319 lines
14 KiB
TypeScript
319 lines
14 KiB
TypeScript
import { readFile } from "fs/promises"
|
|
import path from "path"
|
|
import { describe, expect, test } from "bun:test"
|
|
import { parseFrontmatter } from "../src/utils/frontmatter"
|
|
|
|
async function readRepoFile(relativePath: string): Promise<string> {
|
|
return readFile(path.join(process.cwd(), relativePath), "utf8")
|
|
}
|
|
|
|
describe("ce-code-review contract", () => {
|
|
test("documents explicit modes and orchestration boundaries", async () => {
|
|
const content = await readRepoFile("plugins/compound-engineering/skills/ce-code-review/SKILL.md")
|
|
|
|
expect(content).toContain("## Mode Detection")
|
|
expect(content).toContain("mode:autofix")
|
|
expect(content).toContain("mode:report-only")
|
|
expect(content).toContain("mode:headless")
|
|
expect(content).toContain(".context/compound-engineering/ce-code-review/<run-id>/")
|
|
expect(content).toContain("Do not create residual todos or `.context` artifacts.")
|
|
expect(content).toContain(
|
|
"Do not start a mutating review round concurrently with browser testing on the same checkout.",
|
|
)
|
|
expect(content).toContain("mode:report-only cannot switch the shared checkout to review a PR target")
|
|
expect(content).toContain("mode:report-only cannot switch the shared checkout to review another branch")
|
|
expect(content).toContain("Resolve the base ref from the PR's actual base repository, not by assuming `origin`")
|
|
expect(content).not.toContain("Which severities should I fix?")
|
|
})
|
|
|
|
test("documents headless mode contract for programmatic callers", async () => {
|
|
const content = await readRepoFile("plugins/compound-engineering/skills/ce-code-review/SKILL.md")
|
|
|
|
// Headless mode has its own rules section
|
|
expect(content).toContain("### Headless mode rules")
|
|
|
|
// No interactive prompts (cross-platform)
|
|
expect(content).toContain(
|
|
"Never use the platform question tool",
|
|
)
|
|
|
|
// Structured output format
|
|
expect(content).toContain("### Headless output format")
|
|
expect(content).toContain("Code review complete (headless mode).")
|
|
expect(content).toContain('"Review complete" as the terminal signal')
|
|
|
|
// Applies safe_auto fixes but NOT safe for concurrent use
|
|
expect(content).toContain(
|
|
"Not safe for concurrent use on a shared checkout.",
|
|
)
|
|
|
|
// Writes artifacts but no todos, no commit/push/PR
|
|
expect(content).toContain("Do not create todo files.")
|
|
expect(content).toContain(
|
|
"Never commit, push, or create a PR",
|
|
)
|
|
|
|
// Single-pass fixing, no bounded re-review rounds
|
|
expect(content).toContain("No bounded re-review rounds")
|
|
|
|
// Checkout guard — headless shares report-only's guard
|
|
expect(content).toMatch(/mode:headless.*must run in an isolated checkout\/worktree or stop/)
|
|
|
|
// Conflicting mode flags
|
|
expect(content).toContain("**Conflicting mode flags:**")
|
|
|
|
// Structured error for missing scope
|
|
expect(content).toContain("Review failed (headless mode). Reason: no diff scope detected.")
|
|
|
|
// Degraded signal when all reviewers fail
|
|
expect(content).toContain("Code review degraded (headless mode).")
|
|
})
|
|
|
|
test("documents policy-driven routing and residual handoff", async () => {
|
|
const content = await readRepoFile("plugins/compound-engineering/skills/ce-code-review/SKILL.md")
|
|
|
|
// Routing taxonomy and fixer queue semantics
|
|
expect(content).toContain("## Action Routing")
|
|
expect(content).toContain("Only `safe_auto -> review-fixer` enters the in-skill fixer queue automatically.")
|
|
|
|
// Interactive mode four-option routing structure: each distinguishing word must appear
|
|
// as a routing-option label so truncation-safe menus stay intact.
|
|
// Assert presence rather than exact copy — wording can be improved without breaking the test.
