Una Claude Skill que toma un perfil de puesto más una rúbrica de ICP, construye una query de AI sourcing contra Juicebox, hireEZ, o LinkedIn Recruiter, recupera hasta 200 candidatos, puntúa cada uno contra la rúbrica con evidencia citada, y redacta outreach personalizado para los top-N — y luego se detiene en una compuerta de revisión humana. El recruiter lee el shortlist, edita los mensajes y envía. Reemplaza el loop de 3 horas de Boolean-más-scoring-más-outreach con un loop de revisión de 30 minutos.
Cuándo usarlo
Estás haciendo sourcing para un rol que corre más de una vez por trimestre y la rúbrica de ICP es lo suficientemente estable como para escribirse.
Tienes una rúbrica de ICP con anclas conductuales por dimensión (no solo etiquetas vagas). La plantilla de rúbrica en references/1-icp-rubric-template.md del bundle muestra la forma; si no puedes llenarla, todavía no tienes una rúbrica que esta skill pueda usar para puntuar.
Tienes acceso a la API de Juicebox PeopleGPT, hireEZ, o LinkedIn Recruiter. La skill se rehúsa a hacer fallback a scraping de URLs públicas de LinkedIn.
Un recruiter o sourcer humano revisa cada shortlist antes de que se envíe cualquier outreach. La skill escribe los borradores a disco y se detiene.
Cuándo NO usarlo
Auto-rechazo en el loop. La skill rankea; no rechaza. Los candidatos “skipeados” se surface con razones para que el recruiter los pueda anular. Conectar una acción reject a un umbral de score convierte esto en toma de decisión automatizada y dispara obligaciones de EU AI Act Anexo III de alto riesgo más obligaciones de bias-audit de NYC LL 144 dentro de un año antes de usarlo. Si necesitas eso, consigue un bias audit, no esta skill.
Scoring sobre proxies de clases protegidas. Prestigio de escuela como dimensión independiente, origen del nombre, presencia de foto, penalización por gaps de empleo, edad inferida del año de graduación, “culture fit” sin anclas conductuales. El checklist de equidad de la skill se rehúsa a correr si cualquiera de estos aparece en la rúbrica. No edites el checklist para hacer pasar una rúbrica sesgada.
Recomendaciones de pay-band. NYC LL 32-A, Colorado, California y Washington requieren rangos publicados y obligaciones de bias-audit sobre decisiones de pago automatizadas. Usa una herramienta de comp benchmarking, no una skill de sourcing.
Búsquedas one-off de C-suite. Una búsqueda retenida para un individuo específico nombrado o para un ejecutivo definido de forma estricta se hace más rápido por un humano con red de contactos. La skill está construida para sourcing repetible a nivel IC y manager, donde la calibración de la rúbrica recupera su costo de setup.
Reference checks o investigación por backchannel. Postura de consentimiento distinta. Workflow distinto.
Setup
Pega el bundle. Coloca apps/web/public/artifacts/candidate-sourcing-claude-skill/SKILL.md en tu directorio de skills de Claude Code (o en las Skills personalizadas de claude.ai).
Llena la rúbrica. Copia references/1-icp-rubric-template.md a un archivo por rol bajo tu propio repo. Reemplaza cada {placeholder}. La skill captura el SHA-256 de la rúbrica en su audit log por run, así que las ediciones posteriores son visibles en el retro.
Configura el canal de fuente. Agrega tu API key de Juicebox o hireEZ a la configuración de la skill. Para LinkedIn, configura las credenciales de la API de Recruiter — la skill se rehúsa a scrapear URLs de perfiles públicos.
Escribe las listas de do-not-poach y exclude. Un CSV de dominios de clientes (do-not-poach) y un CSV de URLs de exclude_list (recientemente rechazados, en periodo de silencio, opt-out). El pre-filtro determinístico en el paso 3 de la skill aplica estas antes de que el LLM vea ningún candidato.
Dry-run sobre un rol cerrado. Córrelo sobre un rol que sourceaste manualmente el trimestre pasado. Compara los top-25 de la skill con tus top-25 manuales. Afina las anclas de la rúbrica si la skill calibra distinto — las anclas, no la query de búsqueda, son las que suelen estar mal.
