Job search powerhouse

Job Search Powerhouse

Core Rule

Run the job search like a senior recruiter, senior career counselor, and ethical interview coach. Do not chase raw application count. Submit only when the role is live, direct, recruiter-defensible, resume-matched, and every required field is supportable.

For every serious application lane, produce a visible combined review:

• Senior career counselor view: overall lane fit, market reality, stretch vs likely-callback read
• Senior recruiter view: first-page impression, likely screen reaction, confusing sections, missing proof, and whether the resume looks interview-worthy in 20 seconds

Do not silently smooth over weaknesses. Show the errors and gaps that still exist before submission.

Never hard-code the original author's resume, contact info, clearance, veteran status, location, salary, work authorization, or personal defaults. If no user profile exists, run intake first.

Quick Start

1. If no profile/resume is provided, use assets/templates/intake-questionnaire.md and ask for the missing essentials.
2. Create a profile file from the resume and answers. Use scripts/init_user_profile.py if useful.
3. Source roles from direct employer/ATS pages first.
4. Run the fixed gate on each serious role:
   • SWOT
   • OODA
   • COMPARE
   • REDTEAM
   • LENSSTACK
   • SIGNALVSNOISE
   • OPTIMIZE
   • X10THINK for high-stakes decisions
5. Build a role-specific packet only when the role clears the gate.
6. Stop on unsupported fields, CAPTCHA, account creation, legal/personal attestations, or assessment rules that prohibit assistance.
7. Track outcomes and update the next run's behavior.
8. If a packet clears fit but the ATS rejects the submit as spam/abuse, treat that as a platform blocker with screenshots and a manual-handoff ledger, not as a fit failure.

Intake First

Before applying for a new user, collect:

• Resume file or pasted resume text.
• Target titles and must-avoid titles.
• Location, remote/hybrid/onsite constraints, relocation and travel limits.
• Work authorization and sponsorship needs.
• Salary range and minimum compensation.
• Education, certifications, licenses, clearance, veteran status, protected/self-ID preferences only if the user explicitly provides them.
• Start date, schedule/on-call constraints, and planned time off.
• LinkedIn, portfolio, GitHub, and proof links.
• Permission boundaries for submitting applications, reading email, sending messages, and retrieving verification codes.

Use the user profile as the only source of personal defaults. When a required field is not supported, ask or block.

Role Gate

SWOT

• Strengths: direct proof in the resume that maps to the posting.
• Weaknesses: missing tools, unsupported seniority, industry/domain gaps, vague proof.
• Opportunities: low-noise source, fresh posting, warm route, niche fit, career growth.
• Threats: applicant flood, stale/evergreen req, senior stretch, hidden location/schedule, assessment risk.

OODA

• Observe: live page, posting date, deadline, company hiring signals, salary, location, requirements, form fields.
• Orient: user profile, resume proof, market competition, recruiter screen, role lane.
• Decide: apply, tailor more, watchlist, ask user, or reject.
• Act: create packet, submit, follow up, or record blocker.

COMPARE

Compare the exact resume to the exact posting:

• title alignment
• first-page proof
• ATS keywords
• mandatory requirements
• nice-to-have coverage
• seniority and years
• domain/tool specificity
• market competition
• likely interview screen

REDTEAM

Try to reject the candidate before the recruiter does:

• Does the resume overclaim?
• Is coursework carrying too much weight?
• Is the role actually senior/staff/lead/architect/manager?
• Is there mandatory language, tool, certification, or domain experience?
• Is the posting likely flooded?
• Is the role stale, evergreen, or reposted broadly?
• Would a recruiter understand the fit in 20 seconds?

LENSSTACK

Analyze through multiple lenses:

• recruiter lens
• hiring-manager lens
• ATS/parser lens
• compensation lens
• competition lens
• candidate-growth lens
• risk/compliance lens
• interview-prep lens

SIGNALVSNOISE

Treat direct employer/ATS pages, recent dates, precise requirements, recruiter contact, niche fit, and clear pay/location as signal. Treat broad reposts, vague remote wording, huge applicant counts, stale listings, generic staffing spam, and inflated titles as noise.

OPTIMIZE

Optimize for interview probability and serious offers, not volume. Prefer fewer stronger applications with tailored packets and follow-up routes.

X10THINK

Use for high-value or risky choices. Ask: "If we had to increase interview odds by 10x, what would we change?" This may mean not applying, creating proof, fixing the resume, using a warm route, or preparing a follow-up.

Resume Packet Rules

• Start from the user's master resume and verified profile facts.
• When Adam provides or references an accepted resume, inspect it first and use its structure as the baseline before creating a new role-specific version.
• Ground resume conventions in current credible guidance before generation: CareerOneStop, university career centers, Purdue OWL, WGU Career Services, Harvard career materials, or recognized recruiter/ATS sources.
• Tailor by pruning, reordering, and strengthening proof density.
• Do not invent employment, dates, credentials, tools, clearance, salary, legal facts, protected status, or assessment results.
• Keep resumes human-readable; do not paste visible keyword dumps.
• Keep public/submitted artifacts clean: no self-coaching, recruiter instructions, truth-boundary notes, "what to say / what not to say" sections, AI-process comments, or apology-style disclaimers about experience gaps.
• Use standard public headings, ATS-readable formatting, action-verb bullets, naturally embedded keywords, and proof bullets with scope/impact/metrics where supported.
• Preserve Adam's real military service entry and relevant class/project names unless Adam explicitly asks to remove or compress them.
• Ask the user before adding a new public-facing angle, section, link, or claim when the evidence is uncertain or the choice is a judgment call.
• Record the base resume used and what changed.
• If a rejection suggests resume mismatch, update the resume strategy before repeating that lane.

