Skip to content

armgit / ECC

Public Imported from GitHub
0

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

refactor: fold social graph ranking into lead intelligence

Affaan Mustafaf056952e5051
diff --git a/AGENTS.md b/AGENTS.md
index f8fdee4f3..d92bde107 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -1,6 +1,6 @@
# Everything Claude Code (ECC) — Agent Instructions
-This is a **production-ready AI coding plugin** providing 30 specialized agents, 140 skills, 60 commands, and automated hook workflows for software development.
+This is a **production-ready AI coding plugin** providing 36 specialized agents, 142 skills, 68 commands, and automated hook workflows for software development.
**Version:** 1.9.0
@@ -139,9 +139,9 @@ Troubleshoot failures: check test isolation → verify mocks → fix implementat
## Project Structure
```
-agents/ — 30 specialized subagents
-skills/ — 140 workflow skills and domain knowledge
-commands/ — 60 slash commands
+agents/ — 36 specialized subagents
+skills/ — 142 workflow skills and domain knowledge
+commands/ — 68 slash commands
hooks/ — Trigger-based automations
rules/ — Always-follow guidelines (common + per-language)
scripts/ — Cross-platform Node.js utilities
diff --git a/README.md b/README.md
index 19927e013..b12356710 100644
--- a/README.md
+++ b/README.md
@@ -220,7 +220,7 @@ For manual install instructions see the README in the `rules/` folder. When copy
/plugin list everything-claude-code@everything-claude-code
```
-**That's it!** You now have access to 30 agents, 140 skills, and 60 commands.
+**That's it!** You now have access to 36 agents, 142 skills, and 68 commands.
### Multi-model commands require additional setup
@@ -295,7 +295,7 @@ everything-claude-code/
| |-- plugin.json # Plugin metadata and component paths
| |-- marketplace.json # Marketplace catalog for /plugin marketplace add
|
-|-- agents/ # 30 specialized subagents for delegation
+|-- agents/ # 36 specialized subagents for delegation
| |-- planner.md # Feature implementation planning
| |-- architect.md # System design decisions
| |-- tdd-guide.md # Test-driven development
@@ -1109,9 +1109,9 @@ The configuration is automatically detected from `.opencode/opencode.json`.
| Feature | Claude Code | OpenCode | Status |
|---------|-------------|----------|--------|
-| Agents | PASS: 30 agents | PASS: 12 agents | **Claude Code leads** |
-| Commands | PASS: 60 commands | PASS: 31 commands | **Claude Code leads** |
-| Skills | PASS: 140 skills | PASS: 37 skills | **Claude Code leads** |
+| Agents | PASS: 36 agents | PASS: 12 agents | **Claude Code leads** |
+| Commands | PASS: 68 commands | PASS: 31 commands | **Claude Code leads** |
+| Skills | PASS: 142 skills | PASS: 37 skills | **Claude Code leads** |
| Hooks | PASS: 8 event types | PASS: 11 events | **OpenCode has more!** |
| Rules | PASS: 29 rules | PASS: 13 instructions | **Claude Code leads** |
| MCP Servers | PASS: 14 servers | PASS: Full | **Full parity** |
diff --git a/manifests/install-modules.json b/manifests/install-modules.json
index 0b2704098..8854ea790 100644
--- a/manifests/install-modules.json
+++ b/manifests/install-modules.json
@@ -284,8 +284,7 @@
"skills/investor-materials",
"skills/investor-outreach",
"skills/lead-intelligence",
- "skills/market-research",
- "skills/social-graph-ranker"
+ "skills/market-research"
],
"targets": [
"claude",
diff --git a/skills/lead-intelligence/SKILL.md b/skills/lead-intelligence/SKILL.md
index c7a10a8d5..634448b8d 100644
--- a/skills/lead-intelligence/SKILL.md
+++ b/skills/lead-intelligence/SKILL.md
@@ -93,6 +93,52 @@ For each scored target, analyze the user's social graph to find the warmest path
| Industry alignment | 15% — same vertical = natural intro |
| Mutual's X handle / LinkedIn | 10% — identifiability for outreach |
+### Weighted Bridge Ranking
+
+Treat this as the canonical network-ranking stage for lead intelligence. Do not run a separate graph skill when this stage is enough.
+
+Given:
+- `T` = target leads