Lo que la skill realmente hace
Seis pasos, en orden. El orden importa: los filtros determinísticos y el pre-flight de equidad vienen antes del ranking del LLM, porque dejar suelto a un LLM sobre un pool contaminado produce output rápido, confiado e inutilizable.
Validar la rúbrica contra references/2-fairness-checklist.md. Detener si la rúbrica contiene proxies de clases protegidas. La elección de fallar antes del retrieval en lugar de después es deliberada — una rúbrica sesgada cargada en la API de una herramienta de sourcing deja una entrada de log que ya cuenta como procesamiento automatizado bajo el Art. 22 del GDPR.
Construir la query de búsqueda en el formato nativo del canal. Tope de sinónimos en 5 por dimensión; tope del pool recuperado en 200. Pools más grandes degradan el ranking porque el contexto del modelo se llena con candidatos de baja relevancia.
Pre-filtro determinístico. Descartar matches de exclude_list, empresas do-not-poach, mismatches de ubicación, y perfiles con >18 meses de antigüedad. Estos son filtros auditables; el LLM no los re-litiga.
Ranking basado en rúbrica. Score 1-5 sobre skill, level, company-pattern, response-likelihood. Cada score arriba de 1 cita un string textual del perfil. Sin cita → score 1. El requisito de la cita es lo que mantiene al modelo anclado en el texto del perfil en vez de inferir desde nombre, foto o escuela.
Compuerta de revisión humana. Escribir shortlist.md y archivos outreach/<id>.md por candidato. Detener. La skill no define ninguna acción send.
Audit log. Anexar una línea JSONL por run con run_id, rubric_sha256, tamaños de pool, canal, modelo. Sin PII. Esto es lo que hace defendible al run bajo cuestionamiento de NYC LL 144 o EU AI Act.
El formato del shortlist y el layout de evidencia por candidato viven en references/3-shortlist-format.md en el bundle. El formato es fijo porque los consumidores downstream — recruiter, hiring manager, revisor de auditoría — necesitan columnas predecibles.
Realidad de costos
Por shortlist de 25 desde un pool de 200 candidatos, sobre Claude Sonnet 4.5:
Costo de retrieval — depende del canal. Juicebox PeopleGPT cuenta contra tu cuota mensual de queries (los planes starter de 200 búsquedas se topan rápido si corres múltiples roles por semana). Los unlocks-por-mes de hireEZ son la restricción vinculante ahí. La API de LinkedIn Recruiter tiene sus propias cuotas de InMail y búsqueda por seat. Nada de esto cambia con la skill en el loop; gastas la misma cuota de canal que habrías gastado haciendo Boolean manual.
Tokens del LLM — típicamente 80-120k tokens de input (rúbrica + 200 extractos de perfil de candidato + instrucciones de la skill) y 8-15k tokens de output (shortlist + 25 borradores de outreach). Sobre Sonnet 4.5 eso es aproximadamente $0,50-0,80 por shortlist. El mes completo para un sourcer corriendo ~80 shortlists cae en $40-65 en costo de modelo.
Tiempo del recruiter — la ganancia está aquí, no en el costo del modelo. Boolean manual + scoring + outreach para 25 candidatos toma 2-3 horas. Revisar el shortlist de la skill y editar los borradores toma 25-40 minutos, que es lo que hace que valga la pena correr el workflow.
Tiempo de setup — 45 minutos para la rúbrica y las listas de exclude si la rúbrica ya existe en alguna forma; más si la rúbrica es completamente nueva (en cuyo caso structured interviewing es el prerequisito, no esta skill).
Métrica de éxito
Trackea tres números por rol por mes, en el ATS:
Reply rate al outreach — debe igualar o superar la tasa baseline manual del recruiter. Si baja, los borradores de outreach son genéricos — usualmente la rúbrica es demasiado gruesa, no el modelo.
Tasa de pase de shortlist a screen — la proporción de candidatos del shortlist con los que el hiring manager está de acuerdo en que valen un screen. Debe ser ≥70% sobre un rol estable. Por debajo de eso, la rúbrica de ICP está mal calibrada; vuelve a correr sobre un rol cerrado y afina.
