
|
`.venv\Scripts\python.exe main.py`
|
|
|
|
## Step 0 — Intake (ask yourself the standard's questions)
|
|
1. App name + one-line purpose
|
|
2. Axis 1 — resources: each resource, fields, relationships (or paste openapi.yaml)
|
|
3. Axis 2 — pipelines: multi-step inputs? families + formats (or none)
|
|
4. Auth needed now? (yes/no)
|
|
5. Integrations: Mongo / ingestion formats / email / LLM provider
|
|
|
|
## Step 1 — Answer the intake from khata-app ONLY (answers only)
|
|
Resolve each answer from these artifacts, and nothing else:
|
|
- resources + fields + relationships + operationIds <- openapi.yaml (schemas + operationId names)
|
|
- pipeline families + input formats <- directory layout under pipelines/ (+ each pipeline.yaml name)
|
|
- integrations <- pyproject.toml deps + .env.example variable names
|
|
- auth <- the GLOBAL `security` block in openapi.yaml (this trial: NONE)
|
|
Do NOT read or copy khata-app implementation modules. Write the answers down first.
|
|
|
|
## Step 2 — Build (follow the standard)
|
|
Follow docs://modules/build_recipe step by step:
|
|
- Scaffold from the standard template. Prefer pulling files via get_app_component;
|
|
fallback: copy <docs>/standards/src/templates/app into apps/khata-trial.
|
|
- Author openapi.yaml for the resolved resources, OMITTING auth (no global
|
|
`security`, no bearer scheme).
|
|
- Author every CRUD slice; wire the composition root; add thin routes; bootstrap
|
|
app + worker; add pipeline families; add tests and operator scripts.
|
|
- Tier A = call the library (never reimplement); Tier B = copy the template;
|
|
Tier C = author.
|
|
|
|
## Step 3 — Verify and report
|
|
- python -c "import main" succeeds
|
|
- pytest passes
|
|
- every no-auth khata-app operationId is served -> report matched / missing
|
|
- no domain/ module imports fastapi/starlette; no Tier A library reimplemented
|
|
|
|
## Step 4 — Report standard gaps (primary output)
|
|
List every point where the standard was insufficient, wrong, or missing: a file
|
|
the docs said to copy but that didn't exist, an MCP tool that couldn't retrieve
|
|
a needed file, an intake question that wasn't answerable, etc.
|
|
|
|
Output: apps/khata-trial |