+- `M` = your mutuals / existing connections
+- `d(m, t)` = shortest hop distance from mutual `m` to target `t`
+- `w(t)` = target weight from signal scoring
+
+Compute the base bridge score for each mutual:
+
+```text
+B(m) = Σ_{t ∈ T} w(t) · λ^(d(m,t) - 1)
+```
+
+Where:
+- `λ` is the decay factor, usually `0.5`
+- a direct connection contributes full value
+- each extra hop halves the contribution
+
+For second-order reach, expand one level into the mutual's own network:
+
+```text
+B_ext(m) = B(m) + α · Σ_{m' ∈ N(m) \\ M} Σ_{t ∈ T} w(t) · λ^(d(m',t))
+```
+
+Where:
+- `N(m) \\ M` is the set of people the mutual knows that you do not
+- `α` is the second-order discount, usually `0.3`
+
+Then rank by response-adjusted bridge value:
+
+```text
+R(m) = B_ext(m) · (1 + β · engagement(m))
+```
+
+Where:
+- `engagement(m)` is a normalized responsiveness score
+- `β` is the engagement bonus, usually `0.2`
+
+Interpretation:
+- Tier 1: high `R(m)` and direct bridge paths -> warm intro asks
+- Tier 2: medium `R(m)` and one-hop bridge paths -> conditional intro asks
+- Tier 3: no viable bridge -> direct cold outreach using the same lead record
+
### Output Format
```
diff --git a/skills/social-graph-ranker/SKILL.md b/skills/social-graph-ranker/SKILL.md
deleted file mode 100644
index 0d5431186..000000000
--- a/skills/social-graph-ranker/SKILL.md
+++ /dev/null
@@ -1,199 +0,0 @@
----
-name: social-graph-ranker
-description: Weighted social graph traversal that ranks your network connections by proximity to target leads. Uses exponential decay across hops, parallel execution with lead-intelligence skill, and API-driven outreach prioritization. Replaces Apollo, Clay, and manual networking.
-metadata:
- tags: networking, outreach, leads, graph-theory, x-api, linkedin, exa
----
-
-## When to Use
-
-Use this skill when you need to:
-- Find warm intro paths to specific people or companies
-- Rank your existing connections by networking value
-- Identify which mutuals to ask for introductions
-- Prioritize outbound outreach by warmth and proximity
-- Map your social graph against a target lead list
-
-## How It Works
-
-### Architecture
-
-Two parallel pipelines feed a unified ranking engine:
-
-```
-┌─────────────────────────────────────────────────────────┐
-│ SOCIAL GRAPH RANKER │
-├──────────────────────┬──────────────────────────────────┤
-│ INBOUND PIPELINE │ OUTBOUND PIPELINE │
-│ │ │
-│ Your Connections │ Target Lead List (ICP) │
-│ ┌──────────────┐ │ ┌──────────────────┐ │
-│ │ X Mutuals │ │ │ lead-intelligence │ │
-│ │ X Followers │ │ │ skill (parallel) │ │
-│ │ LI Connections│ │ │ Exa + X API + │ │
-│ └──────┬───────┘ │ │ enrichment agents │ │
-│ │ │ └────────┬─────────┘ │
-│ ▼ │ ▼ │
-│ ┌──────────────┐ │ ┌──────────────────┐ │
-│ │ Connection │ │ │ Ranked Lead List │ │
-│ │ Graph Build │ │ │ (scored by ICP │ │
-│ │ (adjacency) │ │ │ fit + response │ │
-│ └──────┬───────┘ │ │ probability) │ │
-│ │ │ └────────┬─────────┘ │
-│ └────────────┼──────────────┘ │
-│ ▼ │
-│ ┌──────────────────────┐ │
-│ │ GRAPH INTERSECTION │ │
-│ │ Match connections │ │
-│ │ against targets │ │
-│ └──────────┬───────────┘ │
-│ ▼ │
-│ ┌──────────────────────┐ │
-│ │ WEIGHTED RANKING │ │
-│ │ Exponential decay │ │
-│ │ across hops │ │
-│ └──────────┬───────────┘ │
-│ ▼ │
-│ ┌──────────────────────┐ │
-│ │ PRIORITIZED OUTPUT │ │
-│ │ 1. Warm intro asks │ │
-│ │ 2. Direct outreach │ │
-│ │ 3. Network gaps │ │
-│ └──────────────────────┘ │
-└─────────────────────────────────────────────────────────┘
-```
-
-### The Math: Weighted Graph Proximity Score
-
-Given:
-- **T** = set of target leads you want to reach
-- **M** = set of your mutuals/connections
-- **G** = social graph (adjacency)
-- **d(m, t)** = shortest path distance from mutual m to target t
-
-For each mutual m, compute:
-
-**Bridge Score:**
-
-```
-B(m) = Σ_{t ∈ T} w(t) · λ^{d(m,t) - 1}
-```
-
-Where:
-- `w(t)` = target weight (from lead-intelligence signal score: role 30%, industry 25%, activity 20%, influence 10%, location 10%, engagement 5%)
