Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Build a talent sourcing strategy for a hard-to-fill role. Use when asked to create a sourcing strategy, a candidate sourcing plan, a channel plan for hiring, or to figure out where to find candidates for a role. Produces a strategy — the ideal-candidate profile and where they are, prioritised sourcing channels, outreach approach, a pipeline target with funnel math, and a weekly plan — so sourcing is deliberate, not just posting and praying.
.claude/skills/mohitagw15856-sourcing-strategy/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -25% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 466% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 87% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 161% | 0% |
| case-22 | ✗→✓ | ▲ Improved | 124% | 0% |
Hard roles aren't filled by posting a job and waiting — they're filled by knowing who you need, where they are, and how to reach enough of them to fill the funnel. This skill builds that plan: the target profile, the channels ranked by where the talent actually concentrates, and the pipeline math so you know how many to source to make one hire.
Given "we can't fill our staff ML engineer role", build the strategy anyway — infer the candidate profile, where they cluster, and a realistic funnel, labelling assumptions. Use funnel ratios with a worked example rather than inventing exact numbers. Never withhold for missing detail.
Ask for these only if they aren't already provided (else infer and label):
1. Ideal candidate profile — the realistic must-haves vs. nice-to-haves, the adjacent profiles worth considering (to widen the pool), and the signals that identify a strong fit.
2. Where they are — where this talent concentrates: companies to source from (and avoid), communities, platforms, events, and content they engage with.
3. Channel plan — sourcing channels ranked by likely yield for this role:
| Channel | Why it fits | Effort | Approach | |---|---|---|---| | Direct sourcing (LinkedIn/GitHub) | … | high | boolean + personalized outreach | | Referrals | … | low | targeted ask to the team | | Communities / events | … | med | … | | Job posts / inbound | … | low | only part of the mix |
4. Outreach approach — the message angle and cadence (pairs with recruiter-outreach and boolean-search-builder).
5. Pipeline target & funnel — how many to source to make the hire: a funnel with ratios + a worked example (e.g. sourced → replied → screened → onsite → offer → hire), so weekly activity is sized to the goal.
6. Weekly plan — the concrete cadence (X sourced, Y outreach, Z screens per week) and how you'll track it.
Talent-sourcing strategy practice — profile-first sourcing, channel prioritisation by talent concentration, funnel/pipeline math, and a measurable weekly cadence.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 53,711 | 34,439 | -36% | 1 | 1 | 0% | 6,905 | 5,653 | -18% | 0 | 0 | — |
case-02 | fail→fail | 37,829 | 29,739 | -21% | 1 | 1 | 0% | 6,162 | 5,391 | -13% | 0 | 0 | — |
case-03 | fail→pass | 51,657 | 27,661 | -46% | 1 | 1 | 0% | 8,326 | 6,230 | -25% | 0 | 0 | — |
case-04 | pass→pass | 20,220 | 28,238 | +40% | 1 | 1 | 0% | 2,341 | 5,566 | +138% | 0 | 0 | — |
case-05 | pass→pass | 48,102 | 40,979 | -15% | 1 | 1 | 0% | 7,286 | 6,387 | -12% | 0 | 0 | — |
case-06 | pass→pass | 28,615 | 32,155 | +12% | 1 | 1 | 0% | 3,425 | 6,242 | +82% | 0 | 0 | — |
case-07 | pass→pass | 21,414 | 28,450 | +33% | 1 | 1 | 0% | 2,732 | 4,843 | +77% | 0 | 0 | — |
case-08 | fail→pass | 9,582 | 17,498 | +83% | 1 | 1 | 0% | 707 | 4,001 | +466% | 0 | 0 | — |
case-09 | fail→fail | 17,870 | 30,939 | +73% | 1 | 1 | 0% | 2,123 | 4,265 | +101% | 0 | 0 | — |
case-10 | fail→fail | 10,866 | 35,960 | +231% | 1 | 1 | 0% | 1,868 | 4,907 | +163% | 0 | 0 | — |
case-11 | pass→pass | 22,201 | 26,563 | +20% | 1 | 1 | 0% | 2,115 | 4,110 | +94% | 0 | 0 | — |
case-12 | pass→pass | 20,401 | 29,629 | +45% | 1 | 1 | 0% | 2,222 | 4,709 | +112% | 0 | 0 | — |
case-13 | fail→fail | 20,496 | 28,130 | +37% | 1 | 1 | 0% | 2,163 | 4,126 | +91% | 0 | 0 | — |
case-14 | pass→pass | 21,525 | 30,026 | +39% | 1 | 1 | 0% | 2,287 | 4,292 | +88% | 0 | 0 | — |
case-15 | fail→fail | 10,081 | 26,487 | +163% | 1 | 1 | 0% | 1,576 | 4,312 | +174% | 0 | 0 | — |
case-16 | fail→fail | 11,194 | 23,815 | +113% | 1 | 1 | 0% | 1,337 | 4,430 | +231% | 0 | 0 | — |
case-17 | fail→pass | 14,594 | 22,716 | +56% | 1 | 1 | 0% | 2,068 | 3,865 | +87% | 0 | 0 | — |
case-18 | fail→pass | 9,474 | 28,145 | +197% | 1 | 1 | 0% | 1,741 | 4,551 | +161% | 0 | 0 | — |
case-19 | pass→pass | 12,732 | 21,349 | +68% | 1 | 1 | 0% | 1,986 | 4,335 | +118% | 0 | 0 | — |
case-20 | pass→pass | 24,532 | 22,732 | -7% | 1 | 1 | 0% | 3,256 | 4,450 | +37% | 0 | 0 | — |
case-21 | fail→fail | 17,215 | 22,134 | +29% | 1 | 1 | 0% | 2,473 | 4,092 | +65% | 0 | 0 | — |
case-22 | fail→pass | 16,750 | 29,484 | +76% | 1 | 1 | 0% | 2,150 | 4,813 | +124% | 0 | 0 | — |
case-23 | pass→pass | 14,503 | 12,219 | -16% | 1 | 1 | 0% | 2,459 | 2,839 | +15% | 0 | 0 | — |
case-24 | fail→pass | 17,983 | 27,996 | +56% | 1 | 1 | 0% | 2,603 | 5,165 | +98% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 24 cases were attempted. The headline lift of +25 percentage points is the difference between those two pass rates over the 24 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.