myagent-ai 1.0.0

package/skills/image-understand/scripts/image-understand.ts

@@ -0,0 +1,41 @@
+import ZAI, { VisionMessage } from 'z-ai-web-dev-sdk';
+
+async function main(imageUrl: string, prompt: string) {
+	try {
+		const zai = await ZAI.create();
+
+		const messages: VisionMessage[] = [
+			{
+				role: 'assistant',
+				content: [
+					{ type: 'text', text: 'Output only text, no markdown.' }
+				]
+			},
+			{
+				role: 'user',
+				content: [
+					{ type: 'text', text: prompt },
+					{ type: 'image_url', image_url: { url: imageUrl } }
+				]
+			}
+		];
+
+		const response = await zai.chat.completions.createVision({
+			model: 'glm-4.6v',
+			messages,
+			thinking: { type: 'disabled' }
+		});
+
+		const reply = response.choices?.[0]?.message?.content;
+		console.log('Image Understanding Result:');
+		console.log(reply ?? JSON.stringify(response, null, 2));
+	} catch (err: any) {
+		console.error('Image understanding failed:', err?.message || err);
+	}
+}
+
+// Example usage - analyze an image
+main(
+	"https://cdn.bigmodel.cn/static/logo/register.png",
+	"Please analyze this image and describe what you see in detail."
+);

package/skills/interview-designer/README.md

@@ -0,0 +1,70 @@
+# Interview Designer
+
+**Evidence-Based Interview Planning**
+
+Design interview questions using Scorecard → Forensic Scan → Future Simulation. Avoid confirmation bias and produce structured interview guides (Scorecard + Red Flags/Green Signals + Pressure Tests + Future Scenarios) with Geoff Smart, Lou Adler, and Daniel Kahneman as the default expert panel.
+
+---
+
+## When to Use This Skill
+
+Use this skill when:
+
+- You need to design interview questions for a specific role
+- You want to avoid confirmation bias in interview planning
+- You're creating a structured interview guide (Scorecard + Questions + Pressure Tests)
+- You need to balance past validation with future simulation
+
+---
+
+## Methodology: Scorecard → Forensic → Future
+
+| Phase           | Expert        | What You Define                                                                 |
+|-----------------|---------------|-----------------------------------------------------------------------------------|
+| **1. Scorecard**| Geoff Smart   | Mission, Outcomes, Competencies — *before* looking at any resume                 |
+| **2. Forensic Scan** | Smart + Domain | Resume gaps vs. highlights; "Too Good To Be True" / "Driver vs Passenger" heuristics |
+| **3. Future Simulation** | Lou Adler  | Performance problems the candidate would face in your context; week-one scenarios |
+
+---
+
+## What You Get
+
+| Output            | Template                           | Purpose                                              |
+|-------------------|------------------------------------|------------------------------------------------------|
+| **Interview Guide** | `templates/interview_guide_template.md` | Scorecard + Red Flags/Green Signals + Pressure Tests + Future Scenarios |
+
+The guide includes both concerns (**Red Flags**) and highlight verification (**Green Signals**) for objective assessment.
+
+---
+
+## Design Principles
+
+1. **Cannot Be Memorized** — Questions force real-time thinking (simulation) or concrete recall (pressure test).
+2. **Forced Trade-offs** — Choose between two "correct" options to surface values, not just knowledge.
+3. **Detail Granularity** — Probe to "what exact words did you say" or "what diagram did you draw."
+
+---
+
+## Quick Reference
+
+| Interview Goal      | Question Type       | Example                                                                 |
+|---------------------|---------------------|-------------------------------------------------------------------------|
+| Validate past claims| Pressure Test (STAR) | "Walk me through the specific metrics you tracked and how you used them." |
+| Predict future fit  | Future Simulation   | "Here's our Q1 challenge. How would you approach it in your first week?" |
+| Detect blind spots  | Trade-off Question  | "Speed vs. quality — which would you sacrifice here, and why?"         |
+
+---
+
+## Install
+
+**ClawHub (OpenClaw)**:
+```bash
+npx clawhub@latest install interview-designer
+```
+
+**Other (e.g. skills.sh)**:
+```bash
+npx skills add mikonos/interview-designer
+```
+
+Compatible with Cursor, Claude Code, OpenClaw, and other agents that support the skills protocol.