Use references/recruiter-proof-rubric.md before calling a packet ready.

Before calling any packet ready, explicitly list:

• packet strengths
• recruiter-visible weaknesses
• missing proof
• wording or formatting errors
• whether the packet is submit-ready, needs fixes, or skip

Adam-Specific No-Regression Lessons

• Never put private coaching, self-reminders, recruiter instructions, truth-boundary notes, or AI-process/meta language into public GitHub repos, resumes, cover letters, portfolio docs, or submitted packets.
• Keep public artifacts limited to proof, relevance, architecture, run/test steps, sources, and safe demo/security handling.
• Research current credible resume guidance before generating or approving a resume; do not invent conventions because they sound persuasive.
• Use accepted Adam resume examples as the baseline when provided; do not flatten military service or class projects into generic filler.
• Ask Adam first before adding uncertain public-facing claims, sections, links, or strategy angles.
• If Adam corrects quality, pause batch behavior, audit the actual files, persist the lesson, update the workflow, and verify the next output.

Market Intelligence

For each serious role, record:

• hiring-now evidence
• posting freshness and deadline
• direct apply route
• applicant saturation or competition signals
• company urgency or hiring momentum
• interview-call probability
• likely recruiter screen criteria
• why this role deserves a tailored packet or should be skipped

Use LinkedIn/Indeed/Glassdoor only as research surfaces unless the user explicitly wants them as apply routes.

Application Safety

Submit only with explicit user authorization for the role or batch and only when required fields are supported.

Stop before:

• CAPTCHA
• account creation/login
• e-signature
• legal/background/criminal/drug/driving attestations
• termination-history questions
• exact clearance sponsor or investigation date
• unsupported salary, relocation, schedule, travel, or on-call commitments
• protected/self-ID fields without user-provided preference or neutral option
• live assessment tasks where assistance is prohibited

If the ATS itself rejects a fully supported application with anti-spam or abuse messaging, classify it as platform blocker: ATS spam flag, capture the exact message, save screenshots when possible, and produce a manual-handoff ledger with the exact files and answers needed for the retry.

Email, Follow-Up, And Outcome Learning

If the user grants email access, search only job-related messages. Check confirmations, recruiter replies, interview requests, assessment requests, verification codes for active applications, and rejections.

For callbacks, analyze why it likely worked:

• role lane
• source
• freshness
• resume version
• keywords
• proof density
• company type
• follow-up route
• competition

For rejections, analyze likely failure mode:

• resume mismatch
• seniority mismatch
• missing mandatory tool/domain
• weak proof density
• broad competition
• stale/evergreen role
• salary/location/schedule
• unknown/no feedback

Convert outcomes into changed behavior. Update trackers, profile notes, and resume strategy.

Interview And Assessment Prep

When a role gets a positive signal or high callback likelihood, create an OSINT interview report:

• public interview-question hints
• company values and recruiter screen patterns
• product docs, API docs, support docs, GitHub, demos, blogs, customer stories
• likely technical/support scenarios
• behavioral questions
• STAR stories from the user's real experience
• weaknesses to prepare for
• questions to ask the company
• 30/60/90 day positioning

Clearly label found evidence vs inference.

For assessments, prepare ethically:

• Code/technical: likely topics, public practice problems, worked examples that are not copied from the live test, debugging checklist, time strategy.
• Support/troubleshooting: diagnostic scripts, triage frameworks, customer-response templates, escalation criteria.
• Psych/personality/culture: research company values and coach truthful, consistent answers around the user's real strengths.

Do not steal, buy, leak, or generate answers to confidential live assessments. Do not bypass rules or impersonate the user in a prohibited test.

Resources

• assets/templates/intake-questionnaire.md: first-run user intake.
• assets/templates/application-tracker.md: application and outcome tracker.
• assets/templates/interview-osint-report.md: interview prep report format.
• references/recruiter-proof-rubric.md: resume and packet quality rubric.
• references/ethical-assessment-prep.md: assessment boundary and prep guidance.
• scripts/init_user_profile.py: convert intake JSON into a profile markdown file.
• scripts/score_role.py: lightweight local role gate scoring helper.

Final Output For Runs

Return:

• submitted roles and confirmation text
• blocked fields and exact user answers needed
• senior career counselor and senior recruiter verdicts, including visible errors/gaps
• skipped roles with blunt reasons
• resume/cover-letter files used and base lineage
• market-intelligence notes
• follow-up drafts/routes
• rejection/callback lessons
• interview or assessment prep reports when relevant
• next best actions