SkillAgentSearch skills...

relationship-intelligence

Consolidate lead feeds and relationship history into evidence-backed client, recruiting, partnership, sponsorship and acquisition opportunities. Use for cross-source identity reconciliation, prior-touch attribution, independent fit tiers and owner-ready action queues.

Install / Use

npx skills add ericosiu/ai-marketing-skills --skill relationship-intelligence

Installs into whichever agent you are using.

About this skill
📄

SKILL.md

Installable skill definition

Quality Score

85/100

Category

Automation

Supported Platforms

Universal

Tags

Our assessment of relationship-intelligence

relationship-intelligence scores 85/100 on our quality scale, 1924th of 2,790 Automation skills we index.

Its SKILL.md is 5.6 KB long, split into 6 sections with 1 code example: a solid amount of guidance for an agent.

With 3,573 GitHub stars, it is one of the more widely adopted skills in the catalogue.

Substance
26/30
Structure
15/20
Description
15/15
Adoption
15/20
Freshness
15/15

Maintenance, license and trust

  • The repository was last updated about 2 months ago, so relationship-intelligence is actively maintained.
  • It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
  • Its trust signals score 100/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

Safety scan

No issues found

Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.

AI review by kimi-k2.7-code on 2026-11-16. Automated pattern scan on 2026-11-16. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.

relationship-intelligence compared with similar skills

All 4 of these similar skills score higher than relationship-intelligence; compare them before choosing.

SkillScoreStarsUpdatedFormat
relationship-intelligence (this skill)by ericosiu853.6k55d agoSKILL.md
Agent-Reachby Panniantong100112.3k8d agoCLAUDE.md
rufloby ruvnet10081.9ktodayCLAUDE.md
Claude-Code-Game-Studiosby Donchitos10039.7k7d agoCLAUDE.md
Scraplingby D4Vinci100103.6k2d agoMCP Server

Frequently asked questions

How do I install relationship-intelligence?
Run npx skills add ericosiu/ai-marketing-skills --skill relationship-intelligence. The install tabs above show the steps for each supported agent.
Which AI agents does relationship-intelligence work with?
It is written for Universal, as a SKILL.md file. Other agents that read the same format can often use it too.
Is relationship-intelligence safe to use?
Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It is MIT-licensed and scores 100/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is relationship-intelligence still maintained?
The repository was last updated about 2 months ago, so relationship-intelligence is actively maintained.

name: relationship-intelligence description: Consolidate lead feeds and relationship history into evidence-backed client, recruiting, partnership, sponsorship and acquisition opportunities. Use for cross-source identity reconciliation, prior-touch attribution, independent fit tiers and owner-ready action queues. Not an automatic outreach or hiring decision system.

Relationship Intelligence

Repository preamble

When available, run the repository's non-blocking telemetry/version_check.py. Respect existing telemetry consent; never enable remote reporting here. If usage logging is allowed, pass only skill name, elapsed time, success and version to telemetry/telemetry_log.py. Its documented runtime metadata is added by the helper. Never log contacts, evidence, source names, query text or relationship metrics. Standalone copies may skip unavailable helpers.

Turn fragmented signals into a short answer: who matters, for which opportunity, what has already happened, who owns the relationship, and what to do next. Extend an existing lead router rather than create a competing source of truth. This package provides a workflow and offline checks. It does not connect accounts, deploy a service, prove source coverage or start alerts by being installed.

Steps

  1. Freeze the scope. Record sources, accounts, time boundaries, timezone, goals, approved audience and permitted actions. Read the existing router and field ownership before proposing changes. Inventory missing integrations. Keep source coverage alongside every result; no match is not no prior contact.
  2. Normalize without collapsing identities. Follow the data contract. Keep people, companies, assessed properties, evidence and interaction events separate. Verified IDs or corroborated exact addresses can link a person; a shared domain, name or employer cannot by itself. Hold conflicting companies, generic addresses and stale affiliations. Preserve originals and reversible merge provenance.
  3. Reconcile prior touches. Deduplicate stable source events and mirror IDs before counting. Show verified owner-personal, verified team/assistant, automated, inbound and unknown-author outbound separately. Account ownership, writing style and an email signature do not prove personal authorship. Show last verified touch, last reply, pending commitment and coverage dates.
  4. Research and classify five lanes independently. Use approved professional sources and Jev or another authorized provider. Reuse dated evidence where still valid. Read the evaluation and rollout guide. Retain cited facts separately from estimates and model hypotheses. A download, audit request or newsletter signup is not commercial intent. Never propagate one lane's score to another. Unknown is not D or zero.
  5. Propose a useful next action. For each credible opportunity state why this person, why now, one specific angle, evidence, gaps, existing owner, proposed next action and due date. Separate urgency from fit. Existing active deals, promised introductions and replies outrank generic cold outreach suggestions. Keep prior disqualification, opt-outs and owner holds visible and controlling.
  6. Validate and release narrowly. Run offline checks plus representative human review before live routing. The included helper checks event counting; it does not verify evidence or grade the whole system. Preserve calculated CRM fields. Version new advisory fields and record migration/rollback plans. CRM writes, shared dashboards and alerts need the user's specific authorized scope and audience. Research alone grants none of these permissions.
  7. Close the loop. Only approved S-tier alerts go to the approved destination, with evidence, angle, owner and a private record link. Deduplicate alerts by person, lane and material evidence revision. Store attempted/verified/uncertain send states and reconcile uncertainty before retrying. Internal alerts never authorize outreach. Measure reviewed fit, accepted actions, qualified meetings and outcomes, not just enriched rows or messages sent.

Team review view

Provide filters for lane/tier, owner, source, date, evidence freshness, prior touches, reply state, hold reason and next action. Use one row per person with expandable company/property and event evidence, not one row per ingestion event. Show account-level history separately from verified owner-personal history. Display unavailable sources prominently, with direct source links appropriate to the audience. Do not expose private communications or contact data in public exports, alerts or this repository.

Operating boundaries

  • Use only authorized data and enrichment destinations. Minimize private payloads; do not send private messages or sensitive data to a provider without approval.
  • Recruiting tiers are advisory, based on job-related professional evidence. Exclude protected traits and appearance. Humans make hiring decisions.
  • An acquisition fit is not willingness to sell. Sponsorship fit is not budget.
  • Stop on source denials and reconcile access gaps; do not bypass restrictions.
  • Report measured tests and remaining gaps. A requested “95+” is a quality target, not a score to invent or proof of revenue impact.

Offline helper

scripts/touch_summary.py accepts normalized JSON and summarizes deduplicated, verified interaction evidence. It performs no network calls or writes to a CRM. Run synthetic tests with:

python3 -m unittest discover -s relationship-intelligence/tests

Related Skills

View on GitHub
GitHub Stars3.6k
CategoryAutomation
Updated1mo ago
Forks684

Languages

Python

Trust signals

100/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

No cautions
relationship-intelligence — Universal Skill: Install & Safety Check | SkillAgent