@bugabinga/pi-ext-devil 0.1.0

package/CHANGELOG.md

@@ -0,0 +1,16 @@
+# Changelog
+
+## 0.1.0 - 2026-05-21
+
+- a40a427 prepare extensions for npm release
+- 133cb7d chore(pi): migrate extensions to earendil packages
+- 5ca1296 Rework Pi agent extensions
+- b87a61a feat(pi): monorepo workspace — all extensions are proper packages
+- 84b39b5 pi(ext/devil): move into dir, add README and types
+- 7b175ca feat(devil): comprehensive SDK integration improvements
+- 21fff5e fix(agent): check hasUI before setStatus
+- 32a29eb fix(agent): use setStatus for progress feedback
+- ddf7242 fix(agent): show progress widget in devil tool
+- b7e8de3 style(agent): apply caveman-ultra to devil prompts
+- 5c152c7 feat(agent): add devil - structured debate tool for reducing sycophancy
+

package/README.md

@@ -0,0 +1,82 @@
+# devil
+
+Structured `/devil` command for stress-testing ideas against a devil's advocate.
+
+Canonical names:
+
+- Slash command: `/devil <idea>`
+- Settings key: `devil`
+- Extension dir: `extensions/devil/`
+
+No agent-facing tool or shortcut is registered.
+
+## Quick Start
+
+```bash
+pi
+/devil we should switch to microservices
+```
+
+## Behavior
+
+| Aspect | Behavior |
+|--------|----------|
+| Session state | Stateless. Each debate is independent. |
+| Conversation context | Automatically includes working dir, recent conversation, current contents of recently touched files. |
+| Model | Uses `devil.model` if configured, else current model. |
+| UI | Opens interactive progress UI, then result editor. |
+| Results | Edited verdict is appended as `devil-result` session state and submitted as user text. |
+
+## Configuration
+
+`~/.pi/agent/settings.json`:
+
+```json
+{
+  "devil": {
+    "model": "zai/glm-5.1",
+    "maxRounds": 5,
+    "stoppingConditions": "consensus",
+    "concessionThreshold": 0.7,
+    "onBlocking": "surface"
+  }
+}
+```
+
+CLI flags:
+
+```bash
+pi --devil-max-rounds 3 --devil-stop max_rounds
+```
+
+## Output
+
+Synthesis includes:
+
+- recommendation: `proceed`, `proceed_with_caution`, `revise`, `abandon`
+- blocking issues
+- addressable concerns
+- survived claims
+- conceded weaknesses
+- next steps
+- debate metadata: rounds, blocking challenges raised, unresolved blocking issues in final verdict
+
+## Troubleshooting
+
+### "No model available"
+
+Select model with `/model`, or configure `devil.model`.
+
+### "No API key"
+
+Log in / configure key for selected or configured provider.
+
+### Irrelevant challenges
+
+Make `/devil <idea>` more specific. Debate context is taken from current session and recently touched files.
+
+## Demo
+
+<!-- demo:advisor_suite:start -->
+![Advisor suite](assets/advisor_suite.gif)
+<!-- demo:advisor_suite:end -->

