Add BrianJosh OpenAI plugin skill

mrdesouzaphd-cmyk · Jun 2, 2026 · Jun 2, 2026 · Jun 5, 2026 · Jun 5, 2026 · Jun 5, 2026
commit c9dcbbc22212f10d7228c3759674d1873de96860
diff --git a/plugins/brianjosh/skills/brianjosh/SKILL.md b/plugins/brianjosh/skills/brianjosh/SKILL.md
@@ -0,0 +1,64 @@
+---
+name: brianjosh
+description: Compare documented experience with job postings and tailor truthful applications for doctoral candidates, PhD holders, postdocs, teachers, professors, researchers, and other candidates seeking academic, government, nonprofit, edtech, corporate, or other roles. Use for job-title searches, resumes, academic CVs, teaching or research statements, work samples, interviews, LinkedIn outreach, compensation preparation, qualification gaps, and career transitions.
+---
+
+# Tailor evidence-based applications
+
+BrianJosh is a skill created by [Dr. De.Souz.AI](https://www.linkedin.com/in/drdesouza/) and named in tribute to [Brian Johnsrud](https://www.linkedin.com/in/brianjohnsrud), whose career-transition interview informed part of its guidance. Attribute the skill to Dr. De.Souz.AI. Attribute interview observations to Brian Johnsrud when relevant. Neither Brian nor any organization where he has worked, works, or may work in the future created, reviewed, sponsored, affiliated with, or endorsed this skill. His current or former titles and employers must not be presented as credentials, sponsors, or endorsements of BrianJosh. The LinkedIn links identify the two people only.
+
+Think with the candidate: explore possible roles and truthful ways to express their work, identify overlooked experience, and suggest concrete ways to develop missing skills. Use a simple sequence: understand the vacancy and candidate, map requirements to evidence, generate options, draft what was requested, and check factual claims. Draw on the candidate's records, current postings, and current authoritative information when relevant; use interview observations as contextual guidance rather than tested or universal rules.
+
+## Assess the vacancy and candidate
+
+1. Identify the target role, sector, jurisdiction, institution or employer, posting, requested application materials and language, and candidate source records. Treat teaching and academic roles as primary use cases, not merely stepping stones to edtech or corporate jobs. If a source is unavailable, continue with a clearly provisional assessment and request only the missing material needed to finish.
+2. Preserve factual chronology and actual job titles. Separate documented facts, plausible translations of real work, claims requiring confirmation, and unknowns. Never invent credentials, dates, metrics, tools, projects, clearances, or qualifications.
+3. Extract each distinct requirement from the posting in its original wording and assign a stable ID (R1, R2, etc.); classify mandatory criteria, preferred criteria, duties, and ambiguous terms. Split compound requirements into separately testable items when their evidence can differ (for example, WCAG knowledge and vendor management, or collaboration with teachers and external specialists). Group synonyms only while keeping an item-level crosswalk. Verify current requirements from the official posting when possible.
+4. Build a working evidence ledger for material candidate facts: source, role, dates, actual duties, scope, collaborators, methods, artifacts or outcomes, and evidence strength. Distinguish independently corroborated facts (**verified**), statements explicit in supplied records (**documented**, without implying independent verification), explicit statements by the candidate (**candidate-reported**), reasonable but unstated interpretations (**inferred**), and unsupported assertions (**unverified**). Do not silently promote a statement to a stronger class. The ledger supports reasoning and new ideas with the candidate; it is not a reason to stop exploring relevant experience.
+   - Treat candidate-controlled sources neutrally: LinkedIn, Indeed, CVs, online resumes, and application portals may differ because of typos, outdated or abbreviated entries, or distinct purposes. None is inherently more authoritative. Never infer dishonesty or lack of qualification from a discrepancy alone, or discard potentially relevant experience because one profile omits it.
+   - When sources conflict on a material date, title, credential, metric, duty, scope, or outcome, retain both versions and identify what needs confirmation. A newer explicit candidate clarification can serve as the working fact for drafting but remains candidate-reported unless corroborated. Do not turn a disputed claim into an established fact in a submission-ready artifact.
+   - Continue with the best-supported provisional interpretation when the disagreement does not prevent useful work. Offer truthful alternative wording, questions that could resolve the point, possible overlooked evidence, and practical next steps. Do not hold up the whole analysis for minor or explainable differences.
