Guide

Everything you need to get started, in about five minutes.

What it is

Career Memory keeps a quiet record of the work you do with Claude Code, and turns it into things that help your career:

a memory per project: what you built, hard problems solved, decisions, stack, and when;
a timeline of your work, month by month;
resume updates you can review and apply, plus prompt-based editing;
blog post ideas and drafts, and LinkedIn suggestions;
a weekly email with what’s new.

There’s nothing to log or review. You keep working; it remembers.

How it works

1
On your computer
A Claude Code plugin reads the sessions in folders you opt in to. It strips code, tool output, secrets and personal details, then uses your own Claude Code to pick out what's worth remembering and keeps a condensed memory per project.
2
Synced to your account
Only those condensed memories and short work items are sent here. Never your code or transcripts.
3
Turned into suggestions
The app uses your memory, work history and resume to suggest resume updates, blog ideas and LinkedIn edits. Nothing changes until you apply it.

Set up Claude Code

You need Claude Code on macOS or Linux, with python3 (3.9 or newer). In Claude Code, run these one at a time:

1. Install the plugin

/plugin marketplace add Wert1996/career-memory-plugin

/plugin install career-memory

/reload-plugins

2. Choose what to capture, and connect

/career-memory:setup ~/code --backfill-days 30

Use the folder that holds your repos. Backfill also reads your last 30 days of sessions.

/career-memory:connect

Opens this site to approve. Click Approve, and you're connected.

Only sessions in the folder you choose (and its subfolders) are read. Add more later with /career-memory:scope allow <path>, or put an empty .careerignore file in any repo to skip it. Using more than one computer? Repeat these steps on each.

3. In the app

Import your resume (PDF or DOCX). Your history before the plugin starts there, and it seeds your work history.
On Career, check your work history and confirm your current job, so new work is credited to the right employer.

What happens next

The plugin processes sessions in the background, at most twice a day, after a session has been idle for 45 minutes. It runs on your own Claude Code; it never slows you down.
With --backfill-days, your first memories appear shortly after your next Claude Code reply. Otherwise they build up as you work. Run /career-memory:run to process now.
New memories sync to this site automatically, and appear on Memory and the Timeline.
On Memory, set which job each project belongs to (or mark it personal), then refresh your career profile.
Ask for resume suggestions and blog ideas any time, or wait for Monday’s email.

Using the app

Feed: Your overview: setup steps, recent timeline, memory updates, pending resume suggestions and blog ideas.
Timeline: Your work month by month, across every project and job.
Resume: Import your current resume, then review suggested updates from your memory. Edit with prompts ("shorten the summary") and see a live preview. Pick a template and share a public link.
Memory: One condensed memory per project, plus a career profile: your signature strengths and notable work. Set which job each project belongs to.
Blog: Post ideas from the hard problems and decisions in your memory. One click drafts a post you can edit and publish.
LinkedIn: Paste your LinkedIn profile to get suggested improvements, grounded in what you've actually built.
Career: Your work history (the source of truth for your current job), target roles, saved job descriptions and a skills analysis.
Settings: Public handle, plugin connections, this month's AI usage, data export and account deletion.

Steering what’s remembered

Tell it in plain English how to read your work. In Claude Code:

/career-memory:instructions treat repos api and web as one project called Checkout

/career-memory:instructions skip anything from my dotfiles and homelab

/career-memory:instructions emphasise system design; don't name the client Contoso

This writes ~/.career-memory/instructions.md, which you can also edit by hand. A .careermemory.md in a repo adds instructions for that repo. New sessions follow them right away; /career-memory:memory rebuild and regroup re-apply them to what’s already remembered.

Projects: one product can span several repos. New repos join an existing project automatically when they’re part of the same work; /career-memory:memory regroup reorganises everything, and merge FROM INTO does it by hand.

Plugin commands

/career-memory:setup [path] [--backfill-days N]: Opt in for a folder. Nothing is captured before this.
/career-memory:connect: Link this computer to your account (approve in the browser).
/career-memory:status: What's captured, memory counts, next run, connection.
/career-memory:run: Process finished sessions now instead of waiting. --dry-run shows the exact redacted text that would be read.
/career-memory:memory [project]: Show project memories. regroup, merge FROM INTO, rebuild and forget manage them.
/career-memory:instructions [change]: Show or edit your plain-English instructions for how sessions are read.
/career-memory:scope allow|exclude <path>: Change which folders are captured. work <path> --org <Employer> tags employer work. pause / resume.
/career-memory:timeline: Your timeline in the terminal.
/career-memory:uninstall: Delete all local data.

Weekly email and limits

Every Monday morning, if your memory changed that week, you get fresh resume suggestions and blog ideas, and an email listing them. Quiet weeks send nothing.

Career Memory is free while invite-only, with monthly limits on AI features (they reset on the 1st; see Settings for your usage):

resume edits by prompt10
suggestion refreshes5
career profile refreshes1
resume imports1
blog drafts1
tailored resumes1
application answers1

The plugin’s processing runs on your own Claude Code, so it doesn’t count.

Questions

Is my code uploaded?

No. Code, tool output and file contents are dropped on your computer before anything is read, and secrets, URLs, emails and similar details are scrubbed. Only short summaries of what you built are synced.

What about work for my employer?

Only capture what you're allowed to record. Choose folders deliberately, add .careerignore to sensitive repos, and use instructions to keep client or project names out. Memories describe internal systems generically.

Does it use my Claude Code quota?

A little: a few short calls when it processes sessions, at most twice a day.

Can I see exactly what's read?

Yes: /career-memory:run --dry-run prints the exact redacted text for each session, and changes nothing.

How do I remove something?

Forget items with /career-memory:memory forget, or delete them in the app. You can download or delete everything from Settings, and /career-memory:uninstall removes local data.

Is there a browser extension?

Not yet for invitees. For now, paste your LinkedIn profile and job descriptions into the app.

More detail: privacy policy. Questions or problems: apairs.anshumaan@gmail.com.