Skip to content
Conduit

skill

HR-toolkit

Composable Codex plugin for conference papers, domain matching, public person enrichment, and Excel delivery.

SeanX16/HR-toolkitcommunitylistedunreviewed

Set up

Skills are installed by placing their folder where your agent loads skills.

The download comes from github.com/SeanX16/HR-toolkit itself, not from our server, so what you get is exactly what its author published.

Install instructions come from the README

Skills do not use an MCP config. Follow the steps in the README below.

README

synced Oct 23, 2026

HR Toolkit

中文说明

HR Toolkit is a composable, skills-only plugin for Codex. It installs four independent Skills together and lets Codex select one Skill or combine several according to the task.

It supports two related workflows:

  • build a complete, source-backed conference paper and author roster;
  • identify papers related to a research domain, enrich public professional information, and deliver a review-ready Excel workbook.

It can also start from a general list of people, without requiring conference or paper inputs.

Included Skills

StageSkillResponsibilityMain output
Acollect-conference-papersCollect papers and author appearances from declared conference sourcesconference-paper-catalog/v1
Bmatch-papers-to-domainTurn keywords into weighted concepts and match every paper with evidencepaper-domain-matches/v1
Cenrich-person-recordsConfirm a field preset or custom definitions, then enrich public professional informationauthor-enrichment/v1 or person-enrichment/v1
Drender-talent-tablesTurn structured inputs into a standard or natural-language-customized Excel workbook.xlsx

Common routes:

Conference roster:       A
Domain-focused research: A -> B -> C
Conference Excel:        A -> B -> C -> D
General people list:     C
Custom people Excel:     C -> D
Existing JSON to Excel:  D

There is no fifth routing Skill. Each Skill has distinct activation metadata, and Codex decides which one or which combination applies.

B does not weight every supplied phrase mechanically. It can group aliases into one concept, retain conference-wide umbrella terms as non-scoring context, and avoid forcing newer labels onto older papers.

Before C researches any person, it presents Preset A (professional core), Preset B (academic talent full), and Custom. Custom fields require explicit definitions, formats, evidence rules, ambiguity handling, and user confirmation. After a custom run, the user can choose whether to save the field set as a preset.

Install from GitHub

Add this repository as a Codex plugin marketplace, then install the bundled plugin:

codex plugin marketplace add SeanX16/HR-toolkit --ref main
codex plugin add hr-toolkit@hr-toolkit

After installation, start a new Codex task and try one of these prompts:

Collect all papers from this conference and preserve the declared source scope.
Match these papers to neural rendering and show the evidence for every paper.
Enrich this CSV of names with current organization, role, and public profile.
Turn this JSON into a review-ready Excel workbook with the columns I describe.

The repository marketplace entry is stored at .agents/plugins/marketplace.json. The installable package under plugins/hr-toolkit/ is generated from the source and should not be edited manually.

For current plugin and marketplace concepts, see the official Codex plugin packaging documentation.

Inputs and contracts

The conference workflow uses:

conference-paper-catalog/v1
        |
        v
paper-domain-matches/v1
        |
        v
author-research-result/v1
        |
        v
author-enrichment/v1

The general people-list workflow uses:

pasted names / CSV / JSON / profile URLs
        |
        v
person-seed-list/v1
        |
        v
person-field-selection/v1 (user-confirmed)
        |
        v
person-research-result/v1
        |
        v
person-enrichment/v1

All contracts preserve stable IDs, upstream fingerprints, collection time, source URLs, unknown values, and review state.

Safety and intended use

HR Toolkit is a research and delivery workflow, not a hiring decision system.

  • It uses public, person-specific professional evidence.
  • Missing or ambiguous information stays Unknown or Needs Review.
  • It does not infer nationality, ethnicity, age, gender, religion, disability, marital status, political affiliation, or other protected attributes from a name.
  • Domain matching describes paper-topic evidence; it is not a candidate score, hiring recommendation, rejection, or outreach decision.
  • Local runs, researched people data, email lists, and generated workbooks are excluded from Git.

Read PRIVACY.md before using real person data and SECURITY.md before reporting a vulnerability.

Repository layout

.codex-plugin/   Plugin manifest
.agents/         Repository marketplace metadata
skills/          The only editable source for the four Skills
packages/        Shared deterministic Python runtime
schemas/         Versioned cross-Skill JSON schemas
plugins/         Generated installable plugin package committed for distribution
tests/           Offline contract and integration tests
tools/           Validation and packaging utilities
.work/           Local run data, ignored by Git
outputs/         Local Excel deliveries, ignored by Git
distributions/   Optional local build output, ignored by Git

Development

Requirements:

  • Python 3.11 or newer
  • Node.js for the D-stage renderer syntax check
  • A Codex workspace runtime with @oai/artifact-tool when creating Excel files

Run the source validation:

python tools/validate_project.py

Rebuild and verify the repository marketplace package:

python tools/build_distributions.py --output plugins --overwrite
python tools/verify_distributions.py --root plugins

Build an optional standalone local distribution:

python tools/build_distributions.py --overwrite
python tools/verify_distributions.py

See CONTRIBUTING.md for change and pull-request guidance.

License

MIT

↑↓ navigate openesc close