career-manager
Skill24 starsChanged 42 days agoMITView on GitHub
Orchestrate the complete career-manager system across job-search, company-research, job-tracker, and cv-tailor. Use when coordinating multi-step job hunt workflows, routing requests to the right sub-skill, and enforcing shared data ownership boundaries.
In this repository9 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- company-researchSkill
- cv-tailorSkill
- evalsSkill
- interview-prepSkill
- job-searchSkill
- job-trackerSkill
- onboardingSkill
The file
We split this skill into the rules it gives an agent
12 rules. About 0 are common, 8 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈8
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 12
Before starting any workflow, run the briefing script to understand current state:
- Distinctive: line 18
Read the JSON output and present a natural-language summary to the user:
- Distinctive: line 23
Use the briefing to inform which workflow to suggest.
- Distinctive: line 27
"What should I do today?" / "what's my status?" / "briefing": - Run briefing, then suggest highest-priority action: follow-ups first, then …
- Distinctive: line 39
After any skill completes, suggest the logical next action (one suggestion, not a chain):
- Very specific: line 63
Every Agent tool dispatch in the SKILL.md files states its model explicitly.
- Distinctive: line 72
Keep non-career work out of this skill.
- Very specific: line 73
Use company-research for one-company analysis, not bulk pipeline operations.
- Very specific: line 74
Use job-tracker for application status operations, not market/company research.
- Distinctive: line 78
If a sub-skill path or file is missing, stop and report the missing dependency.
- Distinctive: line 79
If requested action spans multiple skills, execute sequentially (don't ask for permission between each step unless a decision is needed).
- Very specific: line 80
If data sources conflict, prefer canonical owners (job-search for target companies, job-tracker for applications) and log the discrepancy.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Career Manager Orchestrator
This is intentionally a router skill (thin coordinator), not a heavy execution skill. Execution lives in sub-skills below.
Before Any Workflow: Status Briefing
Before starting any workflow, run the briefing script to understand current state:
uv run python3 scripts/generate_briefing.pyRead the JSON output and present a natural-language summary to the user:
- First run (first_run=true): "This is your first run. I'll search across your career paths and probably find 15-25 companies. Should take about 20 minutes."
- Ongoing (first_run=false): "You have [total_companies] companies tracked, [stale_companies] are stale. [active_applications] active applications, [followup_needed] need follow-up. [Context-appropriate time estimate]."
Use the briefing to inform which workflow to suggest.
Route by Intent
- "What should I do today?" / "what's my status?" / "briefing" - Run briefing, then suggest highest-priority action: follow-ups first, then monitor if stale companies exist, then prospecting if >7 days since last.
- Target list scoring / hygiene / dedupe / stale checks / broken links -
job-search - "Run job search" / "find companies" / "run the pipeline" -
job-search - Single company deep research dossier -
company-research - "Pursue [company]" / "research and apply to [company]" -
company-researchthencv-tailorthenjob-tracker - Application pipeline CRUD + follow-up reporting -
job-tracker - First-time setup / profile update / regenerate config -
onboarding - "Tailor my CV" / "apply to [role]" -
cv-tailor - "Prep me for the [company] interview" / "interview prep for [company]" -
interview-prep
Cross-Skill Flow
After any skill completes, suggest the logical next action (one suggestion, not a chain):
| Skill completes | Suggest |
| Onboarding | Auto-start job-search pipeline (no prompt needed) |
| Job search phase 3 | "Top 3 are [names]. Want me to research them?" |
| Company research says PURSUE | "Want me to tailor your CV for [role] at [company]?" |
| CV tailor produces artifacts | "Want me to add this to your tracker?" |
| Job tracker shows stale follow-ups | "Want me to draft follow-up text for these?" |
| Job tracker status becomes interviewing | "Want an interview prep doc for [company]?" |
These are suggestions. The user can always decline.
Model Policy
Route by work type, not by habit:
| Work | Model | Why |
| Prospecting, eval, monitor agents (job-search) | sonnet | Bulk search + rubric scoring; volume dominates cost (~500K tokens/wave) |
| Company research dossiers | sonnet | Structured extraction from web sources |
| CV tailoring, onboarding, interview-prep drafting | session default | Human-facing writing; quality over cost |
| Independent review passes (e.g. interview-prep review) | sonnet | Structured checking against explicit criteria |
Every Agent tool dispatch in the SKILL.md files states its model explicitly. If a dispatch instruction has no model, that is a bug — fix the instruction, don't guess.
Source of Truth
job-search/data/target-companies.csvjob-tracker/data/applications.csv
Boundaries
- Keep non-career work out of this skill.
- Use
company-researchfor one-company analysis, not bulk pipeline operations. - Use
job-trackerfor application status operations, not market/company research.
Error Handling
- If a sub-skill path or file is missing, stop and report the missing dependency.
- If requested action spans multiple skills, execute sequentially (don't ask for permission between each step unless a decision is needed).
- If data sources conflict, prefer canonical owners (
job-searchfor target companies,job-trackerfor applications) and log the discrepancy.
Reference
references/apply-workflow.md- end-to-end research, tailor, apply, track, follow-up flow
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.