What this resource does
Core uses
One job-search toolkit, several task modes
Career-Ops uses one shared Skill router with separate modes and supporting scripts. The four modes below are a selection. Only text mode was tested in this review.
Four documented modes described; one mode tested in one English example.
text
Rewrites your summary, reorders experience and selects relevant projects for a supplied job description, using your existing CV as the factual source.
- Typical inputs
- Your current CV; The job description and your target role; Your preferred output language
- Output
- A job-specific Markdown CV and a short summary of sections and keyword coverage.
Best for: Preparing a readable CV draft for one vacancy.
Component for this task
text
Rewrites your summary, reorders experience and selects relevant projects for a supplied job description, using your existing CV as the factual source.
Suggested requests (1)
Run career-ops text mode for this job description, using my current CV. Write the tailored CV in English without inventing experience.
Adapted into complete requests from official trigger wording in modes/text.md.
scan
Searches the configured job portals and company career pages, filters titles and adds new matches to a job pipeline for later evaluation.
- Typical inputs
- Target companies and career-page addresses; Job-title filters and search preferences
- Output
- A pipeline of discovered vacancies, with source links.
Best for: Checking a defined set of hiring sources rather than searching the entire job market.
Component for this task
scan
Searches the configured job portals and company career pages, filters titles and adds new matches to a job pipeline for later evaluation.
Suggested requests (1)
Run the career-ops scan mode and summarize new matches.
From docs/CODEX.md.
interview/practice
Asks one interview question at a time, follows up on your answer and gives feedback grounded in your recorded experience.
- Typical inputs
- The role and interview context; Your CV and prepared examples; Your answers during the practice session
- Output
- A practice transcript and feedback on individual answers.
Best for: Rehearsing how to explain your experience before an interview.
Component for this task
interview/practice
Asks one interview question at a time, follows up on your answer and gives feedback grounded in your recorded experience.
tracker
Displays recorded applications and their statuses, updates a status when requested and summarises the application pipeline.
- Typical inputs
- Your application tracker; Confirmed replies, interviews or other status changes
- Output
- An updated tracker and summaries of recorded application activity.
Best for: Keeping applications organised without treating a drafted CV as a submitted application.
Component for this task
tracker
Displays recorded applications and their statuses, updates a status when requested and summarises the application pipeline.
Suggested requests (1)
Run the career-ops tracker mode and summarize the current statuses.
From docs/CODEX.md.
Use boundaries
Limits and checks
Check every experience claim
A smoother CV can still drop a qualification or overstate an achievement.
Compare it with your original CV before sending it. Confirm numbers, job titles and the limits of your experience.
Your AI provider processes your material
The official README says your CV is sent to the AI provider you choose, even though project files are stored locally.
Remove unnecessary personal or confidential information and check your application settings before sharing a CV.
Other modes need more setup
Finding live jobs and producing PDFs can require browser tools or additional software. Optional account integrations have separate requirements.
Start with a supplied CV and job description in text mode. Review any extra installation or account request before continuing.
More boundaries
- Keyword matching and AI job-fit scores do not establish your chance of being hired.
- The text-mode example does not test live job search, interview practice, PDFs or application submission.
Technical details
- Resource type
- Toolkit
- Author or maintainer
- Santiago Fernández de Valderrama Aparicio
- Version
- 1.32.0
- Latest release
- career-ops-v1.32.0
- Source last updated
- 7 Sept 2026
- Last verified
- 8 Sept 2026
- Licence
- MIT
- Access
- Publicly available
- Resource price
- Free
- Skill instruction language
- English
- Documentation language
- English
- Repository languages
- JavaScript, TypeScript, Go, HTML, Shell, CSS, TeX, Dockerfile, Nix
- Dependencies
- Node.js >=18; Git; Configured AI coding application; Playwright/Chromium for PDF generation (not required by text mode)
- Review status
- One CV example tested
Continue exploring
Related resources
Official package contents
Files
Read the official English instructions without leaving this profile. This view includes the English files represented by this profile. Non-English source files, supporting code, assets, tests, and references may not be shown.
Official package files
8 files available here
.agents/skills/career-ops1
docs1
modes/interview1
modes3
Read on this page
Select a file
Choose a file from the tree to read its official instructions without leaving this page.
Independent testing
Evaluation report
See exactly what was tested, which checks were used, and what the result does not cover.
Evaluations are mainly based on automated AI-assisted checks. Results are for reference only.
One example
Tailoring a data analyst CV without inventing missing skills
Career-Ops turned a fictional data analyst CV into a version tailored to a community-services vacancy. It highlighted relevant experience and kept the checked facts, but removed one explicit limit on that experience.