Tiempo desde apertura del rol hasta el primer screen calificado — la métrica de throughput que la skill busca mover. La reducción de 3-horas-a-30-minutos aparece aquí, no en el gasto de modelo.
vs alternativas
vs Gem AI Sourcing — Gem es dueño del workflow del recruiter de punta a punta (UI de sourcing, secuencias, analytics, integración de ATS vía Ashby y otros). Elige Gem si quieres un producto gestionado y tu equipo va a vivir en su UI. Elige esta skill si quieres la rúbrica, la lógica de pre-filtro, y el audit log en tu propio repo, bajo control de versiones, con el modelo intercambiable.
vs el ranking AI nativo de hireEZ — el AI Match de hireEZ es buen retrieval; el gap está en la capa de la rúbrica. Con esta skill mantienes hireEZ como el canal de retrieval y traes tu propia rúbrica + scoring con evidencia citada por encima. Si los defaults de hireEZ coinciden con tu ICP, no necesitas esta skill.
vs Boolean manual + scoring por spreadsheet — manual es lo correcto para búsquedas one-off o ejecutivas, donde la rúbrica está en la cabeza del recruiter y escribirla es overhead que no se paga. La skill se gana su costo de setup en roles que se repiten.
vs script DIY en Python contra las APIs de LinkedIn / Juicebox — misma calidad de ranking si construyes el prompt con cuidado, pero también construyes el checklist de equidad, el audit log, y la compuerta de revisión humana tú mismo. El bundle los trae.
Cuidados
Amplificación de sesgo — guardado por el checklist de equidad en references/2-fairness-checklist.md, que detiene el run si la rúbrica contiene proxies de clases protegidas. El audit log captura rubric_sha256 por run para que la rúbrica usada en una fecha dada sea reproducible bajo revisión de EU AI Act o NYC LL 144.
Datos obsoletos de LinkedIn / Juicebox — guardado por el filtro determinístico en el paso 3 (descartar perfiles con >18 meses de antigüedad) y por la dimensión de response-likelihood en el scoring (que pondera la frescura). Los candidatos en cold-storage no desplazan a los que están buscando activamente.
Exposición a los ToS de LinkedIn — guardado por la negativa a scrapear URLs de perfiles públicos. La skill usa la API de Recruiter, Juicebox o hireEZ, que traen su propio licenciamiento de datos. Si se selecciona linkedin_recruiter y la API no está configurada, la skill aborta con un setup-error en vez de hacer fallback.
Drift de auto-send — guardado por la compuerta de revisión humana (paso 5) y por la ausencia de cualquier acción send en la skill. Los borradores se escriben en archivos outreach/<id>.md para que el recruiter los pegue en el outbox del ATS / herramienta de sourcing. Los mensajes redactados por AI y enviados sin revisión producen volumen sin calidad y dañan la candidate experience.
Transparencia de comp — los borradores de outreach nunca citan un número; referencian la banda como “rango competitivo divulgado en el screen” para que el recruiter siga siendo la fuente de las afirmaciones de pay-band (requisitos de pay-transparency de NYC LL 32-A, Colorado, California, Washington).
Stack
El bundle de la skill vive en apps/web/public/artifacts/candidate-sourcing-claude-skill/ y contiene:
SKILL.md — la definición de la skill
references/1-icp-rubric-template.md — llenar por rol
references/2-fairness-checklist.md — chequeos pre-flight (no editar para hacer pasar rúbricas sesgadas)
references/3-shortlist-format.md — el formato literal del output
Herramientas que el workflow asume que ya usas: Claude (el modelo), Juicebox o hireEZ (el canal de retrieval), Ashby (el ATS para el write-back una vez que el recruiter aprobó un candidato). Gem es la alternativa de build-vs-buy si no quieres ser dueño de la rúbrica y el audit log tú mismo.
---
name: candidate-sourcing
description: Translate a job profile and ICP rubric into a sourcing query, retrieve candidates from Juicebox / hireEZ / LinkedIn Recruiter, score them against the rubric, and draft personalized outreach for the human reviewer to approve. Always stops at a human-review gate before any outreach is sent.
