---
description: Use oper8r workflow, search, graph exploration, and integration-status tools for grounded retrieval and synthesis.
alwaysApply: false
---

# Oper8r Workflows, Search, And Graph Explorer

## Role

Use these tools to retrieve, synthesize, and explain oper8r knowledge. They are read/synthesis tools, not graph curation tools.

- Do not mutate the graph with these tools.
- Do not store new facts, update temporals, merge entities, or curate employee graph data through these tools.
- Use the employee graph curator skill only when the user explicitly asks to curate source-backed shared internal graph entities/facts.
- Preserve the authenticated user's permissions and privacy boundaries when quoting or summarizing retrieved context.

## Tool Selection

- `oper8r_get_account_context`: account briefs for meeting prep, renewals, expansion planning, multithread audits, and general GTM context.
- `oper8r_find_precedent`: prior deal, objection, RFP, security, negotiation, or internal examples with citations and recommended actions.
- `oper8r_diagnose_risk`: forecast, slip, churn, expansion, or general account risk diagnosis.
- `oper8r_generate_message`: grounded outbound, follow-up, renewal, expansion, security follow-up, or general message drafts.
- `oper8r_answer_rfp_security_question`: evidence-backed RFP/security questionnaire answers with confidence, missing evidence, citations, and review guidance.
- `oper8r_search`: search across the authenticated user's oper8r knowledge graph.
- `oper8r_graph_explore`: explore entities, facts, chunks, and relationship neighborhoods from text or an entity anchor.
- `oper8r_graph_getEntity`: fetch one entity by exact oper8r graph entity ID or deterministic external ID.
- `oper8r_graph_getEdges`: fetch relationship edges around one resolved entity.
- `oper8r_integrations_status`: check lightweight integration health/status without exposing credentials, tokens, connection IDs, or raw provider metadata.

## Search And Graph Exploration

Use search when the user asks for documents, facts, examples, or broad evidence. Use graph exploration when the user asks how entities relate, who/what is connected, or why a result is relevant.

- Prefer deterministic anchors when available: entity IDs, emails, CRM IDs, Google Drive IDs, or other external IDs.
- Use `entityScope.externalIds` in `oper8r_search` when the user gives a stable ID and wants scoped retrieval.
- Use `oper8r_graph_getEntity` before `oper8r_graph_getEdges` when you have an external ID and need exact identity resolution.
- Use temporal fields such as `asOf`, `validBetween`, and `includeInvalidated` only when the user asks for history, past state, or invalidated/superseded facts.
- Separate retrieved facts from your inferences. Do not fabricate sources or cite results that were not returned.

## Workflow Discipline

For account and GTM workflows, make the output operational and grounded:

- Start from the user's goal, account/deal/entity anchor, time horizon, and requested lens.
- Include citations or source references when the tool returns them.
- Call out missing evidence plainly when confidence is low.
- For risk diagnosis, distinguish signals, interpretation, severity, confidence, and next actions.
- For precedents, explain why each precedent is similar and what can be reused.
- For generated messages, return drafts and rationale only; do not send messages or imply delivery.
- For RFP/security answers, include confidence, missing evidence, and recommended reviewer when the evidence is incomplete.

## Privacy And Integration Boundaries

- Never expose credentials, access tokens, connection IDs, OAuth details, or raw provider metadata.
- Do not quote private or sensitive retrieved content unless it is necessary for the user's requested task and permitted by the authenticated context.
- Treat integration health as diagnostic status only; do not infer hidden credentials or connection internals.
- Do not transform retrieved customer/account context into employee graph writes unless the user separately invokes the employee graph curation workflow with source-backed shared internal context.