Skip to content
SkillMixed3,182 tokens in use

Commenter Talent Scan

Pull every commenter on a public hiring post, enrich their profiles in bulk and rank them as hire targets, peers or parked accounts with an eval-gated classifier.

3,862 stars on the repository MIT Licenseericosiu/ai-marketing-skills · commenter-talent-scan
npx skills add ericosiu/ai-marketing-skills -s commenter-talent-scan
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing Commenter Talent Scan costs

28tokens idle
3,182tokens in use

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Counts are approximate, made offline with cl100k_base.

Model dependence Mixed

Script-ledMixedModel-led

Scripts do part of the work and the model's judgement does the rest.

This describes the skill's structure, how much of it is scripts. It does not say how well it works. Scripts: references/sieve-domain-template/build_states.py, references/sieve-domain-template/route.py.

Touches

shelldetectednetworkdetectedwrites filesdetected

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Excellent
  • Frontmatter8/15
  • Structure25/25
  • Examples6/15
  • Output15/15
  • Size15/15
  • Resources12/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Good
  • Reaches the network−10

    does this on the machine, and the frontmatter does not say so

    references/sieve-domain-template/build_states.py, line 39
  • Writes files−8

    does this on the machine, and the frontmatter does not say so

    references/sieve-domain-template/build_states.py, line 62
  • Runs shell commands−5

    does this on the machine, and the frontmatter does not say so

    references/sieve-domain-template/build_states.py, line 63

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Good
  • Development activity88
  • Maintenance44
  • Community56
  • Maturity28
  • Momentum40

A hiring post’s comments are a pre-filtered applicant pool, and this skill reads the whole pool. It pulls every comment, resolves each commenter to a username, enriches their profiles in bulk, builds one state per person with a profile image, and scores everyone with a cheap typed classifier called Jev, so a few hundred people are ranked in minutes for cents.

Scoring is blocked until a calibration eval on at least twelve gold cases built from the real pool passes at 80 percent or better. The output is a CSV sorted by slot (outreach top, outreach, relationships, park), depth, then followers, with every proof number labeled as a bio claim.

What You Get

  • A ranked CSV with username, display name, verdict slot, depth score, role level, specialty, followers, verified flag, website, profile URL and bio
  • A summary naming each recommended person with the one or two proof points their bio carries
  • A sieve domain template with config, routing, state builder and a gold-case guide
  • A cost line for each Jev run that states the classifier cost only
  • A coverage report of captured versus displayed comments and any unresolved commenters

Common Use Cases

  • Find the best hire targets among hundreds of apply comments on a Head of Content post
  • Separate working operators from agency founders and peers in a hiring thread
  • Score podcast guest or partnership candidates from a community post by flipping the routing
  • Reuse the pipeline on your own hiring post where commenters are warmer

What Makes It Unique

The classifier cannot run ungated: scoring refuses to start until a passing calibration eval exists for the current question set, and any question change closes the gate again. Bios are treated as claims, unknown stays unknown, rate limits are detected by parsing each response and checking for usable data, and the cost line never reports a missing usage figure as zero.

Under The Hood

It runs commands on your machine and calls outside services. The agent uses an Instagram command-line tool tied to your account to fetch comments and profiles, an Apify actor for bulk enrichment, and a Jev API key for scoring. The bundled Python state builder downloads each profile picture from the web and converts it with ffmpeg, then writes JSON state files, cases and the output CSV into a working folder. Nothing is sent to the people it ranks; outreach drafts need the hiring manager’s approval of the exact text.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers