{
  "markdown": "# brandketing\n\nA Claude Code plugin that turns raw product and customer data into a living marketing intelligence context store - and compiles it into briefs and copy on demand.\n\n## What it does\n\nInstead of filling out a brief from scratch every time, this plugin maintains a **context store** that accumulates structured marketing objects over time. Each asset you generate pulls from approved, sourced, scored objects - not from memory or imagination.\n\nThe two source documents behind this system:\n- `brief-harness.md` - automated brief compilation from 5 data pipelines\n- `marketing-harness.md` - 6-step strategy workflow + critic pass for copy generation\n\n## Install\n\n```\n/plugin install brandketing@claude-plugins-official\n```\n\nOr from GitHub:\n```\n/plugin install brandketing@github:entro314-labs/brandketing\n```\n\n## Commands\n\n| Command | What it does |\n|---|---|\n| `/brandketing:ingest <source> <pipeline>` | Normalize a raw source file into structured context objects |\n| `/brandketing:review [type]` | Approve, reject, or edit pending objects |\n| `/brandketing:brief <deliverable> [icp]` | Compile a brief from approved objects |\n| `/brandketing:generate [deliverable] [icp]` | Run strategist + critic, output full copy hierarchy |\n| `/brandketing:status` | Show context store freshness and object counts |\n\n## Pipelines\n\n| Pipeline | Source types | Output objects |\n|---|---|---|\n| `product-truth` | Docs, PRDs, changelogs, demos | Capability |\n| `customer-truth` | Call transcripts, tickets, reviews | Pain, Quote, UseCase |\n| `commercial-truth` | CRM, won/lost, account data | Persona |\n| `market-truth` | Competitor pages, comparison content | Competitor, Cliche |\n| `proof-truth` | Analytics, case studies, wins | Proof |\n\n## Agents\n\n- **normalizer** - converts raw sources into typed, scored objects\n- **compiler** - assembles brief sections from approved objects, filtered by ICP\n- **strategist** - runs the 6-step strategy workflow to produce a messaging hierarchy\n- **critic** - red-teams every output line and rewrites what's weak\n\n## Context store\n\nEach project gets its own store at `.brandketing/context/store/` (not committed to git).\n\nObjects: `capabilities`, `pains`, `quotes`, `use_cases`, `personas`, `competitors`, `cliches`, `proofs`, `brand`\n\nEach object carries: confidence score, source references, freshness date, and approval status.\n\n## Workflow\n\n```\n1. /brandketing:ingest call-transcripts.txt customer-truth\n2. /brandketing:ingest product-docs.md product-truth\n3. /brandketing:review\n4. /brandketing:generate homepage-hero icp-1\n```\n\nFor a lightweight version without the context store, see **brandketing-quick**.\n\n## Design principles\n\n- **Machine assembles, human approves.** The system proposes; a human decides ICP, category, and positioning bet.\n- **Every claim has a source.** Objects without source references are not created.\n- **Freshness is tracked.** Static / slow-moving / fast-moving layers refresh on different cadences.\n- **Copy goes through a critic.** Every generation runs a red-team pass before output.\n",
  "bytes": 3058,
  "sha": "d083ed9320ac6438f9d3aba3728655003929cced19da59741a9cbe42317e654c",
  "repo_slug": "entro314-labs/brandketing",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_entro314_labs_brandketing_brandketing_d1edb215/readme"
}