reggiechan74/jobops/createrubric

Configuration

Read .jobops/config.json. If missing, stop with:

JOBOPS NOT CONFIGURED

Run /jobops:setup to initialize your workspace.

Use config.directories.<key> for all file paths in this skill. Use config.preferences.cultural_profile if this skill generates resume-style content. Use config.preferences.default_jurisdiction if this skill has jurisdiction-sensitive logic (crisis/legal skills accept --jurisdiction=<ISO-3166-2> to override).

Templates

For each template used by this skill, resolve the full path as:

{config.templates.base_dir}/{config.templates.active.<template_name>}/<filename>

Templates referenced by this skill: assessment_rubric_framework

Application Path Resolution

This skill writes to a per-application folder. Before writing any output:

1. Parse {Company}_{Role}_{YYYYMMDD} from the job-posting filename, or honor --app=<slug> if supplied. The slug MUST be canonical: a leading PascalCase {Company} token (matching the Company_Intelligence/{Company}/ folder so OSINT links), a PascalCase {Role} (underscores between words allowed), and a trailing compact 8-digit date (20260519 — no hyphens, no time). Reject leading date/time prefixes such as 2026-04-15_214414_...; if the source filename carries one, recompose it into canonical form ({Company}_{Role}_{YYYYMMDD}) before composing the folder path.
2. Compose the app folder: {config.directories.applications_root}/{app_slug}/.
3. Resolve this skill's sub-folder by category:

• resume-development (buildresume, provenance-check) → resume/
• cover-letter (coverletter) → cover-letter/
• rubric / assessment (createrubric, assessjob, assesscandidate, auditjobposting) → assessment/
• briefing / interview prep (briefing, interviewprep) → interview/

1. If the app folder does not exist, mkdir -p it, then copy

{config.directories.job_postings}/{filename} → {app_slug}/job_posting.md so the pinned JD cannot silently change under completed work. Ensure the pinned copy begins with YAML front matter carrying output_type: job_posting: if the source JD already has a front-matter block, add the key to it; otherwise wrap a new block (--- / output_type: job_posting / source_jd: {filename} / ---) above the JD body.

1. Exact-slug collisions (same Company+Role+Date) are not auto-suffixed. If the folder

already contains the same output type, require the user to pass --app=<distinct-slug>.

1. Output filename (fixed; only the sub-folder above is resolved dynamically): write the rubric to assessment/rubric.md.

Your Task

Analyze the {{ARG1}} job posting and generate a detailed, reusable scoring rubric that extracts all requirements and creates standardized evaluation criteria.

WORKFLOW ARCHITECTURE

Phase 1 (Parallel batch):   Load templates + Load job posting (3 parallel reads)
Phase 2 (PARALLEL):         Domain Research (subagent) while main agent determines role variant
Phase 3 (Sequential):       Create rubric (synthesizes job posting + domain research)
Phase 4 (Sequential):       Validate and save rubric

Dependency Rules:

• Phase 2 starts after job posting is loaded (Phase 1)
• Domain research subagent and role variant analysis run concurrently
• Phase 3 WAITS for domain research results
• Phase 4 WAITS for rubric completion

PROGRESS TRACKING (MANDATORY)

Before starting any work, create all tasks for user visibility:

Task Update Rules:

• Mark each task in_progress BEFORE starting work on it
• Mark each task completed AFTER finishing it

YAML FRONT MATTER

Write the rubric to {applications_root}/{app_slug}/assessment/rubric.md and begin the file with:

---
job_file: {config.directories.job_postings}/{{ARG1}}
role: <role title>
company: <company name>
role_variant: <Technical IC | People Manager | Executive>
total_points: 200
generated_by: /createrubric
generated_on: <ISO8601 timestamp>
output_type: rubric
status: final
version: 2.0
---

Insert this before the first heading and bump version if you update the rubric later.

PHASE 1: LOAD INPUTS (Parallel batch)

Task: Mark task 1 in_progress.

Read all three files in a single parallel batch using three Read tool calls:

• {config.templates.base_dir}/{config.templates.active[assessment_rubric_framework]}/assessment_rubric_framework.md - Master 200-point rubric structure with role variants
• {config.templates.base_dir}/{config.templates.active[evidence_verification_framework]}/evidence_verification_framework.md - Evidence-based scoring protocols
• {config.directories.job_postings}/{{ARG1}} (add .md extension if needed)

If the job posting doesn't exist in {config.directories.job_postings}/, check the root directory for legacy files.

Task: Mark task 1 completed.

PHASE 2: PARALLEL ANALYSIS

Mark tasks 2 and 3 as `in_progress` simultaneously.

Dispatch domain research subagent AND begin role variant analysis in the SAME message.

