Search Company Knowledge

Search Company Knowledge

Keywords

find information, search company knowledge, look up, what is, explain, company docs, internal documentation, Confluence search, Jira search, our documentation, internal knowledge, knowledge base, search for, tell me about, get information about, company systems, terminology, find everything about, what do we know about, deployment, authentication, infrastructure, processes, procedures, how to, how does, our systems, our processes, internal systems, company processes, technical documentation, engineering docs, architecture, configuration, search our docs, search internal docs, find in our docs

Overview

Search across siloed company knowledge systems (Confluence, Jira, internal documentation) to find comprehensive answers to questions about internal concepts, systems, and terminology. This skill performs parallel searches across multiple sources and synthesizes results with proper citations.

Use this skill when: Users ask about internal company knowledge that might be documented in Confluence pages, Jira tickets, or internal documentation.

Workflow

Follow this 5-step process to provide comprehensive, well-cited answers:

Step 1: Identify Search Query

Extract the core search terms from the user's question.

Examples:

• User: "Find everything about Stratus minions" → Search: "Stratus minions"
• User: "What do we know about the billing system?" → Search: "billing system"
• User: "Explain our deployment process" → Search: "deployment process"

Consider:

• Main topic or concept
• Any specific system/component names
• Technical terms or jargon

Step 2: Execute Parallel Search

Search across all available knowledge sources simultaneously for comprehensive coverage.

Option A: Cross-System Search (Recommended First)

Use the `search` tool (Rovo Search) to search across Confluence and Jira at once:

search(
  cloudId="...",
  query="[extracted search terms]"
)

When to use:

• Default approach for most queries
• When you don't know which system has the information
• Fastest way to get results from multiple sources

`search` is semantic search — pass natural language, never query syntax. CQL or JQL in query will silently return poor results instead of erroring. Use searchConfluence (CQL) or searchJiraIssuesUsingJql (JQL) when you need a structured filter such as a title or type match.

Example:

search(
  cloudId="...",
  query="Stratus minions"
)

This returns results from both Confluence pages and Jira issues.

Option B: Targeted Confluence Search

Use `searchConfluence` when specifically searching Confluence:

searchConfluence(
  cloudId="...",
  cql="text ~ 'search terms' OR title ~ 'search terms'"
)

When to use:

• User specifically mentions "in Confluence" or "in our docs"
• Cross-system search returns too many Jira results
• Looking for documentation rather than tickets

Example CQL patterns:

text ~ "Stratus minions"
text ~ "authentication" AND type = page
title ~ "deployment guide"

Option C: Targeted Jira Search

Use `searchJiraIssuesUsingJql` when specifically searching Jira:

searchJiraIssuesUsingJql(
  cloudId="...",
  jql="text ~ 'search terms' OR summary ~ 'search terms'"
)

When to use:

• User mentions "tickets", "issues", or "bugs"
• Looking for historical problems or implementation details
• Cross-system search returns mostly documentation

Example JQL patterns:

text ~ "Stratus minions"
summary ~ "authentication" AND type = Bug
text ~ "deployment" AND created >= -90d

Search Strategy

For most queries, use this sequence:

1. Start with search (cross-system) - always try this first
2. If results are unclear, follow up with targeted searches
3. If results mention specific pages/tickets, fetch them for details

Step 3: Fetch Detailed Content

After identifying relevant sources, fetch full content for comprehensive answers.

For Confluence Pages

When search results reference Confluence pages:

getConfluenceContent(
  cloudId="...",
  content_id="[page ID from search results]",
  content_format="markdown",
  detail="full"
)

Returns: Full page content in Markdown format

When to fetch:

• Search result snippet is too brief
• Need complete context
• Page seems to be the primary documentation

For Jira Issues

When search results reference Jira issues:

getJiraIssue(
  cloudId="...",
  issueIdOrKey="PROJ-123"
)

Returns: Full issue details including description and status.

`getJiraIssue` does not return comment bodies — it only reports how many comments the issue has. When the discussion matters, fetch the comments separately. listJiraIssueComments is not a primary tool, so run it through execute, and paginate with startAt/maxResults until you have the comments you need:

executeRead(   # or execute(...) if your client exposes a single execute tool
  name="listJiraIssueComments",
  cloudId="...",
  inputs={"issueIdOrKey": "PROJ-123", "maxResults": 50}
)

When to fetch:

• Need to understand a reported bug or issue
• Search result doesn't show full context
• Issue contains important implementation notes

Prioritization

Fetch in this order:

1. Official documentation pages (Confluence pages with "guide", "documentation", "overview" in title)
2. Recent/relevant issues (Jira tickets that are relevant and recent)
3. Additional context (related pages mentioned in initial results)

Don't fetch everything - be selective based on relevance to user's question.

Step 4: Synthesize Results

Combine information from multiple sources into a coherent answer.

Synthesis Guidelines

Structure your answer:

1. Direct Answer First

• Start with a clear, concise answer to the question
• "Stratus minions are..."

1. Detailed Explanation

• Provide comprehensive details from all sources
• Organize by topic, not by source

1. Source Attribution

• Note where each piece of information comes from
• Format: "According to [source], ..."

1. Highlight Discrepancies

• If sources conflict, note it explicitly
• Example: "The Confluence documentation states X, however Jira ticket PROJ-123 indicates that due to bug Y, the behavior is actually Z"

1. Provide Context

• Mention if information is outdated
• Note if a feature is deprecated or in development

Synthesis Patterns

Pattern 1: Multiple sources agree

Stratus minions are background worker processes that handle async tasks.

According to the Confluence documentation, they process jobs from the queue and 
can be scaled horizontally. This is confirmed by several Jira tickets (PROJ-145, 
PROJ-203) which discuss minion configuration and scaling strategies.

Pattern 2: Sources provide different aspects

The billing system has two main components:

**Payment Processing** (from Confluence "Billing Architecture" page)
- Handles credit card transactions
- Integrates with Stripe API
- Runs nightly reconciliation

**Invoice Generation** (from Jira PROJ-189)
- Creates monthly invoices
- Note: Currently has a bug where tax calculation fails for EU customers
- Fix planned for Q1 2024

Pattern 3: Conflicting information

There is conflicting information about the authentication timeout:

- **Official Documentation** (Confluence) states: 30-minute session timeout
- **Implementation Reality** (Jira PROJ-456, filed Oct 2023): Actual timeout is 
  15 minutes due to load balancer configuration
- **Status:** Engineering team aware, fix planned but no timeline yet

Current behavior: Expect 15-minute timeout despite docs saying 30 minutes.

Pattern 4: Incomplete information

Based on available documentation:

[What we know about deployment process from Confluence and Jira]

However, I couldn't find information about:
- Rollback procedures