Codex CLI · Model not recorded · 8 Sept 2026
What we asked it to do
Use an existing CV to prepare an English application for a Data and Insight Analyst job. Highlight the relevant experience without adding skills or achievements the candidate does not have.
Exact request used in the test
Run career-ops text mode. Tailor my cv.md for the Data and Insight Analyst vacancy at Riverton Community Trust in documents/job-description.md. Write the CV in English, keeping my existing section structure and using only my actual experience. Save the tailored Markdown CV and give the short completion summary described by text mode. Do not browse, contact anyone or submit an application.
What was provided
A fictional CV for Alex Morgan and a fictional community-services vacancy. The candidate has SQL, Power BI and survey-analysis experience; the vacancy also mentions Azure, Tableau and predictive modelling, which the CV does not claim.
Original input
Fictional test materials. All names, organisations and contact details are invented. ORIGINAL CV # Alex Morgan Manchester, UK | [email protected] ## Professional Summary Data analyst with experience preparing service reports, cleaning operational data and explaining findings to non-technical colleagues. Interested in using data to improve public services. ## Skills - Analysis: Excel, SQL (PostgreSQL), Python (pandas), descriptive statistics - Reporting: Power BI dashboards, accessible written summaries, stakeholder workshops - Working practices: data-quality checks, Git, documented analysis steps ## Professional Experience ### Data Analyst | Northbridge Community Services | September 2023 - Present - Prepared monthly service reports for 12 community centres using Excel and SQL. - Cleaned 18,400 appointment records and documented missing-value and duplicate checks. - Built a Power BI dashboard used by 6 service managers to review appointment attendance. - Reduced report preparation time from 6 hours to 4 hours per month by reusing SQL queries; the service manager confirmed these timings over three monthly reporting cycles. - Presented findings at a workshop with 8 colleagues and wrote a two-page summary for staff without an analytics background. ### Research Assistant | Westford University | October 2022 - August 2023 - Analysed 240 completed survey responses in Python and checked summary tables against spreadsheet calculations. - Co-authored an internal briefing about survey response patterns. The study was descriptive and did not establish that any service caused better outcomes. ## Projects ### Appointment attendance dashboard - Combined appointment and service-location tables using SQL and displayed attendance trends in Power BI. This was an internal reporting project, not a predictive model. ### Community survey summary - Used pandas to clean survey responses and produce descriptive summaries; documented missing responses and small group sizes. ### Reporting handover guide - Wrote a step-by-step guide explaining SQL query inputs, quality checks and monthly report updates so colleagues could repeat the process. ## Education MSc Social Research Methods | Westford University | 2022 BA Sociology | Westford University | 2021 JOB DESCRIPTION # Riverton Community Trust: Data and Insight Analyst Location: Manchester, UK. Hybrid working. This is a fictional vacancy for a test, not an open job. We want an analyst who can turn service data into clear information for operational colleagues. You will develop Power BI dashboards, write SQL queries, carry out data-quality checks and explain findings to non-technical stakeholders. You will document repeatable reporting processes and work with service managers to understand their questions. Essential experience: - SQL for joining and summarising operational datasets. - Power BI dashboards and monthly reporting. - Data cleaning, including missing values and duplicate checks. - Clear written communication, stakeholder workshops and accessible summaries. - Descriptive statistics and careful interpretation of observational evidence. - Process improvement and documentation. Desirable, but not essential: - Microsoft Azure data services. - Tableau. - Predictive modelling. Please provide an English CV showing relevant experience and projects. Formal line-management experience is not required.
Output from the Skill
Career-Ops wrote the CV below for the fictional candidate Alex Morgan. This is its complete, unedited output.
Alex Morgan
Manchester, UK | [email protected]
Professional Summary
Data analyst focused on community service reporting, using SQL, Power BI dashboards and data-quality checks to turn operational data into clear information for non-technical colleagues. Prepared monthly reporting for 12 community centres and built an attendance dashboard used by 6 service managers. Brings experience in accessible written summaries, stakeholder workshops and documenting repeatable reporting processes. Interested in using data and insight to improve public services.
Skills
- Analysis: SQL (PostgreSQL), Excel, descriptive statistics, Python (pandas)
- Reporting: Power BI dashboards, monthly reporting, accessible written summaries, stakeholder workshops
- Working practices: data cleaning and data-quality checks, missing values and duplicate checks, process improvement, documentation of repeatable reporting processes, Git
Professional Experience
Data Analyst | Northbridge Community Services | September 2023 - Present
- Prepared monthly service reports for 12 community centres using Excel and SQL to summarise operational data.
- Built a Power BI dashboard used by 6 service managers to review appointment attendance.
- Cleaned 18,400 appointment records and documented data-quality checks for missing values and duplicates.