2.1 Research Domain and Industry Context (Task 2 - Subagent)

Dispatch a domain research subagent to run while you analyze the role variant:

Use Task tool with subagent_type=general-purpose, model=sonnet, and prompt:
"Research the following for the role of [ROLE TITLE] at [COMPANY NAME]:
 
1. Industry standards and typical role expectations for this specific position
2. Required vs nice-to-have skills based on current market standards
3. Typical responsibilities and seniority indicators for this role level
4. Company context: culture, values, technology stack, recent developments, size, reputation
5. Current market conditions: salary ranges, demand, competitive landscape
6. Industry-specific terminology, certifications, and best practices
7. What differentiates strong vs average candidates for this type of role
 
Provide a structured research summary organized by these 7 areas.
Focus on actionable intelligence that would help calibrate a scoring rubric.
Be specific - cite sources and data points where possible."

2.2 Determine Role Variant and Extract Requirements (Task 3 - Main agent)

While the domain research subagent runs, analyze the job posting:

Determine Role Variant

Document the selected variant in the rubric YAML header.

Parse Job Posting to Extract Requirements:

• Hard Skills: All technical competencies with proficiency levels required
• Experience Requirements: Relevance, recency, and domain specifics (NOT years-based scoring)
• Impact Evidence: Demonstrated achievements, outcomes, and metrics
• Credentials: Required and preferred education/certifications
• Fit & Readiness: Communication, values alignment, and role readiness indicators

Task: Mark task 3 completed when extraction is done.

Task: Mark task 2 completed when domain research subagent returns.

PHASE 3: CREATE RUBRIC (Sequential - needs domain research)

Task: Mark task 4 in_progress.

Prerequisite: Domain research (task 2) must be completed.

3.1 Synthesize Job Posting + Domain Research

Use domain research findings to calibrate rubric thresholds:

• Set proficiency levels appropriate to industry standards
• Weight skills based on market reality, not just posting emphasis
• Identify implicit requirements the posting may not state
• Calibrate experience expectations to seniority level

3.2 Create Job-Specific Rubric Following Template Structure

MANDATORY DETAILED SCORING REQUIREMENT: You MUST create a comprehensive 200-point scoring rubric following the EXACT structure defined in the assessment rubric framework template. This is NON-NEGOTIABLE.

CRITICAL: Use the assessment rubric framework template as your MANDATORY reference template. The rubric structure is fixed - you must maintain:

1. Five Main Categories (default point allocations - adjust based on role variant):

• Skills Inventory (50 points default / 25%)
• Experience Relevance (40 points default / 20%)
• Demonstrated Impact (60 points default / 30%)
• Credentials (20 points default / 10%)
• Fit & Readiness (30 points default / 15%)

1. Role Variant Weight Adjustments:

1. Proficiency-Based Skill Scoring (7-level scale, NOT years-based):

• 6 - Expert: Can teach/design novel solutions; recognized authority
• 5 - Advanced: Solves complex problems independently; mentors others
• 4 - Proficient: Handles standard work with minimal guidance
• 3 - Competent: Performs with occasional supervision
• 2 - Developing: Requires regular guidance; growing capability
• 1 - Novice: Basic understanding; needs significant support
• 0 - None: No demonstrated capability

1. Mandatory Rubric Components:

• Alignment Statement: Construct definition explaining what the rubric measures
• Critical Barriers Table: Minimum thresholds with consequences for failure
• Anchor Examples: Concrete examples for each scoring level
• Confidence Flagging: Guidance for marking low-confidence scores
• Weight Justification Table: Rationale for category weight allocations

1. Customize Job-Specific Content:

• Replace [bracketed placeholders] with actual job requirements
• Extract specific skills, technologies, and responsibilities from job posting
• Define role-specific thresholds and criteria (scope, budget size, team size, etc.)
• Add company-specific values and cultural indicators
• Include industry-specific success metrics and KPIs

1. Maintain Template Components:

• Overall assessment ranges for each section
• Evaluation frameworks with specific metrics
• Evidence-based scoring verification protocols
• Usage guidelines and quality control checklists
• Scoring interpretation guidelines

Task: Mark task 4 completed.

PHASE 4: VALIDATE AND SAVE

4.1 Validate Rubric Completeness

Task: Mark task 5 in_progress.

ENFORCEMENT CHECK: After creating the rubric, verify:

• Role variant selected and documented in YAML header
• Point allocations match selected variant (total = 200)
• Five categories present: Skills, Experience, Impact, Credentials, Fit
• Alignment Statement with construct definition included
• Critical Barriers table with thresholds and consequences
• Skills use 7-level proficiency scale (0-6), NOT years-based
• No redundancy (years only in Experience, achievements only in Impact)
• Anchor examples for each scoring level
• Confidence flagging guidance included
• Weight justification table present
• Evidence verification framework is included