|
|
expect(content).toMatch(/\(A\)\s*`Review each finding one by one/)
|
|
expect(content).toMatch(/\(B\)\s*`LFG\./)
|
|
expect(content).toMatch(/\(C\)\s*`File a \[TRACKER\] ticket/)
|
|
expect(content).toMatch(/\(D\)\s*`Report only/)
|
|
|
|
// The new routing question dispatches to focused reference files, not inline prose.
|
|
expect(content).toContain("references/walkthrough.md")
|
|
expect(content).toContain("references/bulk-preview.md")
|
|
expect(content).toContain("references/tracker-defer.md")
|
|
|
|
// Stem is third-person (AGENTS.md:127 — no first-person "I" / "me" in the new routing question).
|
|
// The Interactive branch of After Review Step 2 must not reintroduce the removed bucket-policy wording.
|
|
expect(content).not.toContain("What should I do with the remaining findings?")
|
|
expect(content).not.toContain("What should I do?")
|
|
|
|
// Zero-remaining case: routing question is skipped with a completion summary.
|
|
expect(content).toMatch(/skip the routing question entirely/i)
|
|
|
|
// Stage 5 tie-breaking rule — the walk-through's recommendation is deterministic.
|
|
expect(content).toMatch(/Skip\s*>\s*Defer\s*>\s*Apply/)
|
|
|
|
// Autofix-mode residual todo handoff is preserved (mode isolation).
|
|
expect(content).toContain(
|
|
"In autofix mode, create durable todo files only for unresolved actionable findings whose final owner is `downstream-resolver`.",
|
|
)
|
|
expect(content).toContain("If only advisory outputs remain, create no todos.")
|
|
|
|
// Tracker fallback chain explicitly forbids extending the internal todos system.
|
|
const trackerDefer = await readRepoFile(
|
|
"plugins/compound-engineering/skills/ce-code-review/references/tracker-defer.md",
|
|
)
|
|
expect(trackerDefer).toContain(".context/compound-engineering/todos/")
|
|
expect(trackerDefer).toMatch(/Never fall back to `\.context\/compound-engineering\/todos\//)
|
|
|
|
// Subagent template carries the why_it_matters framing guidance that replaces the
|
|
// rejected synthesis-time rewrite pass. Assert presence of the observable-behavior
|
|
// rule and the required-field reminder without pinning exact prose.
|
|
const subagentTemplate = await readRepoFile(
|
|
"plugins/compound-engineering/skills/ce-code-review/references/subagent-template.md",
|
|
)
|
|
expect(subagentTemplate).toMatch(/observable behavior/i)
|
|
expect(subagentTemplate).toMatch(/required/i)
|
|
|
|
// walkthrough.md carries the four per-finding option labels (Apply / Defer / Skip /
|
|
// LFG the rest). Assert presence of each distinguishing word so renaming an option
|
|
// breaks the test. Exact label wording may be refined for clarity — these assertions
|
|
// check the structural contract, not the prose.
|
|
const walkthrough = await readRepoFile(
|
|
"plugins/compound-engineering/skills/ce-code-review/references/walkthrough.md",
|
|
)
|
|
expect(walkthrough).toContain("Apply the proposed fix")
|
|
expect(walkthrough).toContain("Defer — file a [TRACKER] ticket")
|
|
expect(walkthrough).toContain("Skip — don't apply, don't track")
|
|
expect(walkthrough).toMatch(/LFG the rest/)
|
|
|
|
// bulk-preview.md contract: exactly Proceed / Cancel, no third option.
|
|
const bulkPreview = await readRepoFile(
|
|
"plugins/compound-engineering/skills/ce-code-review/references/bulk-preview.md",
|
|
)
|
|
expect(bulkPreview).toContain("Proceed")
|
|
expect(bulkPreview).toContain("Cancel")
|
|
|
|
// Step 5 final-next-steps flow is gated on fixes-applied count, not routing option.
|
|
expect(content).toContain("fixes_applied_count")
|
|
expect(content).toMatch(/Step 5 runs only when `fixes_applied_count > 0`/i)
|
|
|
|
// Final-next-steps wording preserved.