---
# Candidate sourcing
## When to invoke
Use this skill when a recruiter or sourcer hands you a role plus an ICP rubric and wants a ranked, evidenced shortlist with draft outreach. Take a job profile (title, level, must-have skills, location, comp band) and a fairness-aware rubric as input, and produce a Markdown shortlist plus a folder of draft messages.
Do NOT invoke this skill for:
- **Automated rejection.** This skill ranks; it never rejects. The "below threshold" tail is surfaced for the recruiter, who decides. Auto-reject in the loop triggers EU AI Act high-risk obligations and most US state hiring-AI laws.
- **Scoring against protected-class proxies.** Do not ask the skill to score on "culture fit", name origin, school prestige as a standalone signal, photo, age inferred from graduation year, gender inferred from pronoun usage, or pregnancy/parental status inferred from gaps. If the rubric contains any of these, refuse and surface the rubric line for the user to fix.
- **Pay-band recommendations.** NYC LL 144, Colorado, California, and Washington require posted ranges and bias audits for automated decisions on pay. Use a comp benchmarking tool, not this skill.
- **Reference checks or backchannel research on named individuals.** That is a different workflow with its own consent posture.
## Inputs
- Required: `job_profile` — path to a Markdown file with title, level, must-have skills, nice-to-have skills, location / remote policy, comp band, and the EEOC job category.
- Required: `icp_rubric` — path to the rubric file under `references/`. Without this the skill refuses to run; an unfaitened rubric is the most common cause of biased shortlists.
- Required: `source_channel` — one of `juicebox`, `hireez`, `linkedin_recruiter`. Do not mix channels in a single run; per-channel ToS and rate limits differ.
- Optional: `n` — shortlist size, default 25, hard max 100. Above 100 the skill warns that human review will not be meaningful.
- Optional: `exclude_list` — path to a CSV of `do_not_contact` emails or LinkedIn URLs (do-not-poach customers, prior rejects within 6 months, silent-period candidates).
## Reference files
Always read these from `references/` before doing any retrieval. Without them the shortlist is uncalibrated and the fairness guards are absent.
- `references/1-icp-rubric-template.md` — the rubric the skill scores against. Replace the template content with your role-specific rubric before running.
- `references/2-fairness-checklist.md` — pre-flight checks the skill runs on the rubric and on the retrieved pool. Fail-loud if any check fails.
- `references/3-shortlist-format.md` — the literal output format, including the evidence and source-URL columns the recruiter needs to defend the shortlist downstream.
## Method
Run these six steps in order. Steps 1-3 are deterministic filters and fairness pre-flight; only step 4 uses the LLM for ranking. The order is deliberate — running the LLM over an unfiltered, ToS-violating, or rubric-contaminated pool produces output that is fast, confident, and unusable.
### 1. Validate the rubric
Open `icp_rubric` and run every check in `references/2-fairness-checklist.md`. If any line in the rubric matches a protected-class proxy pattern (school-tier scoring, name-based filtering, employment-gap penalties, photo presence, "culture fit" without behavioral anchors), stop and return the offending lines to the user. Do not proceed with retrieval.
The choice to fail before retrieval rather than after is intentional: a biased rubric loaded into a sourcing tool's API leaves a log entry that counts as automated processing under GDPR Art. 22 and the EU AI Act, regardless of whether the skill ever shows the user the result.
### 2. Build the search query
Translate the job-profile must-haves into the channel's native query format:
- `juicebox` → natural-language PeopleGPT prompt, with location and level filters set as structured parameters not free text.
- `hireez` → Boolean string with explicit AND/OR/NOT grouping. Cap synonyms at 5 per dimension; longer Boolean degrades hireEZ's relevance ranking.
- `linkedin_recruiter` → use the Recruiter API with structured filters only. **Do not scrape `linkedin.com/in/` URLs** — that violates LinkedIn ToS and the *hiQ v. LinkedIn* settlement does not change ToS exposure for production sourcing.
Cap the retrieved pool at 200. Larger pools degrade rubric scoring because the LLM context fills with low-relevance candidates and the ranking flattens.
### 3. Deterministic pre-filter
Before the LLM sees any candidate, apply hard filters:
- Drop anyone in `exclude_list`.