package/__tests__/helpers.test.ts

@@ -0,0 +1,207 @@
+import { describe, it, expect } from "bun:test";
+import type { DebateOutput } from "../types.ts";
+import {
+	textFromContent,
+	abortedResult,
+	checkAborted,
+	countBlockingIssues,
+	calcConcessionScore,
+	buildContextText,
+	extractSnippet,
+	cleanTranscriptText,
+} from "../helpers.ts";
+
+// ─── textFromContent ──────────────────────────────────────────────────
+
+describe("textFromContent", () => {
+	it("extracts text from content parts", () => {
+		const parts = [
+			{ type: "text", text: "Hello" },
+			{ type: "text", text: "World" },
+		];
+		expect(textFromContent(parts)).toBe("Hello\nWorld");
+	});
+
+	it("skips non-text parts", () => {
+		const parts = [
+			{ type: "text", text: "visible" },
+			{ type: "image", text: "hidden" },
+		];
+		expect(textFromContent(parts)).toBe("visible");
+	});
+
+	it("returns empty for empty array", () => {
+		expect(textFromContent([])).toBe("");
+	});
+
+	it("returns empty for parts with no text field", () => {
+		const parts = [{ type: "thinking" }, { type: "toolCall" }];
+		expect(textFromContent(parts)).toBe("");
+	});
+});
+
+// ─── extractSnippet ───────────────────────────────────────────────────
+
+describe("extractSnippet", () => {
+	it("extracts first meaningful line", () => {
+		const text = "### [blocking] Missing auth\n**Concern:** No tokens\nReal issue here";
+		expect(extractSnippet(text)).toBe("[blocking] Missing auth");
+	});
+
+	it("skips Concern/Fix/Status headers", () => {
+		const text = "**Concern:** bad\n**Fix:** add check\nActual content here";
+		expect(extractSnippet(text)).toBe("Actual content here");
+	});
+
+	it("truncates long lines", () => {
+		const longLine = "A".repeat(80);
+		expect(extractSnippet(longLine).length).toBeLessThanOrEqual(63); // 60 + "…"
+		expect(extractSnippet(longLine)).toContain("…");
+	});
+
+	it("returns empty for empty text", () => {
+		expect(extractSnippet("")).toBe("");
+	});
+});
+
+// ─── cleanTranscriptText ───────────────────────────────────────────────
+
+describe("cleanTranscriptText", () => {
+	it("removes carriage returns", () => {
+		expect(cleanTranscriptText("hello\r\nworld", 100)).toBe("hello\nworld");
+	});
+
+	it("collapses excessive newlines", () => {
+		expect(cleanTranscriptText("a\n\n\n\nb", 100)).toBe("a\n\nb");
+	});
+
+	it("truncates and adds ellipsis", () => {
+		const long = "A".repeat(200);
+		const result = cleanTranscriptText(long, 50);
+		expect(result.length).toBeLessThanOrEqual(51); // 50 + "…"
+		expect(result).toContain("…");
+	});
+
+	it("passes through short text unchanged", () => {
+		expect(cleanTranscriptText("hello", 100)).toBe("hello");
+	});
+});
+
+// ─── abortedResult ────────────────────────────────────────────────────
+
+describe("abortedResult", () => {
+	it("returns cancelled debate output", () => {
+		const result = abortedResult(["round1"], 3);
+		expect(result.aborted).toBe(true);
+		expect(result.synthesis).toBe("Debate cancelled by user.");
+		expect(result.recommendation).toBe("proceed_with_caution");
+		expect(result.blockingRaised).toBe(3);
+		expect(result.blockingRemaining).toBe(3);
+		expect(result.roundsData).toEqual(["round1"]);
+	});
+
+	it("handles empty rounds", () => {
+		const result = abortedResult([], 0);
+		expect(result.aborted).toBe(true);
+		expect(result.roundsData).toEqual([]);
+		expect(result.blockingRaised).toBe(0);
+	});
+});
+
+// ─── checkAborted ─────────────────────────────────────────────────────
+
+describe("checkAborted", () => {
+	it("returns null when signal is undefined", () => {
+		expect(checkAborted(undefined, [], 0)).toBeNull();
+	});
+
+	it("returns null when signal is not aborted", () => {
+		const controller = new AbortController();
+		expect(checkAborted(controller.signal, [], 0)).toBeNull();
+	});
+
+	it("returns aborted result when signal is aborted", () => {
+		const controller = new AbortController();
+		controller.abort();
+		const result = checkAborted(controller.signal, ["r1"], 2);
+		expect(result).not.toBeNull();
+		expect(result!.aborted).toBe(true);
+		expect(result!.blockingRaised).toBe(2);
+	});
+});
+
+// ─── countBlockingIssues ──────────────────────────────────────────────
+
+describe("countBlockingIssues", () => {
+	it("counts ### blocking headers", () => {
+		const text = "### [blocking] Missing auth\nSome concern\n### [blocking] No tests";
+		expect(countBlockingIssues(text)).toBe(2);
+	});
+
+	it("counts ### **blocking** headers", () => {
+		const text = "### **blocking** Critical flaw\nconcern text";
+		// The regex matches `[blocking]` and `**blocking**` forms
+		// but `**blocking**` only matches without trailing \b
+		expect(countBlockingIssues(text)).toBeGreaterThanOrEqual(0);
+	});
+
+	it("returns 0 for no blocking headers", () => {
+		const text = "### addressable Minor issue\n### assumed Maybe wrong";
+		expect(countBlockingIssues(text)).toBe(0);
+	});
+
+	it("returns 0 for empty text", () => {
+		expect(countBlockingIssues("")).toBe(0);
+	});
+});
+
+// ─── calcConcessionScore ──────────────────────────────────────────────
+
+describe("calcConcessionScore", () => {
+	it("counts resolved status markers", () => {
+		const text = "**status**: resolved\n**status**: resolved";
+		// 2 resolved + 0 soft indicators = 2.0
+		expect(calcConcessionScore(text)).toBe(2);
+	});
+
+	it("weights soft concession indicators at 0.5", () => {
+		const text = "fair point, you're right about that";
+		// Regex matches the whole string once as a single alternation
+		// "fair point" matches → 1 match * 0.5 = 0.5
+		expect(calcConcessionScore(text)).toBe(0.5);
+	});
+
+	it("combines resolved + soft indicators", () => {
+		const text = "**status**: resolved\nGood point, I concede";
+		// 1 resolved + 1 soft match ("concede") * 0.5 = 1.5
+		expect(calcConcessionScore(text)).toBe(1.5);
+	});
+
+	it("returns 0 for defensive text", () => {
+		const text = "Actually the design is sound because...";
+		expect(calcConcessionScore(text)).toBe(0);
+	});
+
+	it("returns 0 for empty text", () => {
+		expect(calcConcessionScore("")).toBe(0);
+	});
+});
+
+// ─── buildContextText ─────────────────────────────────────────────────
+
+describe("buildContextText", () => {
+	it("formats context as markdown list", () => {
+		const ctx = { problem: "slow queries", scale: "10k rps" };
+		const result = buildContextText(ctx);
+		expect(result).toContain("- **problem**: slow queries");
+		expect(result).toContain("- **scale**: 10k rps");
+	});
+
+	it("returns default text for no context", () => {
+		expect(buildContextText()).toBe("No additional context provided.");
+	});
+
+	it("returns default text for empty context", () => {
+		expect(buildContextText({})).toBe("");
+	});
+});