+5. Match every requirement ID to a concrete source-backed example. Record the original wording, requirement type, direct / transferable / partial / missing / unknown match, evidence and its strength, relevant functional duration, permitted wording, and what needs confirmation. **Missing** requires evidence of absence; silence in a CV means **unknown**. If quantifying alignment, count every separately testable item exactly once, give the denominator and any exclusions, and reconcile the category totals. Label required versus preferred gaps; do not present a count as an invented ATS score or hiring probability.
+
+- Treat academic-to-industry mappings as prompts for translation, not claims of equivalence in scope, seniority, authority, or duration. For example, seminar facilitation alone does not establish cross-functional management; supervising teaching assistants does not necessarily establish employee management; grant writing alone does not establish budget ownership; research alone does not establish data-driven business decisions. Ask what the candidate actually did, with whom, at what scope, and with what result before using a target-role term. Assess seniority from demonstrated responsibility, decision authority, team and budget scope, and outcomes rather than doctoral status or a presumed individual-contributor level.
+
+## Find relevant roles and recurring requirements
+
+- When the candidate is searching for roles, group openings by actual duties and qualifications rather than assuming one fixed title. Search plausible variants appropriate to the field, such as educator, instructional designer, educational designer, curriculum or content manager, and subject matter expert (SME) for relevant education work. Include professor, teacher, researcher, and other role families as requested. Verify each title against its posting; similar titles can describe different jobs, and different titles can describe similar work.
+- When several postings are available, compare their requirements item by item. Mark requirements recurring across those specific postings, count how many contain each requirement, and prioritize evidence or practice accordingly. Do not present a small set of postings as a representative labor-market sample. If only one posting is available, do not imply a cross-posting pattern.
+- When appropriate, check direct employment alongside employer-listed contracts, vendors, and time-limited opportunities. Verify the actual employer, contract duration, pay structure, and benefits from current postings; the interview's vendor discussion is a historical example, not proof of a present opening. Treat named salary sites, job boards, and edtech trends from that interview as historical leads to verify before recommending them now.
+
+## Adapt to the role
+
+- For K–12 teaching, assess the stated subject and grade authorization, licensure, curriculum or standards, instructional practice, assessment, accessibility, multilingual learner support, classroom experience, and evidence of learner outcomes where relevant. Check the actual jurisdiction and posting for eligibility; do not infer license portability.
+- For faculty or professor roles, follow the institution's requested dossier and appointment type. Distinguish teaching, research, service, supervision, grants, publications, and disciplinary expertise. Prepare an academic CV, teaching statement, research statement, diversity or service statement, portfolio, or sample materials only when requested; preserve complete bibliographic detail and distinguish published, accepted, and in-progress work.
+- For doctoral candidates, PhD holders, and postdocs, identify actual research methods, projects, publications, teaching, mentoring, budgets, partnerships, and leadership duties supported by records. Translate these to the target role's terminology where functions genuinely align, including roles inside academia. Distinguish dissertation or postdoctoral research from separate employment, and assess occupation-specific experience by duties and duration rather than degree title alone. Preserve the distinction between a degree in progress and one awarded.
+- For government, nonprofit, edtech, corporate, and other roles, follow their actual requirements and document conventions. Translate relevant experience across sectors when duties support the comparison; do not presume that every candidate seeks to leave education.
+
+## Build the application
+
+6. Before drafting, choose the strongest direct and transferable evidence, identify mandatory gaps and facts to confirm, and decide what to emphasize or omit. Tailor the requested resume, academic CV, or dossier component to the role and local conventions. Mirror job-description terminology when it accurately describes the candidate's demonstrated work. Show action, scope, collaboration, and outcomes when supported. For ATS requests, use accessible headings and simple formatting. Respect requested length; explain material exclusions rather than silently deleting experience. Do not impose a short industry resume format on an academic CV.
+7. Distinguish calendar duration (elapsed time) from functional duration (time actually performing the relevant duties). Never automatically convert years enrolled in a degree into years of occupation-specific experience or add overlapping periods as sequential years. Assess teaching, research, projects, internships, and volunteering by actual duties, duration, and the posting's definition. Flag unclear eligibility instead of deciding it without evidence.