|
|
expect(content).toContain("**On the resolved review base/default branch:**")
|
|
expect(content).toContain("git push --set-upstream origin HEAD")
|
|
expect(content).not.toContain("**On main/master:**")
|
|
})
|
|
|
|
test("keeps findings schema and downstream docs aligned", async () => {
|
|
const rawSchema = await readRepoFile(
|
|
"plugins/compound-engineering/skills/ce-code-review/references/findings-schema.json",
|
|
)
|
|
const schema = JSON.parse(rawSchema) as {
|
|
_meta: { confidence_thresholds: { suppress: string } }
|
|
properties: {
|
|
findings: {
|
|
items: {
|
|
properties: {
|
|
autofix_class: { enum: string[] }
|
|
owner: { enum: string[] }
|
|
requires_verification: { type: string }
|
|
}
|
|
required: string[]
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
expect(schema.properties.findings.items.required).toEqual(
|
|
expect.arrayContaining(["autofix_class", "owner", "requires_verification"]),
|
|
)
|
|
expect(schema.properties.findings.items.properties.autofix_class.enum).toEqual([
|
|
"safe_auto",
|
|
"gated_auto",
|
|
"manual",
|
|
"advisory",
|
|
])
|
|
expect(schema.properties.findings.items.properties.owner.enum).toEqual([
|
|
"review-fixer",
|
|
"downstream-resolver",
|
|
"human",
|
|
"release",
|
|
])
|
|
expect(schema.properties.findings.items.properties.requires_verification.type).toBe("boolean")
|
|
expect(schema._meta.confidence_thresholds.suppress).toContain("0.60")
|
|
|
|
const fileTodos = await readRepoFile("plugins/compound-engineering/skills/ce-todo-create/SKILL.md")
|
|
expect(fileTodos).toContain("/ce-code-review mode:autofix")
|
|
expect(fileTodos).toContain("/ce-todo-resolve")
|
|
|
|
const resolveTodos = await readRepoFile("plugins/compound-engineering/skills/ce-todo-resolve/SKILL.md")
|
|
expect(resolveTodos).toContain("ce-code-review mode:autofix")
|
|
expect(resolveTodos).toContain("safe_auto")
|
|
})
|
|
|
|
test("documents stack-specific conditional reviewers for the JSON pipeline", async () => {
|
|
const content = await readRepoFile("plugins/compound-engineering/skills/ce-code-review/SKILL.md")
|
|
const catalog = await readRepoFile(
|
|
"plugins/compound-engineering/skills/ce-code-review/references/persona-catalog.md",
|
|
)
|
|
|
|
for (const agent of [
|
|
"review:ce-dhh-rails-reviewer",
|
|
"review:ce-kieran-rails-reviewer",
|
|
"review:ce-kieran-python-reviewer",
|
|
"review:ce-kieran-typescript-reviewer",
|
|
"review:ce-julik-frontend-races-reviewer",
|
|
]) {
|
|
expect(content).toContain(agent)
|
|
expect(catalog).toContain(agent)
|
|
}
|
|
|
|
expect(content).toContain("## Language-Aware Conditionals")
|
|
expect(content).not.toContain("## Language-Agnostic")
|
|
})
|
|
|
|
test("stack-specific reviewer agents follow the structured findings contract", async () => {
|
|
const reviewers = [
|
|
{
|
|
path: "plugins/compound-engineering/agents/review/ce-dhh-rails-reviewer.agent.md",
|
|
reviewer: "dhh-rails",
|
|
},
|
|
{
|
|
path: "plugins/compound-engineering/agents/review/ce-kieran-rails-reviewer.agent.md",
|
|
reviewer: "kieran-rails",
|
|
},
|
|
{
|
|
path: "plugins/compound-engineering/agents/review/ce-kieran-python-reviewer.agent.md",
|
|
reviewer: "kieran-python",
|
|
},
|
|
{
|
|
path: "plugins/compound-engineering/agents/review/ce-kieran-typescript-reviewer.agent.md",
|
|
reviewer: "kieran-typescript",
|
|
},
|
|
{
|
|
path: "plugins/compound-engineering/agents/review/ce-julik-frontend-races-reviewer.agent.md",
|
|
reviewer: "julik-frontend-races",
|
|
},
|
|
]
|
|
|
|
for (const reviewer of reviewers) {
|
|
const content = await readRepoFile(reviewer.path)
|
|
const parsed = parseFrontmatter(content)
|
|
const tools = String(parsed.data.tools ?? "")
|
|
|
|
expect(String(parsed.data.description)).toContain("Conditional code-review persona")
|
|
expect(tools).toContain("Read")
|
|
expect(tools).toContain("Grep")
|
|
expect(tools).toContain("Glob")
|
|
expect(tools).toContain("Bash")
|
|
expect(content).toContain("## Confidence calibration")
|
|
expect(content).toContain("## What you don't flag")
|
|
expect(content).toContain("Return your findings as JSON matching the findings schema. No prose outside the JSON.")