package/skills/interview-designer/SKILL.md

@@ -0,0 +1,53 @@
+---
+name: interview-designer
+description: Analyze resumes and design interview strategies using evidence-based methodology. Transforms interview prep from "read resume → ask questions" into "define standard → forensic evidence → future simulation". Combines Geoff Smart's Topgrading, Lou Adler's performance-based hiring, and Daniel Kahneman's bias control. Use when preparing for interviews, creating structured interview guides, or designing questions to validate candidate competencies.
+---
+
+# Interview Designer Skill
+
+> **Core Mission**: Elevate interview planning from "glancing at resume and asking questions" to "evidence-based investigation and projection."
+> **Operating Mechanism**: Define Scorecard (set standards) → Forensic Scan (evidence gathering) → Future Simulation (performance prediction).
+> **Prompt Strategy**: This skill uses \<Chain of Thought\>. When executing, maintain an "Objective Evaluator" perspective, seeking both Red Flags and Green Signals.
+
+## 1. Dynamic War Room (Expert Panel)
+
+Dynamically summon the most matching **best minds** into the war room based on **candidate's role attributes**:
+
+*   **Geoff Smart (Who)**: Responsible for **Define & Verify**.
+    *   *Principle*: Scorecard First. Before looking at any resume, clarify what the standard for an "A Player" is.
+*   **Lou Adler (Performance-based)**: Responsible for **Predict**.
+    *   *Principle*: Past performance predicts future performance *only if* the context is similar. Must design simulations for future scenarios.
+*   **Daniel Kahneman (Bias Control)**: Responsible for **De-bias**.
+    *   *Principle*: Beware of "confirmation bias." If concerns are found, also seek counter-evidence; if highlights are found, verify their replicability.
+*   **Domain Expert**: Responsible for **Depth**.
+
+## 2. Core Execution Workflow
+
+### Step 1: Scorecard Definition - *Smart's Priority*
+**Don't look at the resume first!** Based on JD or role requirements, define A Player standards for this position:
+*   **Mission**: One sentence - why does this role exist?
+*   **Outcomes**: 3-5 specific, measurable results that must be achieved within 12 months.
+*   **Competencies**: Hard/soft skills required to achieve the above outcomes.
+
+### Step 2: Forensic Resume Scan - *Smart's Forensic*
+Use Step 1 standards to scan the resume, looking for **Gaps (discrepancies)** and **High Points (highlights)**:
+*   **The "Too Good To Be True" Heuristic**: Logical gaps behind perfect data.
+*   **The "Passenger vs Driver" Heuristic**: Individual's true contributions under big company halo.
+*   **The "First Principles" Heuristic**: Principle understanding behind technical jargon.
+
+### Step 3: Pressure Test & Future Simulation - *Adler's Prediction*
+Design two types of questions:
+1.  **Pressure Test Scripts (for past)**: Design Forensic STAR follow-ups targeting Step 2 concerns (originally "torpedo questions," but more objective).
+2.  **Future Simulation (for future)**: Design a specific Performance Problem.
+    *   *Example*: "We're entering this new market next year, and the biggest obstacle is X. If you join, how would you analyze this problem in your first week?"
+
+## 3. Question Design Principles
+
+1.  **Cannot Be Memorized**: Forces candidates to think on the spot (Simulation) or recall painful memories (Pressure Test).
+2.  **Forced Trade-offs**: Choose between two "correct" options to test values.
+3.  **Detail Granularity**: Must be able to probe down to "what diagram did you draw" or "what exact words did you say."
+
+## 4. Output Format
+
+Directly call `templates/interview_guide_template.md` to generate the report.
+**Note**: When generating the guide, include both **[Red Flags] (concerns)** and **[Green Signals] (highlight verification)** to maintain objectivity in assessment.

package/skills/interview-designer/_meta.json

@@ -0,0 +1,6 @@
+{
+  "ownerId": "kn73h39f2c8arzkcvwmafak9mn80wq7d",
+  "slug": "interview-designer",
+  "version": "1.0.0",
+  "publishedAt": 1770749339448
+}