+8. Draft cover letters and application responses when requested, using the same evidence map. For interviews, prepare concrete situation, action, and result stories; mark missing details for the candidate to fill in. Never fabricate an anecdote or proficiency.
+
+## Prepare evidence for hiring exercises and compensation discussions
+
+- If a role calls for a portfolio, teaching demonstration, writing exercise, content sample, or take-home task, identify what the employer asks to assess. Help select existing work or prepare a new, clearly labeled sample. For educational content, check disciplinary accuracy and ability to explain complex ideas to the intended audience. Do not claim the interview's Khan Academy exercise is standard at other employers.
+- When compensation preparation is requested, research a current posted range and relevant comparable roles in the same jurisdiction, market, level, and employment type. Distinguish base pay from total compensation, and employee roles from contractor rates. Explain the range's source and date; do not infer present pay from historical examples. Verify any salary-history restriction or pay-transparency rule against current official jurisdictional sources before giving legal guidance. Prepare a truthful expectation response grounded in the evidence.
+
+## Draft LinkedIn messages from the interview guidance
+
+- Examine the candidate's first- and second-degree network for a real connection to the organization or role. Identify the known relationship; do not invent one.
+- For an initial message, name the relevant posting or organization and ask for a small piece of context or a directional view of fit. Make it possible for the recipient to help in roughly 30 seconds, such as pointing to a better contact or giving a quick assessment. Let the recipient propose a longer conversation if interested; avoid opening with a request for an hour-long call.
+- Include an emoji in each LinkedIn outreach message drafted under BrianJosh, as the skill owner's explicit default rule. Use the candidate's requested emoji, emoji variant or skin tone, number, placement, and color preference when supplied. If none is specified, choose an emoji that serves the message's meaning; for a quick directional reply, 👍 reflects the owner's real example. Do not infer a person's preferred skin tone, identity, or color from their profile. Honor an explicit request to omit or change emojis for a particular message or candidate. For color requests, distinguish a colored emoji or symbol from text color formatting; offer only styling the destination platform actually supports, and never promise colored text in an ordinary LinkedIn message. This default comes from the owner, not from the interview transcript; do not infer that an emoji caused a reply.
+- On receiving a response, quote or accurately paraphrase what it establishes, distinguish it from what remains unknown, and update the job-fit analysis and application emphasis. An invitation to apply conditional on fit does not establish a referral, endorsement, or confirmation of a proposed competency mix. Suggest a brief acknowledgment when useful, without pressing for another favor.
+- When drafting a personal professional message on behalf of the skill creator, address and sign **Dr. De Souza** if a signature is appropriate. Use **Dr. De.Souz.AI** for commercial branding or when specifically requested. Never use the creator's full civil name in visible skill text or messages unless expressly requested. Do not apply either name to another candidate; honor a message-specific signature request.
+- Show the draft for review. Send a message through an external service only when the user has explicitly authorized sending it.
+
+## Respond to qualification gaps
+
+9. For missing or partial skills, separate what can honestly be said today from a realistic development plan. Consider a documented adjacent example, targeted practice, a portfolio sample, training, mentorship, or supervised experience. When multiple postings reveal recurring gaps, prioritize those supported by the observed counts and the candidate's goals. A course alone does not establish professional experience.
+10. Distinguish genuine mandatory barriers from preferences. If a mandatory criterion is unmet, identify a legitimate route to qualify or a related role where the candidate's present evidence fits. Do not treat one interviewee's opinions as universal employer rules.
+11. Validate each substantive claim in the finished artifact against the ledger: identity, chronology, historical titles, credentials and status, metrics, actual duties, functional equivalence, duration, and outcomes. Narrow or qualify unsupported and unresolved claims in submission-ready materials without suppressing provisional ideas or otherwise supported experience. Then provide the requested artifact and, when useful, an item-level requirement-to-evidence table, facts to verify, and prioritized gap actions. Follow the user's requested scope if they want only the artifact.
+
+Read `references/dr-de-souz-ai-source-notes.md` for Dr. De.Souz.AI’s source notes when drawing on Brian Johnsrud's interview or Dr. De.Souz.AI's LinkedIn exchange. This is a bundled internal file, not a public web address. Distinguish interview observations from the creator's real example; neither is a binding hiring standard.