camel-knowledge

name	camel-knowledge
description	Look up Camel documentation, components, CVEs/security advisories, and versions.
user_invocable	false

Camel Knowledge — Apache Camel Documentation Reference

Invocable standalone via /camel-knowledge or loaded by pipeline skills for documentation lookup.

When NOT to use this skill

• Implementation tasks (generating routes, writing code) — use /camel-execute instead
• Route validation or quality analysis — use /camel-validate instead
• Designing integrations — use /camel-brainstorm for design work; this skill answers factual questions

Purpose

Provides access to Apache Camel documentation via MCP knowledge tools. Used during:

• Brainstorm: checking component availability, finding migration guides
• Execute: verifying component availability during quality review, checking for CVEs

MCP Tools

The generated MCP allowlist is defined in workflow/camel-kit-workflow.yaml under the camel-knowledge MCP server. Keep this table aligned with that manifest.

Tool	Purpose	When to Use
camel_docs_search	General documentation search	Migration guides, getting started, EIP patterns, general questions
camel_docs_component_info	Component documentation and CVE lookup	Verify component exists, get configuration reference, check security
camel_docs_cve_search	CVE security advisory search	Security review, vulnerability check by CVE ID, component, or severity
camel_docs_release_info	Release notes for a version	Version selection, understanding what changed
camel_docs_jira_lookup	JIRA issue lookup by ID	Find which release fixed a specific issue

Usage Pattern

Component Availability Check

camel_docs_component_info(component="kafka", version="4.14")

Returns availability status.

CVE Check (quality review)

camel_docs_cve_search(cve_id="CVE-2024-22369")
camel_docs_cve_search(component="sql", severity="HIGH")

Migration Guide Search

camel_docs_search(query="migrating from Camel 3 to Camel 4.14", version="4.14", max_results=5)

JIRA Issue Lookup

camel_docs_jira_lookup(jira_id="CAMEL-22784")

Subagent Dispatch Pattern

When invoked from within another skill (not standalone), the orchestrator should dispatch knowledge queries as a knowledge-researcher subagent (from agents/knowledge-researcher.md). This keeps full MCP search results out of the orchestrator context — only the synthesized answer flows back.

Standalone invocation (/camel-knowledge): runs inline in the current context.

Pipeline invocation (from camel-brainstorm, camel-execute, etc.): dispatch as subagent:

1. Build the subagent prompt with:
   • The knowledge-researcher persona (full text from agents/knowledge-researcher.md)
   • The specific question or lookup request
   • Relevant context (Camel version, component name, CVE ID, etc.)
2. The subagent runs the appropriate MCP tools
3. Only the structured answer (see agents/knowledge-researcher.md output format) flows back

This pattern prevents ~2000-5000 tokens of raw search results per query from accumulating in the orchestrator context. For a migration brainstorm with 5-10 knowledge lookups, this saves 10,000-50,000 tokens.

Important Notes

• Pass max_results=5 unless more results are needed
• If a tool call fails, warn and continue — don't block the pipeline
• Never fabricate documentation content — if no results, say so