SuperAPI AI Directory

aisdlc-project-discover-preflight

General & other

View repository

Updated 10-16 08:50

About this skill

aisdlc-project-discover-preflight (Step 0: Preparation and evidence entry point inventory)

Overview

The goal of Preflight is not to "understand all the code," but to first findexecutable evidenceandauthoritative entry pointsto provide a traceable source for each link in the subsequent .aisdlc/project/* provides a traceable source for each link.

At the start, announce:"I am using the aisdlc-project-discover-preflight skill to inventory evidence entry points and build a traceable checklist."

Input / Output

  • Input: repository directory structure, build scripts, dependency files, configuration files, CI/CD configuration, contract files (OpenAPI/Proto/Schema/migrations), observability entry points (if any).
  • Output (minimum): an "evidence entry point inventory" (first as a draft, later distributed and backfilled into Memory/index/module pages/contract pages/ops).

Evidence entry point inventory (minimum items)

Record the entry points you find under the following categories (if none, write "not found", do not guess):

  • Runtime entry points
    • Local startup command/script path
    • Environment variables/configuration entry points (examples:.env.example、config/*、values.yaml etc.)
  • Test entry points
    • Command or script path for unit/integration/E2E tests
    • Coverage/quality gate entry points (if any)
  • Build and release entry points
    • build command/script path
    • Location of pipeline/job configuration files
  • Contract entry points
    • File paths for OpenAPI/Proto/JSON Schema/GraphQL schema (if any)
    • DB schema/DDL/migration directories (if any)
    • ORM model definition entry points (if any)
  • Key directories (map clues)
    • Service/application boundaries (apps/services/packages/src etc.)
    • Gateway/routing/handler entry points
    • job/consumer/worker entry points
  • Observability entry points (if any)
    • dashboard / alerts / logs query entry point links or configuration locations
    • runbook/on-call/rollback entry points (if any)

Recommended evidence sources (by priority)

Prioritize finding "executable/locatable" evidence, then descriptive documentation:

  1. scripts/commands:package.json、Makefile、justfile、*.ps1、*.sh
  2. CI/CD:.github/workflows/*, Azure/GitLab/Jenkins pipeline files
  3. Contract files:openapi.*、proto/、schema/、migrations/
  4. README/Docs: used only as navigation, not as final evidence

Where to first place the "evidence entry point inventory" (without adding directories)

First write the draft into .aisdlc/project/memory/structure.md a temporary subsection (later it will be split apart and backfilled):

  • ## 证据入口清单(临时)
    • Entry point list for runtime/tests/build release/contracts/observability/key directories (each item includes a path or link)

Do not invent a new directory under .aisdlc/project/ just to "place the draft" (for example docs/、inbox/); it expands the maintenance surface and encourages detail accumulation.

Red flag checklist (if any item appears: stop and correct)

  • Start writing "how to run/test the project/where the contract is" when no evidence has been found (should write "not found + how to find it next")
  • For delivery speed, first generate a large number of field-level descriptions (a field encyclopedia is not a project-level artifact of Discover)
  • Treat the README description as fact without linking to scripts/CI/contract files

A good example (one record of the evidence entry checklist)

  • Test entry:package.json of scripts.test(path:package.json; command:npm test)
  • CI gate: GitHub Actions ci.yml(path:.github/workflows/ci.yml; key job:test)