# Intelligence Pillar (placeholder)

The Intelligence pillar turns the clean, comparable, traceable dataset
produced by the Baseline pillar into analyst-quality insight.

## How to add an Intelligence agent

1. Create `src/agents/intelligence/<agent-name>/` with the same five-file
   structure used by Baseline agents:
   - `schema.ts` — zod input/output schemas
   - `runbook.ts` — numbered processing steps (Std 6)
   - `tools.ts` — declared toolset (Std 5, approved + connected)
   - `index.ts` — the executor; returns `AgentResult<T>` (Std 11/12)
   - `README.md` — objective, inputs, outputs, runbook, tools, triggers, failure modes
2. Implement an `AgentStandardsContract` and the 12 universal standards.
3. Layer the **5 Intelligence-specific standards** on top (to be defined when
   the pillar is built — e.g. insight derivation rules, traceability of
   inferred claims, anti-fabrication guardrails for synthesis).
4. Register the executor + contract in `src/orchestrator.ts` (single
   switch case) and append the step to `src/pipeline.ts`:
   ```ts
   { kind: 'agent', pillar: 'intelligence', agent: '<agent-name>' }
   ```
5. Optionally insert a HITL checkpoint between Baseline and Intelligence:
   ```ts
   { kind: 'checkpoint', name: 'records-review' }
   ```

Nothing in the Baseline pillar or the orchestrator's outer loop has to
change to add Intelligence — the pipeline is data.