mirror of
https://github.com/EveryInc/compound-engineering-plugin.git
synced 2026-06-19 15:41:46 +02:00
96 lines
4.8 KiB
TypeScript
96 lines
4.8 KiB
TypeScript
import { readFileSync } from "fs"
|
|
import path from "path"
|
|
import { describe, expect, test } from "bun:test"
|
|
|
|
const BODY = readFileSync(
|
|
path.join(
|
|
process.cwd(),
|
|
"plugins/compound-engineering/skills/ce-brainstorm/references/brainstorm-sections.md",
|
|
),
|
|
"utf8",
|
|
)
|
|
|
|
// The 2026-05-13 cloakbrowser brainstorm dogfood put Key Decisions late in
|
|
// the artifact, where it got lost in the details. The decisions in that doc
|
|
// were framing choices that constrained Requirements / Flows / Scope — they
|
|
// belong near the top so the reader encounters them as the doc's narrative
|
|
// spine, not as bottom-of-doc reference material. Under the agency-driven
|
|
// section contract, "Include when material" lists Key Decisions early in
|
|
// the catalog (right after Problem Frame) and the catalog item's prose
|
|
// states the placement rationale.
|
|
describe("ce-brainstorm Key Decisions placement", () => {
|
|
test("Key Decisions appears early in the Include-when-material catalog (after Problem Frame, before Actors / Key Flows / Scope)", () => {
|
|
const catalogStart = BODY.indexOf("## Include when material")
|
|
expect(catalogStart).toBeGreaterThan(-1)
|
|
const catalogEnd = BODY.indexOf("\n## ", catalogStart + 5)
|
|
const catalog = BODY.slice(catalogStart, catalogEnd)
|
|
|
|
const problemFrame = catalog.indexOf("**Problem Frame**")
|
|
const keyDecisions = catalog.indexOf("**Key Decisions**")
|
|
const actors = catalog.indexOf("**Actors**")
|
|
const keyFlows = catalog.indexOf("**Key Flows**")
|
|
const scopeBoundaries = catalog.indexOf("**Scope Boundaries**")
|
|
|
|
expect(problemFrame).toBeGreaterThan(-1)
|
|
expect(keyDecisions).toBeGreaterThan(-1)
|
|
expect(actors).toBeGreaterThan(-1)
|
|
expect(keyFlows).toBeGreaterThan(-1)
|
|
expect(scopeBoundaries).toBeGreaterThan(-1)
|
|
|
|
// Key Decisions must sit AFTER Problem Frame and BEFORE Actors / Key
|
|
// Flows / Scope Boundaries in the catalog.
|
|
expect(keyDecisions).toBeGreaterThan(problemFrame)
|
|
expect(keyDecisions).toBeLessThan(actors)
|
|
expect(keyDecisions).toBeLessThan(keyFlows)
|
|
expect(keyDecisions).toBeLessThan(scopeBoundaries)
|
|
})
|
|
|
|
test("Key Decisions catalog item explains its placement to the agent", () => {
|
|
// The placement rationale must be visible at the catalog item itself
|
|
// so the agent reading the contract sees the why inline. Defends
|
|
// against a future edit that drops Key Decisions to a later catalog
|
|
// position because "audit content typically goes at the end."
|
|
const keyDecIdx = BODY.indexOf("**Key Decisions**")
|
|
expect(keyDecIdx).toBeGreaterThan(-1)
|
|
const itemRegion = BODY.slice(keyDecIdx, keyDecIdx + 700)
|
|
expect(
|
|
/constrain Requirements|sits high|sits here|near the top|opinionated|framing choices/i.test(itemRegion),
|
|
"Key Decisions catalog item must include placement rationale (constrains Requirements / Flows / Scope; sits near the top; opinionated framing choices).",
|
|
).toBe(true)
|
|
})
|
|
})
|
|
|
|
// Mirror of the plan-sections.md metadata test on the ce-plan side
|
|
// (`plan-sections.md enumerates the required plan metadata fields by name`).
|
|
// PR #826 split the prescriptive requirements-capture.md into a section
|
|
// contract (brainstorm-sections.md) + format-rendering refs.
|
|
// markdown-rendering.md now says "Per-skill frontmatter fields are defined
|
|
// in each skill's section contract" — so brainstorm-sections.md MUST
|
|
// actually list them or downstream consumers that key on these field names
|
|
// (filename construction via `date`+`topic`, Phase 0.1 resume detection)
|
|
// break silently when agents compose brainstorms from the new refs.
|
|
describe("ce-brainstorm metadata field contract", () => {
|
|
test("brainstorm-sections.md enumerates the required brainstorm metadata fields by name", () => {
|
|
// Required field names that downstream consumers depend on.
|
|
for (const field of ["date", "topic"]) {
|
|
expect(
|
|
new RegExp(`\\b${field}\\b`).test(BODY),
|
|
`brainstorm-sections.md must name the required '${field}' metadata field — downstream tooling keys on it (filename construction, resume detection).`,
|
|
).toBe(true)
|
|
}
|
|
})
|
|
|
|
test("brainstorm-sections.md states that the status flip mechanic does not apply", () => {
|
|
// The shared HTML rendering reference describes `<span class="status">`
|
|
// as a load-bearing hook for ce-work's active → completed flip. That
|
|
// mechanic is plan-side; brainstorm has no status lifecycle. The
|
|
// contract must say so explicitly so an agent reading
|
|
// brainstorm-sections.md + html-rendering.md together doesn't invent
|
|
// a status field for brainstorm artifacts.
|
|
expect(
|
|
/no `?status`? field|status flip does not apply|no.*active.*completed|plan-side mechanic/i.test(BODY),
|
|
"brainstorm-sections.md must explicitly state that the `status` field / `active → completed` flip mechanic does not apply to brainstorms.",
|
|
).toBe(true)
|
|
})
|
|
})
|