Display Name
career-ops
Category
Tooling
Sub-Category
Tooling: Orchestrators
Primary Link
https://github.com/santifer/career-ops
Author Name
santifer
Author Link
https://github.com/santifer
License
MIT
Other License
No response
Description
career-ops is an AI-powered job search orchestrator built on Claude Code: 14-skill pipeline that evaluates JDs, generates ATS-tailored PDFs, batches outreach, and tracks applications.
Companies use AI to filter candidates. career-ops gives candidates AI to choose companies.
46K stars (top 500 GitHub), 9.7K forks, 3.000+ Discord, MIT, local-first (CV never leaves machine).
Site: https://career-ops.org
Validate Claims
Three validation paths, in increasing investment:
30 seconds — Verify the metrics from terminal (no install):
gh api repos/santifer/career-ops --jq '{stars: .stargazers_count, forks: .forks_count, license: .license.spdx_id, created: .created_at}'
gh api 'search/repositories?q=topic:job-search&sort=stars&per_page=3' --jq '.items[] | "\(.stargazers_count) ⭐ \(.full_name)"'
Confirms: 46K+ stars, 9.7K forks, MIT license, created April 2026, and that career-ops is #1 by stars in the #job-search topic globally (≈10× ahead of #2).
2 minutes — Browse the 14-skill orchestration in the GitHub web UI (no install):
https://github.com/santifer/career-ops/tree/main/.claude/skills/career-ops
Each skill mode has its own SKILL.md with the prompt, scope, and use case. The orchestrator logic that composes them is in the same tree.
5 minutes — Read the quickstart, no install required:
https://github.com/santifer/career-ops#readme
The README documents setup (git clone + npm install + playwright install + Anthropic API key), the full 14-skill mode list, the Go dashboard, and end-to-end workflows for scan / evaluate / batch / pdf / tailor / track.
Optional 15 minutes for full local verification:
Clone, set ANTHROPIC_API_KEY, run npm run dev -- evaluate <any-job-url>. The skill produces a score (0–5), an ATS-optimized PDF in ./output/, and never sends the local CV anywhere — only the candidate vector embedding is sent to the Anthropic API for evaluation. Network calls can be audited via claude --verbose (no third-party endpoints, no telemetry).
The local-first claim ("CV never leaves the machine") is enforced by the architecture: every skill reads CV from ~/cv.pdf or ~/profile.md locally, computes embeddings on-device, and only transmits vectors + JD text to Anthropic. There is no remote storage of personal data in the codebase — grep -r "telemetry\|analytics\|tracking" . returns zero matches.
External press validation:
Specific Task(s)
Install career-ops following the README, then ask Claude Code to evaluate a single real job posting end-to-end against your own CV/profile. The demo proves three properties at once:
- Candidate-side ATS-style scoring using the same vector machinery employers use to filter candidates
- Automatic generation of an ATS-optimized PDF resume tailored to that specific job
- The "CV never leaves the machine" claim — only embeddings are sent to api.anthropic.com, verifiable via
claude --verbose or any system network monitor
Total time after install: ~90 seconds to see all three outputs.
Specific Prompt(s)
I have career-ops installed and my profile at ~/career-ops/profile.md. Use the /career-ops evaluate skill mode against this job URL: https://job-boards.greenhouse.io/anthropic [or any public job posting]. Return:
- The alignment score on the 0-5 scale (threshold for "apply" set at 4.0)
- The three highest-impact ATS keyword gaps between my profile and the JD
- The path to the tailored PDF resume generated in ./output/
- Confirmation that no network calls were made outside api.anthropic.com
Additional Comments
career-ops emerged from my own job search after looking for a job — the 14 skill modes mirror workflows I actually used during 631 evaluated postings, dozens generated PDFs, and one signed offer (Head of Applied AI). A protocol spec (SHVP draft 0.1 coming Q3 2026) is being designed on top together with the Discord community (3,000+ devs), aiming to standardize candidate-side AI hiring infrastructure across implementations ahead of EU AI Act enforcement (Aug 2026).
Thanks for maintaining this list.
Recommendation Checklist
Display Name
career-ops
Category
Tooling
Sub-Category
Tooling: Orchestrators
Primary Link
https://github.com/santifer/career-ops
Author Name
santifer
Author Link
https://github.com/santifer
License
MIT
Other License
No response
Description
career-ops is an AI-powered job search orchestrator built on Claude Code: 14-skill pipeline that evaluates JDs, generates ATS-tailored PDFs, batches outreach, and tracks applications.
Companies use AI to filter candidates. career-ops gives candidates AI to choose companies.
46K stars (top 500 GitHub), 9.7K forks, 3.000+ Discord, MIT, local-first (CV never leaves machine).
Site: https://career-ops.org
Validate Claims
Three validation paths, in increasing investment:
30 seconds — Verify the metrics from terminal (no install):
Confirms: 46K+ stars, 9.7K forks, MIT license, created April 2026, and that career-ops is #1 by stars in the
#job-searchtopic globally (≈10× ahead of #2).2 minutes — Browse the 14-skill orchestration in the GitHub web UI (no install):
Each skill mode has its own SKILL.md with the prompt, scope, and use case. The orchestrator logic that composes them is in the same tree.
5 minutes — Read the quickstart, no install required:
The README documents setup (git clone + npm install + playwright install + Anthropic API key), the full 14-skill mode list, the Go dashboard, and end-to-end workflows for scan / evaluate / batch / pdf / tailor / track.
Optional 15 minutes for full local verification:
Clone, set ANTHROPIC_API_KEY, run
npm run dev -- evaluate <any-job-url>. The skill produces a score (0–5), an ATS-optimized PDF in./output/, and never sends the local CV anywhere — only the candidate vector embedding is sent to the Anthropic API for evaluation. Network calls can be audited viaclaude --verbose(no third-party endpoints, no telemetry).The local-first claim ("CV never leaves the machine") is enforced by the architecture: every skill reads CV from
~/cv.pdfor~/profile.mdlocally, computes embeddings on-device, and only transmits vectors + JD text to Anthropic. There is no remote storage of personal data in the codebase —grep -r "telemetry\|analytics\|tracking" .returns zero matches.External press validation:
Specific Task(s)
Install career-ops following the README, then ask Claude Code to evaluate a single real job posting end-to-end against your own CV/profile. The demo proves three properties at once:
claude --verboseor any system network monitorTotal time after install: ~90 seconds to see all three outputs.
Specific Prompt(s)
I have career-ops installed and my profile at ~/career-ops/profile.md. Use the /career-ops evaluate skill mode against this job URL: https://job-boards.greenhouse.io/anthropic [or any public job posting]. Return:
Additional Comments
career-ops emerged from my own job search after looking for a job — the 14 skill modes mirror workflows I actually used during 631 evaluated postings, dozens generated PDFs, and one signed offer (Head of Applied AI). A protocol spec (SHVP draft 0.1 coming Q3 2026) is being designed on top together with the Discord community (3,000+ devs), aiming to standardize candidate-side AI hiring infrastructure across implementations ahead of EU AI Act enforcement (Aug 2026).
Thanks for maintaining this list.
Recommendation Checklist