-- `λ` = decay factor, typically 0.5 (halves value each hop)
-- `d(m,t)` = hop distance (1 = direct connection, 2 = mutual-of-mutual, etc.)
-- Convention: `d(m,t) = 1` for direct connection, so `λ^0 = 1` (full value)
-
-**Properties:**
-- Direct connection to target: contributes `w(t) · 1.0`
-- One hop away: contributes `w(t) · 0.5`
-- Two hops: contributes `w(t) · 0.25`
-- Three hops: contributes `w(t) · 0.125`
-- Effectively zero beyond ~6 hops (Gaussian/exponential decay → 0)
-
-**Extended Score (second-order network value):**
-
-For deeper traversal, also consider the mutual's own network reach:
-
-```
-B_ext(m) = B(m) + α · Σ_{m' ∈ N(m) \ M} Σ_{t ∈ T} w(t) · λ^{d(m',t)}
-```
-
-Where:
-- `N(m) \ M` = connections of m that you DON'T already have
-- `α` = second-order discount (typically 0.3)
-- This captures: "even if m doesn't know my targets directly, m knows people I don't, who might"
-
-**Final Ranking:**
-
-```
-R(m) = B_ext(m) · (1 + β · engagement(m))
-```
-
-Where:
-- `engagement(m)` = normalized score of how responsive m is (reply rate, interaction frequency)
-- `β` = engagement bonus weight (typically 0.2)
-
-### Execution Steps
-
-1. **Build Target List**
- - Run lead-intelligence skill in parallel to generate scored ICP leads
- - Or provide manual target list with names/handles
-
-2. **Harvest Your Graph**
- - X API: `GET /2/users/:id/following` and `GET /2/users/:id/followers`
- - LinkedIn: connection export CSV or browser-use scraping
- - Build adjacency map: `mutual → [their connections]`
-
-3. **Intersect and Score**
- - For each mutual, check which targets they follow/connect with
- - Compute B(m) with decay
- - For top-k mutuals, expand one more hop and compute B_ext(m)
-
-4. **Generate Output**
- - Tier 1: Mutuals with B(m) > threshold → warm intro requests
- - Tier 2: Targets with no warm path → direct cold outreach via lead-intelligence
- - Tier 3: Network gaps → suggest who to follow/connect with to build bridges
-
-5. **Draft Messages**
- - Warm intro: "Hey [mutual], I saw you're connected to [target]. I'm working on [context]. Would you be open to making an intro?"
- - Uses outreach-drafter agent from lead-intelligence for personalization
-
-### Configuration
-
-```yaml
-# Target definition
-targets:
- - handle: "@targetperson"
- platform: x
- weight: 0.9 # override signal score
-
-# Decay parameters
-decay_factor: 0.5 # λ — halve value per hop
-max_depth: 3 # don't traverse beyond 3 hops
-second_order_discount: 0.3 # α — discount for network-of-network
-engagement_bonus: 0.2 # β — bonus for responsive mutuals
-
-# API configuration
-x_api:
- bearer_token: $X_BEARER_TOKEN
- rate_limit_delay: 1.1 # seconds between API calls
-linkedin:
- method: csv_export # or browser_use
- csv_path: ~/Downloads/Connections.csv
-```
-
-### Integration with lead-intelligence
-
-This skill runs IN PARALLEL with lead-intelligence:
-- lead-intelligence generates the target list (T) with signal scores
-- social-graph-ranker maps your network against those targets
-- Combined output: prioritized outreach plan with warm paths where available
-
-### Example Output
-
-```
-BRIDGE RANKING — Top 10 Mutuals by Network Value
-═══════════════════════════════════════════════════
-
-#1 @alex_quant (B=4.72)
- Direct → @kalshi_ceo (w=0.9), @polymarket_shayne (w=0.85)
- 1-hop → @a16z_crypto (w=0.7, via @defi_mike)
- Action: Ask for intros to Kalshi + Polymarket
-
-#2 @sarah_vc (B=3.15)
- Direct → @sequoia_partner (w=0.95)
- 1-hop → @yc_gustaf (w=0.8, via @batch_founder)
- Action: Ask for Sequoia intro
-
-#3 @dev_community (B=2.88)
- Direct → @cursor_ceo (w=0.6), @vercel_guillermo (w=0.6)
- 2-hop → @anthropic_dario (w=0.95, via @cursor_ceo → @anthropic_team)
- Action: Ask for Cursor intro, mention Anthropic angle
-
-NETWORK GAPS — No warm path exists
-═══════════════════════════════════
-@target_x — Suggest following @bridge_person_1, @bridge_person_2
-@target_y — Direct cold outreach recommended (lead-intelligence draft ready)
-```