thehawg-bigdawg

name	thehawg-bigdawg
description	GTM engineering operating skill for theHAWG-BIGDAWG. Fires when user asks about prospecting, enrichment, ICP scoring, outbound campaigns, or competitive positioning. Follows SENSE → ORIENT → ACT → DEPOSIT protocol. Use when user says "find accounts", "score leads", "run outbound", "check competitive position", or "update the GTM knowledge graph".

theHAWG-BIGDAWG GTM Engineering Skill

Operate the GTM engineering loop for TheHog.ai using the Context OS knowledge graph.

First-read order

1. README.md
2. progress.md
3. plans/findings.md
4. plans/task_plan.md
5. docs/mcp-setup.md
6. docs/thehog-api-reference.md
7. knowledge_base/
8. 00_foundation/

Protocol: SENSE → ORIENT → ACT → DEPOSIT

SENSE — Read the Environment

Before any action, verify current state:

1. Check graph health:

   context-os graph-exec --graph knowledge_base '(() => {
     const r = codemode.graph_query({ filter: {}, limit: 500 });
     const orphans = r.nodes.filter(n => n.link_count.outbound === 0 && n.link_count.inbound === 0);
     return JSON.stringify({ total: r.total, orphans: orphans.length });
   })()'
   

2. Check what exists: Read 00_foundation/_synthesis/gtm-context-os-synthesis.md for current state overview.

3. Check credentials: Verify THEHOG_ACCESS_KEY and THEHOG_SECRET_KEY are set.

4. Check credit budget: If making API calls, verify budget headroom first.

ORIENT — Find What Matters

1. Identify hub nodes:

   context-os graph-exec --graph knowledge_base '(() => {
     const r = codemode.graph_query({ filter: {}, limit: 500 });
     return JSON.stringify(r.nodes.sort((a,b) =>
       (b.link_count.outbound+b.link_count.inbound)-(a.link_count.outbound+a.link_count.inbound)
     ).slice(0,5).map(n => ({ name: n.name, links: n.link_count.outbound+n.link_count.inbound })));
   })()'
   

2. Read relevant ICP segment: Match the user's intent to [[icp-founder-solo]], [[icp-gtm-engineer]], or [[icp-revops-head]].

3. Read scoring criteria: Check 00_foundation/positioning/icp-scoring-engine.md for current weights.

4. Pick the owner document:

   • Atomic company/product/persona facts → knowledge_base/
   • Strategy, scoring, playbooks, reusable systems → 00_foundation/
   • Session observations → plans/findings.md
   • Status and next steps → progress.md

ACT — Execute the Task

For prospecting:

• Use [[hog-api-search]] patterns to construct queries
• Execute via [[mcp-integration]] or REST API
• Score results using [[icp-scoring-engine]] weights

For outbound:

• Select template track from 00_foundation/messaging/outbound-playbook.md
• Personalize using enrichment data
• Log campaign to knowledge graph

For competitive research:

• Start with 00_foundation/positioning/competitive-battlecards.md
• Update with new intelligence
• Link findings to relevant ICP segments

DEPOSIT — Reinforce the Graph

Every action must produce a deposit:

1. Create or update knowledge node with proper frontmatter
2. Link to at least 2 existing nodes via wiki-links (e.g. [[gtm-engineering]])
3. Update plans/findings.md if gaps are discovered
4. Update synthesis doc if material state changes

MCP Connection

Use environment variables only:

• THEHOG_ACCESS_KEY
• THEHOG_SECRET_KEY
• THEHOG_MCP_URL

Default hosted MCP URL:

https://mcp.thehog.ai/mcp

Hosted MCP uses OAuth when the client supports remote MCP. Local stdio uses npx -y @thehog/mcp@latest with THEHOG_ACCESS_KEY and THEHOG_SECRET_KEY in the process environment.

Do not hardcode credentials. Do not paste credentials into prompts or markdown.

Rate-limit and Async Discipline

• Treat 5 requests per second as an unconfirmed ceiling; prefer lower throughput until official limits are known.
• Poll async jobs with backoff.
• Handle terminal statuses: succeeded, failed, partial_success, and cancelled.
• Deep research polling should be slower than enrichment polling.
• Stop on repeated failures, 402s, 429s, or unclear credit usage.

Safe First Ingestion

1. Run companies/search for TheHog.ai or Hog.ai.
2. Poll until terminal state.
3. Add confirmed facts to knowledge_base/business/thehog-ai-company.md.
4. Log operation metadata and summary in plans/findings.md.
5. Update progress.md.

Quality Gates

• SENSE phase completed (graph health checked, state verified)?
• ORIENT phase completed (hub nodes identified, relevant segments read)?
• ACT phase produced verifiable output?
• DEPOSIT phase linked output to graph (no orphan nodes)?
• Attribution tags on every claim?

Intent Matching

User says	Do this
"find accounts" / "prospect"	Search → Enrich → Score → Deposit
"score leads"	Read enrichment data → Apply scoring → Deposit ranked list
"run outbound"	Read scored accounts → Select templates → Generate outbound
"competitive intel"	Read battlecards → Research → Update → Deposit
"graph health"	Run graph-exec health queries → Report → Fix orphans
"what's missing"	Read plans/findings.md → Run gap analysis → Update

Never Do

• Never commit .env.
• Never hardcode secrets.
• Never run credit-heavy jobs without explicit intent.
• Never overwrite existing graph facts without preserving provenance.
• Never skip the DEPOSIT step after discovering reusable information.