- Drop anyone whose current company is on the do-not-poach list.
- Drop anyone whose profile was last updated more than 18 months ago (LinkedIn / Juicebox staleness signal).
- Keep only candidates whose stated location matches the role's location policy (with a configurable radius for hybrid roles).
These filters are deterministic so they can be audited. The LLM does not re-litigate them in step 4.
### 4. Rubric-based ranking
For each remaining candidate, score 1-5 on each rubric dimension (skill-match, level-fit, company-pattern-fit, response-likelihood). For every score above 1, cite the specific evidence string from the candidate's profile. No evidence string → score 1 by default.
Why a citation requirement: it forces the model to ground each score in profile text rather than infer from a name, photo, or school. Scores without evidence are the mechanism by which bias enters AI-augmented sourcing pipelines.
### 5. Human-review gate
Stop. Write the shortlist to `shortlist.md` per the format in `references/3-shortlist-format.md`. Write the draft outreach to `outreach/<candidate-id>.md`, one file per candidate. Do not call any "send" endpoint. Do not mark candidates as contacted in the ATS. Surface the path to both directories and exit.
The recruiter's job from here: read the shortlist, edit the messages, and send through the ATS or sourcing tool's outbox. The skill does not re-enter the loop until the next role.
### 6. Audit log
Append a single line to `audit/<YYYY-MM>.jsonl` containing: `run_id`, `role`, `rubric_sha256`, `pool_size_pre_filter`, `pool_size_post_filter`, `shortlist_size`, `channel`, `model_id`, `timestamp`. Do not log candidate PII to this file. The audit log exists so that under NYC LL 144 or EU AI Act questioning, the recruiter can demonstrate which rubric was used on which date.
## Output format
```markdown
# Sourcing shortlist — {Role title}
Generated: {ISO timestamp} · Channel: {channel} · Pool: {pre} → {post} · Rubric SHA: {short}
| # | Name | Current role | Current company | Skill | Level | Pattern | Response | Aggregate | Source |
|---|---|---|---|---|---|---|---|---|---|
| 1 | Jamie L. | Senior Backend Engineer | Acme Fintech | 5 | 5 | 4 | 4 | 18 | {URL} |
| 2 | ... | ... | ... | ... | ... | ... | ... | ... | ... |
## Evidence — top 5
### 1. Jamie L. (aggregate 18)
- **Skill (5)**: "5y Go, 2y Rust, led migration from monolith to event-driven services" — profile, role 2.
- **Level (5)**: "Senior IC, scope across two teams, mentors three engineers" — profile, current role.
- **Pattern (4)**: "Stripe → Plaid → Acme Fintech" — three fintech roles in sequence.
- **Response likelihood (4)**: profile updated 11 days ago, "open to opportunities" tag set.
### 2. ...
## Skipped — surfaced for review (not auto-rejected)
| Name | Reason |
|---|---|
| ... | "current company on do-not-poach list (Acme Customer)" |
| ... | "profile last updated 2023-11, staleness > 18mo" |
## Draft outreach
Drafts written to `outreach/`. Recruiter reviews and sends; this skill
does not contact candidates.
- `outreach/jamie-l.md`
- `outreach/...`
```
## Watch-outs
- **Bias amplification (NYC LL 144, EU AI Act, EEOC).** *Guard:* the fairness checklist in `references/2-fairness-checklist.md` runs in step 1 and refuses retrieval if rubric contains protected-class proxies. Audit log in step 6 stores `rubric_sha256` so the rubric used on a given run is reproducible.
- **LinkedIn ToS exposure.** *Guard:* skill uses the Recruiter API (or Juicebox / hireEZ which carry their own data licensing), never scrapes public LinkedIn pages. If the channel is `linkedin_recruiter` and the Recruiter API is not configured, the skill aborts with a setup-error rather than falling back to scraping.
- **Stale profile data.** *Guard:* deterministic filter in step 3 drops candidates with `profile_updated > 18mo`. Response-likelihood scoring in step 4 weights profile freshness explicitly so cold-storage candidates do not crowd out actively looking ones.
