for coding agents · free, open, no auth

Your agent can turn every duty into a results-driven achievement narrative, embellish. Teach it to screen.

One install gives your agent 13 job-search commands, every finding run through 5 recruiter filters: speed, relevance, evidence, risk, clarity. What it reads is what a screener would see.

13 commands 5 recruiter filters ~2,500 lines of instruction 0 API keys

install

Copy one line. Get screened first.

Detects Claude Code, Cursor, Codex, Qoder and more, then writes the skill into the right folder. Add --dry-run to see the plan first.

npx skills add arm-pz/recruiter-lens

see it work

The screener's eye, on paper.

Real shapes of what each command returns. Pick a command.

input: resume, top third

Senior Software Engineer · Jan 2019 – Present
Led cross-functional initiatives and contributed to architectural decisions. Experienced with multiple frameworks and tools; passionate about clean code.

what the first pass caught

  • evidenceThree duty verbs, zero outcomes. "Led", "contributed", "experienced" say the candidate was present, not that anything changed.
  • speedStack, domain and seniority are not findable at a glance. Three facts a screener hunts for should be the three that land first.
6–8 s first pass: a reported heuristic, not a rulenamed filter per finding

input: one bullet

Responsible for managing the migration of legacy systems to cloud infrastructure for the department.

critique, in the candidate's own wording

  • evidenceManage the migration of what, from what, to what, with what result? Name the systems; a screener can't map vagueness onto their stack.
  • clarityStart with the verb, not the job description. "Migrated the department's [N] legacy services to [platform]" is the same claim, screened in half a second.
every fix shown on your wording"make this stronger" is not a finding

input: weak bullet

Improved application performance significantly, earning recognition from leadership.

impact-boost

Rewritten shape: "Cut checkout latency by [metric: ?] on the store's main flow." Then one question back: what did latency actually drop to?

  • risk"Earning recognition from leadership" is an unverifiable soft claim. It reads as padding and costs credibility on the bullets around it.
numbers come from youno metric, no guess: a question or a [metric: ?]

the full vocabulary

13 commands, 5 filters.

Or skip the commands. Ask "why would this resume get rejected?" in plain language and the skill routes it, naming the command it picked.

Analysis 5

See the material the way the screener does.

  • scan
  • critique
  • ats-check
  • reject-reasons
  • linkedin-audit

Preparation 4

Walk in already knowing what they'll ask.

  • predict-questions
  • gap-analysis
  • tailor
  • ai-guidelines

Optimization 4

Better presentation of real experience, never misrepresentation.

  • format-fix
  • keyword-match
  • impact-boost
  • red-flag-review

house rules

What the lens never does.

Invent nothing

Numbers, titles, employers and dates come from the candidate. A missing metric gets a question or a visible [metric: ?], never a guess.

Tend, not rule

Screening behavior is reported tendency, not insider knowledge. Region, industry and seniority change what applies, and the skill says when.

Stay honest

No claim that would collapse under interview follow-up. Gaps and career changes get a brief, honest context line, never concealment.

Know what they see in the first six seconds.

Free, open, and it runs entirely inside the agent you already use.

Install recruiter-lens Read the source