Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add benskamps/career-compass-mcp --skill feedbackgit clone --depth 1 https://github.com/benskamps/career-compass-mcpWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/benskamps/career-compass-mcp/feedback)<a href="https://agentmods.dev/skills/benskamps/career-compass-mcp/feedback"><img src="https://agentmods.dev/badge/skills/benskamps/career-compass-mcp/feedback/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/benskamps/career-compass-mcp/feedback"><img src="https://agentmods.dev/badge/skills/benskamps/career-compass-mcp/feedback.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00048 | $0.00850 |
| Opus 5.5 | $0.00019 | $0.00340 |
| Sonnet 5.5 | $0.00010 | $0.00170 |
| Haiku 4.5 | $0.00005 | $0.00085 |
Grade A, and why
feedback scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 15d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
The user wants to report something Career Compass got wrong or could do better. Career Compass has no telemetry, so this report is the only way its maker hears about it. What they said is below.
- Know what happened. Use what they said and, when it happened in this conversation, what you can see of it. If you can't tell what went wrong, ask one thing: "What did it say, and what should it have said?" That is the whole reply.
- Draft the report in general terms. It will be a public GitHub issue, so it carries
no one's identity:
- No names, emails, phone numbers, addresses, links to their profiles, or salary figures.
- No employer, school, client or product names, theirs or the posting's. Describe them instead: "a mid-size insurer", "a senior data role at a fintech", "a state university".
- No pasted résumé, posting, email or card text. Paraphrase the part that matters ("it said I had no SQL; my résumé lists SQL under Skills"). The draft has four parts:
- Where: claude.ai chat (web, desktop or mobile), Cowork, Claude Code, another MCP app, or "Not sure". Only what they said; never guess it.
- What went wrong: the one closest option from this list, word for word: "The verdict was wrong (strong fit / stretch / long shot)", "It missed something I have, or said I lack something I have", "It said something about me that isn't true", "The gaps or advice weren't useful", "It asked too many questions, or the wrong one", "A draft (résumé, cover letter, reply) was off", "A role search listed bad, dead or made-up postings", "It forgot something, or the Career Compass card was wrong", "It talked about setup or tools I don't have", "Something else".
- Details: what they asked for, what it said, and what it should have said, in three to six plain lines. Only what they told you or what you can see: no theories about why it went wrong, no numbers you worked out, and no metrics from their résumé (say "a high-volume event pipeline", not its throughput).
- Version: with the Career Compass tools, the version
check_setupreports; else leave it out.
- Show it, then the link. Show the whole draft in one code block, then one line: "This is everything the report says. Change anything you like, then open https://github.com/benskamps/career-compass-mcp/issues/new?template=fit-check-wrong.yml, pick the matching options, paste the details into the box and press Submit (you'll need a free GitHub account). Nothing is sent unless you send it." Give the link exactly as written, with nothing added to it. They can also say it in Discussions instead.
- Fix it now if you can. If what went wrong happened in this conversation and you can see the right answer, give the corrected version after the link, in a line or two, or offer to redo it.
Don't open the link, submit, or post anything for them, and never add a detail they didn't give. If they'd rather not use GitHub, give the draft alone and stop.
$ARGUMENTS
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 15d ago First seen · 55 lines · 48 tokens per session scan A 1ba723eb7068
feedback is a skill published in the GitHub repository benskamps/career-compass-mcp (6 stars, last pushed 3d ago), licensed MIT. It adds 48 tokens to every session and 850 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-13.
Other skills, from other repositories
kochab
Kochab, a personalized resume-based job-search assistant that runs a recurring scan and everything downstream of it. Use when the user wants to start or run a job search, shares a resume, or says "set up my job scan," "find roles that fit my resume," or "what should I apply to." Also use to tailor a resume, draft a…
interviewing
Use when the user gets an interview invitation, moves into an interview process, wants to track interview stages, capture post-interview notes, draft follow-up messages, or update an application to interviewing.
get-started
Use when the user is new to Job Hunt Skills and wants to see what it does or set up their workspace — including "I'm new to Job Hunt Skills", "help me get started", "set me up", "first time using this", "here's my resume, tailor it for this job", "can you look at my resume", pasting a resume and a job posting…
Job Description Analyzer
Analyze job postings, calculate match scores, identify gaps, and create application strategy.
yc-job-tracker
Track and filter YC startup job listings for AI/ML engineering roles. Builds a daily-refreshed local database of relevant jobs with scoring and alerts for new postings.
Application Form Filler
Fill out job application form fields with context-aware, tailored answers drawn from the candidate's CV and the job description.