Agent skill
signal-sourcing-engine
Run a systematic sourcing engine: network loops, signal harvesting, and thesis-driven outbound. Use when you need to generate proprietary dealflow, keep a pipeline current, or expand coverage.
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/data/signal-sourcing-engine
Metadata
Additional technical details for this skill
- author
- evalops
- version
- 0.2
SKILL.md
Signal sourcing engine
When to use
Use this skill when you need to:
- Generate new at-bats (companies and founders) in a specific thesis area
- Build a repeatable sourcing system (not ad hoc coffee chats)
- Turn weak signals into qualified founder meetings
Inputs you should request (only if missing)
- Thesis wedge (or ask for a short description)
- Stage focus + geography constraints (if any)
- How the firm defines a "qualified" first meeting (traction, team, ICP, etc.)
- Existing CRM rules (Salesforce fields/stages) if available
Outputs you must produce
- Sourcing loops (3 compounding systems, not ad hoc searches)
- Signal rubric (what to watch + how to score)
- Weekly pipeline update (10 new names, 3 meetings, 1 partner-ready candidate)
- Monthly network refresh (top 50 humans + 10 new additions)
- Outreach experiments (2 variants with response tracking)
Hard gate: No "qualified" company exists unless there's a Salesforce Account/Lead + next task + owner.
Templates:
- assets/signal-rubric.md
- assets/outreach-templates.md
- assets/weekly-pipeline-update.md
Cadence and metrics (required)
Weekly deliverables
- 10 new names added to pipeline
- 3 meetings scheduled or completed
- 1 partner-ready candidate surfaced
- Salesforce updated with all entries
Monthly deliverables
- Refresh top 50 humans list (remove stale, add emerging)
- Add 10 new humans to network cultivation list
- Review outreach experiments: what's working, what's not
- Update signal rubric based on hit rate
Procedure
Loop A: Earned network loop (highest signal, first-class object)
Goal: become "first call" for a cluster of builders and operators.
Build the system:
- Identify 2-3 "talent pools" for the wedge:
- ex-employee clusters from relevant companies
- OSS communities (maintainers + power users)
- buyer/operator communities (CISOs, data eng leads, RevOps, etc.)
- Build a "Top 50 humans" list with columns:
- Name, role, company, thesis relevance
- Last touch date, relationship strength (1-5)
- Value provided to them, value received
- Next action + due date
- Provide small, concrete value before asking for meetings:
- 1 targeted intro
- a short market map excerpt
- a recruiting assist on a specific role
- Track compounding metrics:
- Touches -> warm conversations -> referrals -> founder intros
- Target: 20% of Top 50 should generate a referral per quarter
Monthly refresh:
- Review all 50: who's gone stale? Who's risen?
- Add 10 new humans, remove 10 lowest-value
- Log changes in Salesforce (Contacts with "Network" tag)
Loop B: Signal harvesting loop (volume, disciplined)
Goal: produce a weekly list of candidates worth human qualification.
Build the system:
- Define 8-12 signals tied to the wedge (avoid generic hype).
Examples:
- hiring for "founding AE" or "head of sales" (GTM transition)
- design partners mentioned in job posts
- OSS adoption with maintainer activity and downstream usage
- specific buyer pain appearing in forums where buyers complain
- Score each signal (1-5) for:
- Relevance to thesis
- Timeliness
- Uniqueness (are competitors seeing this too?)
- Human qualification pass:
- 10 minutes scan -> 50 names -> 5 worth work -> 1 worth meeting.
- Log "why" for every pass to improve the rubric.
- Weekly output: 10 new names with signal scores
Loop C: Thesis-driven outbound loop (precision outreach)
Goal: outreach that feels like "I did the work," not spam.
Build the system:
- Create 2 outreach variants per thesis:
- Variant A: insight-led ("We mapped this space and noticed...")
- Variant B: value-led ("We can intro you to [specific customer]...")
- Track response rates per variant (aim for >20% response, >10% meeting)
- Rules for every message:
- Must include: the wedge you believe is emerging
- Must include: what looks distinct about their approach
- Must include: a concrete offer (customer intro, operator feedback, recruiting, etc.)
- If it could be sent to 20 companies with minimal edits, rewrite it.
Monthly review:
- Which variant is winning?
- What offers get responses? What offers fall flat?
- Update templates based on data.
Qualification checklist (first pass)
A candidate is "qualified for a first meeting" if you can answer:
- Who buys? Who uses?
- What changes the buyer's mind?
- What do they replace?
- What proof of pull exists (even weak)?
- What's the likely wedge expansion path?
- Must be true: one sentence stating what makes this worth partner time
- Fastest falsification test: what's the 1-call test?
Salesforce logging (MANDATORY GATE)
Hard rule: A company does not exist in your pipeline unless it has:
- A Lead (founder contact) or Account (company)
- A next-step Task with owner and due date
- Source field populated (Loop A/B/C)
Minimum fields to capture:
- Source (Loop A/B/C)
- Thesis tag / segment
- Signal score (1-5)
- Must be true (one sentence)
- Next step + due date
- Status (meet / watch / pass)
- Pass reason (if passing) + "what would change our mind"
Use salesforce-crm-ops for API logging patterns.
Examples
- Input: "Source seed-stage AI security posture management."
- Output: Top 50 humans list with refresh schedule, 10 new names this week (all in Salesforce), 5 outreach targets with 2 variants, 2 warm intros, weekly metrics (3 meetings, 1 partner-ready), outreach experiment results.
Edge cases
- If signals are noisy: tighten wedge and require buyer clarity before meetings.
- If outbound response is low: your offer is not concrete enough; fix the offer, not the subject line.
- If you can't log to Salesforce: stop and fix the integration before continuing. No exceptions.
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
Didn't find tool you were looking for?