package/skills/interview-designer/references/design_rationale.md

@@ -0,0 +1,43 @@
+# Expert Critique: Interview Designer Skill
+
+> **Simulated Review Panel**:
+> *   **Geoff Smart** (Author of "Who", Topgrading methodology)
+> *   **Lou Adler** (Founder of Performance-based Hiring)
+> *   **Daniel Kahneman** (Nobel Laureate, Behavioral Economics, Decision Noise Research)
+
+---
+
+## 1. Geoff Smart's Perspective: Only "Autopsy," No "Definition"
+*   **Comment**: "The starting point of this design is good (Forensic investigation), which aligns well with Topgrading's spirit—digging for truth. **However, you made a fatal error: the sequence is reversed.**"
+*   **Critique**: 
+    *   The current flow is `Resume Scan` → `Scorecard`. This is **reactive**. You're setting standards based on the candidate's resume, which is the trap of "creating positions around people."
+    *   **The A Method's** first step is always **Scorecard**—before looking at any resume, you must define the role's mission (Mission), outcomes (Outcomes), and competencies (Competencies).
+    *   **Risk**: Without an independent Scorecard first, your "investigation" becomes "nitpicking," not "validation of fit." You might prove they lied, but not that they can deliver.
+*   **Recommendation**: Mandate Step 0 as **"Define Success"**, not **"Scan Resume"**.
+
+## 2. Lou Adler's Perspective: Overemphasis on "Past," Neglecting "Future"
+*   **Comment**: "I see you're very enthusiastic about uncovering resume 'inflation.' That's interesting, but **can someone perform the job just because their resume is perfect?**"
+*   **Critique**:
+    *   Current `Torpedo Questions` mainly expose past lies.
+    *   **Performance-based Hiring** believes the best prediction is having candidates solve **future problems**.
+    *   **Gap**: Lacks **Project-based Problem Solving**. Beyond asking "how did you coordinate in the past," also ask "this is our new project, if you were responsible, what would you do in the first week?"
+*   **Recommendation**: Add **"Future Performance Simulation"** section.
+
+## 3. Daniel Kahneman's Perspective: Breeding Ground for Confirmation Bias
+*   **Comment**: "You're calling this skill 'Forensic' and throwing 'torpedoes.' This plants very strong **negative priming** in the interviewer's mind."
+*   **Critique**:
+    *   Once an interviewer enters the room with "this person might be lying" colored glasses, they'll unconsciously seek evidence to confirm this (Confirmation Bias), while ignoring the candidate's genuine highlights.
+    *   This is the source of **Noise**.
+*   **Recommendation**: Balance the mindset. Change "Torpedo" to **"Evidence Stress Test"**, and explicitly require seeking **"Green Signals"** simultaneously, not just red flags.
+
+---
+
+## 4. Comprehensive Optimization Recommendations (Action Plan)
+
+1.  **Architecture Adjustment (Re-order)**: 
+    *   `Step 1: Define Scorecard` (based on JD/business pain points, independent of resume)
+    *   `Step 2: Resume Forensic` (scan resume gaps based on Scorecard)
+2.  **Content Enhancement (Add Future Focus)**:
+    *   Add `Problem Solving Case` generation logic.
+3.  **Tone Correction (Neutrality)**:
+    *   Maintain sharpness, but remove "presumption of guilt" undertones. Goal is Truth-seeking, not Witch-hunting.

package/skills/interview-designer/templates/interview_guide_template.md

@@ -0,0 +1,62 @@
+# {Candidate_Name}_Targeted_Interview_Guide
+
+> [!IMPORTANT]
+> **Planning Context**: Based on war room simulation with {Expert_List}.
+> **Objective**: Verify competency match (Step 1), gather evidence on resume concerns (Step 2), project future performance (Step 3).
+
+---
+
+## 1. Competency Scorecard
+
+*Before looking at the resume, THIS is what success looks like.*
+
+*   **Mission (One-sentence mission)**: ...
+*   **Outcomes (12-month must-achieve results)**:
+    1.  ...
+    2.  ...
+*   **Core Competencies**:
+    *   **{Competency 1}**: {Description}
+    *   **{Competency 2}**: {Description}
+
+---
+
+## 2. Forensic Resume Scan
+
+*Scan resume against Scorecard, looking for Gaps (concerns) and Evidence (matches).*
+
+### 🔴 Red Flags (Concerns/Gaps)
+*   **{Concern 1}**: ...
+    *   *Expert Challenge*: ...
+*   **{Concern 2}**: ...
+
+### 🟢 Green Signals (Highlights/Matches)
+*   **{Highlight 1}**: ...
+    *   *Evidence*: ...
+
+---
+
+## 3. Interview Battle Scripts
+
+### Part A: Pressure Validation (Past Performance)
+*Forensic STAR follow-ups designed for Red Flags.*
+
+**Q1 (targeting {Concern 1})**:
+*   **The Setup**: "{Question...}"
+*   **The Drill**: "{Follow-up...}"
+
+### Part B: Future Projection (Future Scenario)
+*Performance simulations designed for Outcomes.*
+
+**Q2 (targeting Outcome 1)**:
+*   **Scenario**: "{Set up a specific challenge highly relevant to future work...}"
+*   **Question**: "If this is the situation you face in your first week, how would you handle it?"
+*   **Bar Raiser**: "{What would an A Player do?}" vs "{What would a B Player do?}"
+
+---
+
+## 4. Decision Matrix
+
+| Dimension | No Hire (Kill) | HIRE (Pass) |
+| :--- | :--- | :--- |
+| **Integrity** | ... | ... |
+| **Competency Match** | ... | ... |