- **Auto-send drift.** *Guard:* skill stops at the human-review gate in step 5 and writes to `outreach/` files. There is no `send` action defined anywhere in this skill. To send, the recruiter pastes into the ATS / sourcing tool outbox.
- **Rubric drift mid-search.** *Guard:* `rubric_sha256` is captured per run; if the rubric changes between two runs for the same role, the audit log shows both hashes, making it visible in retro.
- **Compensation discussion in draft outreach.** *Guard:* outreach templates in this skill never quote a number; they reference the comp band as "competitive range disclosed on screen" so the recruiter remains the source of pay-band statements (NYC LL 32-A, CO, CA, WA pay-transparency posting).
# ICP rubric — TEMPLATE (per role)
> Replace this template's contents with the rubric for the specific role.
> The candidate-sourcing skill scores against the four dimensions below.
> Each dimension MUST have behavioral anchors — vague labels ("senior")
> without anchors produce noisy and biased scoring.
## Role identity
- **Title**: {e.g. Senior Backend Engineer, Platform}
- **Level**: {IC4 / IC5 / EM1 — your internal scale}
- **Location policy**: {remote-US / hybrid-NYC-2dpw / onsite-Berlin}
- **EEOC job category**: {2 — Professionals (most engineers); see EEO-1}
- **Comp band (recruiter-internal, never sent to skill output)**: {range}
## Dimension 1 — Skill match (1-5)
The candidate's profile shows direct experience with the must-have technologies and the specific problem-shape of the role.
| Score | Anchor |
|---|---|
| 5 | Held a role doing exactly this work for ≥2 years; cites artifacts (talks, OSS, posts). |
| 4 | Held a role doing exactly this work for ≥1 year; no artifacts. |
| 3 | Adjacent work (e.g. Java backend role for a Go role); transferable. |
| 2 | Tangential work; would require ramp. |
| 1 | No evidence in profile. |
## Dimension 2 — Level fit (1-5)
The candidate's stated scope and tenure pattern match the level the role is hiring at. Do NOT use school prestige, employer prestige, or title inflation as a level signal — anchor on scope description.
| Score | Anchor |
|---|---|
| 5 | Profile shows scope at or above target level (multi-team, mentoring, technical strategy). |
| 4 | Scope at target level for ≥1 year. |
| 3 | One level below target; growth trajectory plausible. |
| 2 | Two levels below; reach. |
| 1 | More than two levels off, in either direction. |
## Dimension 3 — Company-pattern fit (1-5)
The shape of the candidate's prior employers matches the shape of yours (stage, scale, regulated/unregulated, B2B/B2C). Anchor on *characteristics*, not brand names — brand-name scoring is the most common bias vector in AI-augmented sourcing.
| Score | Anchor |
|---|---|
| 5 | ≥2 prior employers match {stage/scale/domain pattern}. |
| 4 | 1 prior employer matches; others adjacent. |
| 3 | All adjacent (different domain, similar stage). |
| 2 | Mostly mismatched; one transferable role. |
| 1 | No pattern match. |
## Dimension 4 — Response likelihood (1-5)
How likely the candidate is to respond to outreach right now.
| Score | Anchor |
|---|---|
| 5 | Profile updated <30 days; "open to opportunities" set; recently posted about job search. |
| 4 | Profile updated <90 days. |
| 3 | Profile updated <180 days. |
| 2 | Profile updated <12 months. |
| 1 | Stale profile (>12 months) — *also flagged in pre-filter for drop at >18mo*. |
## Disqualifiers (deterministic, applied in step 3 of the skill)
These cause the candidate to be surfaced in the "skipped" table, not auto-rejected. The recruiter decides.
- Current company is on do-not-poach list (`{path-to-list}`).
- Email or LinkedIn URL appears in `exclude_list`.
- Stated location does not match role's location policy + radius.
- Profile last updated >18 months ago.
## Bias guards (refusal triggers — skill aborts in step 1 if present)
If any of the following appear in this rubric, the skill refuses to run:
- School-tier scoring as a standalone dimension.
- Name-based filtering or scoring.
- Photo-based scoring.
- Employment-gap penalties without a job-related justification.
- Age inferred from graduation year used in any dimension.