- Presented findings at a workshop with 8 colleagues and wrote an accessible two-page summary for staff without an analytics background.
- Reduced report preparation time from 6 hours to 4 hours per month by reusing SQL queries; the service manager confirmed these timings over three monthly reporting cycles.
Research Assistant | Westford University | October 2022 - August 2023
- Analysed 240 completed survey responses in Python using descriptive statistics and checked summary tables against spreadsheet calculations.
- Co-authored an internal briefing about survey response patterns, interpreting observational evidence carefully. The study was descriptive and did not establish that any service caused better outcomes.
Projects
Appointment attendance dashboard
- Joined appointment and service-location tables using SQL and displayed attendance trends in Power BI for internal reporting.
Community survey summary
- Used pandas to clean survey responses and produce descriptive summaries; documented missing responses and small group sizes.
Reporting handover guide
- Wrote a step-by-step guide documenting SQL query inputs, quality checks and monthly report updates so colleagues could repeat the reporting process.
Education
MSc Social Research Methods | Westford University | 2022 BA Sociology | Westford University | 2021
Our assessment
Relevant experience comes first
The opening paragraph now highlights SQL, Power BI and community-service reporting. It also moves the dashboard and workshop experience higher, making the relevant work easier to find.
Facts retained; missing skills not invented
The checked facts stayed intact: 18,400 appointment records, 240 survey responses, and a reduction from 6 to 4 hours per month over three reporting cycles. It did not invent Azure, Tableau or predictive-modelling experience to fill the job requirements.
One boundary was dropped
The original CV explicitly said the dashboard was not a predictive model. The rewrite still calls it internal reporting, but removes that explicit boundary. Put it back if it matters for explaining your level of experience.
What the 85% match actually means
Its 85% match means 17 of 20 keyword concepts selected by the agent appeared in the CV. It does not mean an 85% chance of an interview or that a recruiter's system has accepted it.
A setup limitation in this test
The CV was produced despite a missing software dependency during the initial setup check. We had not installed the full project dependencies in this temporary checkout. This test therefore does not verify a complete first-time installation.
How this example was scored
90/100 for this one example. Five points were deducted for recoverable errors in the run, including our incomplete dependency setup; five for dropping the explicit non-predictive-dashboard caveat. The complete CV below is the actual output, not an edited demonstration. One English CV rewrite using the official text mode in Codex CLI. No real application was sent. Other Career-Ops modes are not covered by this score.
Five scoring dimensions
Passed checks receive 1 point, partly met checks 0.5, and failed checks 0. Checks outside this example are excluded, not counted as passes. The five dimension percentages are combined using the weights shown below.
Did it complete the task?
100.0%A complete English Markdown CV was saved at the documented path. All five original sections remain in order; there are no tables, HTML or code fences. The summary foregrounds service reporting, SQL and Power BI. Dashboard and workshop bullets move ahead of the reporting-time bullet. All three relevant projects remain.
Did it run without problems?
75.0%The CV was saved in one invocation. The initial setup check failed because js-yaml was absent from this temporary checkout; a first self-check assertion also failed before a corrected check passed. These are observed run/setup issues, not a failure to produce the CV. All five supplied files and tracked source files were unchanged. Retained commands show local reading, CV creation and checks, not an application submission or PDF status update. This is not a network-security assessment.
Did it preserve the supplied information?
75.0%Identity, employers, dates, degrees and all checked quantities are preserved, including 18,400 records, 240 responses and the 6-to-4-hour change over three cycles. It does not add Azure, Tableau or predictive-modelling experience and preserves the descriptive-study limitation. However, it drops the original explicit sentence that the dashboard was not a predictive model; internal reporting remains stated.
Did it do the documented task in this setup?
100.0%The recorded commands read the pinned official router, text mode, shared context, candidate files and required writing guidance. The actual output follows text mode; no replacement workflow was used. The completion summary gives the path, five sections, 17/20 (85%) and three unmatched skills. Its twenty keyword concepts and matching strings are visible in the execution log. This is a count chosen by the agent, not a hiring probability or certified ATS score.
Were the setup and use instructions clear?
100.0%The official router, text-mode guide and supplied candidate files identify the inputs and output without invented installation commands. The README separately documents dependency installation; this test did not exercise that installer.
Limits of this example
- Live job search, company research and job-fit scoring (scan, discover, oferta, ofertas and auto-pipeline).
- PDF/LaTeX exports, cover letters, interview practice, application tracking and application forms.
- External API integrations and account connections: not tested; contact us to request a future test.
- Other AI applications, languages, real applicants and hiring outcomes.
- Complete first-time installation and setup checks: the temporary checkout did not have all npm dependencies installed.
Want to see more tests?
Tell us which module or task you would like us to test next.