|
|
expect(content).toContain(`"reviewer": "${reviewer.reviewer}"`)
|
|
}
|
|
})
|
|
|
|
test("leaves data-migration-expert as the unstructured review format", async () => {
|
|
const content = await readRepoFile(
|
|
"plugins/compound-engineering/agents/review/ce-data-migration-expert.agent.md",
|
|
)
|
|
|
|
expect(content).toContain("## Reviewer Checklist")
|
|
expect(content).toContain("Refuse approval until there is a written verification + rollback plan.")
|
|
expect(content).not.toContain("Return your findings as JSON matching the findings schema.")
|
|
})
|
|
|
|
test("fails closed when merge-base is unresolved instead of falling back to git diff HEAD", async () => {
|
|
const content = await readRepoFile("plugins/compound-engineering/skills/ce-code-review/SKILL.md")
|
|
|
|
// No scope path should fall back to `git diff HEAD` or `git diff --cached` — those only
|
|
// show uncommitted changes and silently produce empty diffs on clean feature branches.
|
|
expect(content).not.toContain("git diff --name-only HEAD")
|
|
expect(content).not.toContain("git diff -U10 HEAD")
|
|
expect(content).not.toContain("git diff --cached")
|
|
|
|
// PR mode still has an inline error for unresolved base
|
|
expect(content).toContain('echo "ERROR: Unable to resolve PR base branch')
|
|
|
|
// Branch and standalone modes delegate to resolve-base.sh and check its ERROR: output.
|
|
// The script itself emits ERROR: when the base is unresolved.
|
|
expect(content).toContain("references/resolve-base.sh")
|
|
const resolveScript = await readRepoFile(
|
|
"plugins/compound-engineering/skills/ce-code-review/references/resolve-base.sh",
|
|
)
|
|
expect(resolveScript).toContain("ERROR:")
|
|
|
|
// Branch and standalone modes must stop on script error, not fall back
|
|
expect(content).toContain(
|
|
"If the script outputs an error, stop instead of falling back to `git diff HEAD`",
|
|
)
|
|
})
|
|
|
|
test("orchestration callers pass explicit mode flags", async () => {
|
|
const lfg = await readRepoFile("plugins/compound-engineering/skills/lfg/SKILL.md")
|
|
expect(lfg).toContain("/ce-code-review mode:autofix")
|
|
})
|
|
})
|
|
|
|
describe("testing-reviewer contract", () => {
|
|
test("includes behavioral-changes-with-no-test-additions check", async () => {
|
|
const content = await readRepoFile("plugins/compound-engineering/agents/review/ce-testing-reviewer.agent.md")
|
|
|
|
// New check exists in "What you're hunting for" section
|
|
expect(content).toContain("Behavioral changes with no test additions")
|
|
|
|
// Check is distinct from untested branches check
|
|
expect(content).toContain("distinct from untested branches")
|
|
|
|
// Non-behavioral changes are excluded
|
|
expect(content).toContain("Non-behavioral changes")
|
|
})
|
|
})
|