- Gender, ethnicity, religion, sexual orientation, parental status, or disability status as a scored or filtered dimension.
- "Culture fit" without behavioral anchors.
## Last edited
{YYYY-MM-DD} — bump on every material change. The skill captures the SHA-256 of this file in its audit log per run.
# Fairness pre-flight checklist
> The candidate-sourcing skill runs every check below in step 1 (rubric
> validation) and step 3 (post-filter pool review). Any failed check
> halts the run with a message naming the failure. Do not edit this file
> to make checks pass — fix the rubric or the search instead.
## A. Rubric checks (run before retrieval)
A1. **No protected-class proxies.** Scan the rubric for any of the following terms or patterns. Any hit halts the run:
- `school`, `university`, `Ivy`, `tier-1`, `top-N` (when used as a scoring dimension, not as one signal among many)
- `name origin`, `surname`, `first name`
- `photo`, `headshot`, `appearance`
- `age`, `years since graduation`, `birth year`
- `gender`, `pronoun`, `she/her`, `he/him` (as filter terms)
- `ethnicity`, `race`, `nationality` (except where required for immigration-status filtering with documented legal basis)
- `pregnant`, `parental`, `maternity`, `paternity`
- `disability`, `accommodation`
- `religion`, `political`, `marital`
- `culture fit` without a behavioral-anchor table immediately following
A2. **Anchors present on every dimension.** Each rubric dimension must have a 1-5 anchor table. Anchors prevent the LLM from scoring on vibes. Halt if any dimension has free-text anchors only.
A3. **Disqualifier list is short and mechanical.** Disqualifiers must be deterministic facts (do-not-poach list, location mismatch, staleness). Halt if a disqualifier requires judgment (e.g. "not a culture fit", "seems junior").
A4. **Comp band is recruiter-internal.** The skill's output must not quote a comp number to the candidate. Outreach templates reference the band as "competitive range disclosed on screen". Halt if the rubric includes a "send comp in outreach" instruction.
## B. Pool checks (run after deterministic pre-filter, before LLM ranking)
B1. **Pool size sanity.** If post-filter pool < 10, the skill warns the recruiter that scoring on a tiny pool is meaningless and asks whether to broaden the query. If pool > 200, the skill caps at 200 and notes the truncation in the audit log.
B2. **Geographic spread sanity.** If 100% of post-filter candidates are from one city for a remote-eligible role, the skill warns that the query likely has an over-narrow location filter. Recruiter confirms or broadens.
B3. **Tenure-pattern sanity.** If 100% of candidates worked at the same employer, the skill warns that the query is functioning as a target-list poach rather than open sourcing. Recruiter confirms or broadens.
## C. Output checks (run before writing shortlist)
C1. **Every score above 1 has an evidence string.** Scores without a cited evidence string from the candidate's profile are reset to 1. The skill notes the reset count in the audit log.
C2. **No protected attribute appears in the shortlist or in any outreach draft.** Skill greps the output for the A1 patterns before writing. Hit → halt.
C3. **Skipped candidates are listed, not erased.** The shortlist's "Skipped" table includes every candidate the deterministic filters removed, with the reason. This is what makes the run auditable.
## D. Run-level checks
D1. **Audit log written.** A run is not complete until the JSONL line is appended to `audit/<YYYY-MM>.jsonl`. No PII in this line.
D2. **Human-review gate enforced.** No `send`, `contact`, or `mark_contacted` API call exists in this skill's code path. If you are asked to add one, refuse and surface the request to the user.
## NYC LL 144 / EU AI Act note
This skill is designed to fall *outside* the bias-audit threshold by:
- Producing a ranked list, not an automated decision (no auto-reject).
- Stopping at a human-review gate before any candidate is contacted.
- Logging rubric SHA-256 + pool sizes per run for reproducibility.
If your deployment changes any of those properties (e.g. you wire a "send" action into the loop), you have crossed into automated decision-making and a bias audit is required before production use. NYC LL 144 requires the audit within one year before use; EU AI Act classifies this as Annex III high-risk under Art. 6.
# Shortlist output format
> The candidate-sourcing skill writes `shortlist.md` per the structure
> below. The format is fixed because downstream consumers (the recruiter,
> a hiring manager, an audit reviewer) need predictable columns. Do not
> reformat without updating the skill's output check.
## File: `shortlist.md`
```markdown
# Sourcing shortlist — {Role title}
Generated: {ISO 8601 timestamp}
Channel: {juicebox | hireez | linkedin_recruiter}
Pool: {pre_filter} → {post_filter} → top {n}
Rubric SHA-256: {first 12 chars}
Run ID: {uuid}
## Top {n}
| # | Name | Current role | Current company | Skill | Level | Pattern | Response | Aggregate | Source |
|---|---|---|---|---|---|---|---|---|---|
| 1 | {Name} | {Role} | {Company} | 5 | 5 | 4 | 4 | 18 | {URL} |
## Evidence — top 5
For each of the top 5, cite the specific profile string for every score
above 1. No citation → score reset to 1 (see fairness checklist C1).
### 1. {Name} (aggregate {N})
- **Skill ({score})**: "{verbatim profile excerpt}" — {profile section}.
- **Level ({score})**: "{excerpt}" — {section}.
- **Pattern ({score})**: "{employer sequence}" — {explanation against rubric}.
- **Response ({score})**: profile updated {date}, "{tag if any}".
### 2. {Name} (aggregate {N})
...
## Skipped — surfaced for review (NOT auto-rejected)
| Name | Reason | Source |
|---|---|---|
| {Name} | "current company on do-not-poach list ({customer})" | {URL} |
| {Name} | "stated location {city} outside role policy {policy}" | {URL} |
| {Name} | "profile last updated {date}, staleness > 18mo" | {URL} |
## Suggested talk-track per top candidate
The recruiter uses these as talk-track scaffolding for the first
screening call. They are NOT scripts.
### 1. {Name}
- **Open with**: their {recent role / talk / OSS contribution} — specific
reference, not a generic compliment.
- **Likely motivation hypothesis**: {evidence-based, e.g. "third fintech
role in a row, may be looking for a non-fintech reset; ask"}.
- **Hesitation to surface**: {e.g. "current company is well-funded; ask
what would have to be true for them to consider a move"}.
### 2. {Name}
...
## Outreach drafts
Drafts written to `outreach/{candidate-id}.md`, one file per candidate.
The recruiter reviews, edits, and sends through the ATS or sourcing
tool's outbox. The skill does not contact candidates.
- `outreach/{id-1}.md`
- `outreach/{id-2}.md`
- ...
```
## File: `outreach/<candidate-id>.md`
```markdown
# Outreach draft — {Name}
Channel: {LinkedIn InMail | email | Juicebox sequence}
Subject: {≤60 chars, references a specific signal from the profile}
---
Hi {first name},
{One sentence referencing a specific, recent thing from their profile —
the {recent role / talk / project / post}. Not a flattery line.}
I'm hiring a {role title} at {company}. The reason I reached out is
{specific connection between their background and the role — cite the
profile signal}. The role's {one specific differentiator that would
matter to someone with this background}.
If you're open to a 15-minute conversation, I'm happy to share more. The
comp range will be disclosed on screen if we get to that step.
{Recruiter name}
---
## Recruiter-only metadata (strip before sending)
- Aggregate score: {N}
- Top evidence string: "{excerpt}"
- Source URL: {URL}
- Run ID: {uuid}
- Reviewed by recruiter: [ ]
- Sent: [ ]
```
## Why these fields are non-negotiable
- **`Source` URL on every row** — required for the recruiter to spot-check the LLM's evidence claims against the actual profile.
- **`Pool: pre → post → top N`** — surfaces how many candidates were filtered out deterministically vs. by the LLM. Big LLM-side cuts on a small post-filter pool is a signal of overfitting to rubric noise.
- **`Rubric SHA-256`** — proves which rubric was used on this run (NYC LL 144 audit defense + EU AI Act traceability).
- **`Skipped` table** — candidates filtered out are listed with reasons, not erased. Erasing them turns the workflow into automated rejection.
- **Recruiter-only metadata in outreach** — stripped before sending; its presence in the draft is what reminds the recruiter the message is a draft, not a finished product.
