czlonkowski
1530f09bbc
feat: Add indirect prompt injection guidance to AI agent pattern (v1.6.0)
...
Adds a "Security: Treat Tool Output as Untrusted Input" section to
ai_agent_workflow.md covering guardrails for agents that fetch
third-party content (HTTP, Serper, Wikipedia, GitHub, MCP Client).
Guidance: gate destructive tools behind human approval, use
least-privilege scopes, constrain the system prompt, validate
structured outputs, and keep tool calls auditable.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com >
2026-04-19 19:26:46 +02:00
czlonkowski
eca0d13eef
feat: Add production learnings to skills and optimize triggering (v1.5.0)
...
Incorporate real-world n8n workflow building experience into all 7 skills:
- SplitInBatches loop semantics, cross-iteration data, nested loops
- Google Sheets/Drive integration gotchas
- pairedItem, correct node references, float precision patterns
- Batch Processing as 6th core workflow pattern
- MCP tool parameter corrections (updates vs parameters, credential format)
- $env blocking awareness, bracket notation for special chars
Optimize all skill descriptions for better triggering accuracy:
- n8n-mcp-tools-expert: triggers on any n8n MCP tool usage
- n8n-workflow-patterns: triggers on workflow creation/design
- n8n-code-javascript: triggers on Code node usage
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-04-13 11:36:13 +02:00
czlonkowski
0f0c473165
chore: Bump plugin/marketplace version to 1.4.0
...
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-04-04 13:17:46 +02:00
czlonkowski
3db664c1bc
fix: Correct zip structure for Claude.ai/Desktop skill uploads (v1.3.0)
...
Individual skill zips now have SKILL.md at the top level (<skill-name>/SKILL.md)
instead of nested under skills/ directory, fixing uploads to Claude.ai and
Claude Desktop. Also adds missing n8n-code-javascript and n8n-code-python
individual zips. Closes #19
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com >
2026-04-03 08:50:25 +02:00
Romuald Członkowski
d9c2872029
feat: Update skills for n8n-mcp unified tool API (v1.1.0)
...
BREAKING: Updated all skills to reflect n8n-mcp tool consolidation:
## Tool API Changes
- get_node_essentials → get_node({detail: "standard"})
- get_node_info → get_node({detail: "full"})
- get_node_documentation → get_node({mode: "docs"})
- search_node_properties → get_node({mode: "search_properties"})
- validate_node_minimal → validate_node({mode: "minimal"})
- validate_node_operation → validate_node({mode: "full"})
- get_property_dependencies → REMOVED (use get_node modes)
## New Features Documented
- Workflow activation via API (activateWorkflow/deactivateWorkflow operations)
- n8n_deploy_template - deploy templates directly to n8n
- n8n_workflow_versions - version history and rollback
- n8n_test_workflow - trigger execution
- n8n_executions - manage executions
- Smart parameters (branch, case) for IF/Switch connections
- Intent parameter for better AI responses
## Documentation Updates
- Added YouTube video introduction with thumbnail
- Added GitHub stars badge (1.2k milestone)
- Added build.sh script for dist packaging
- Fixed "5 skills" → "7 skills" inconsistency in README
## Files Updated
- n8n-mcp-tools-expert: Complete rewrite of SKILL.md, SEARCH_GUIDE.md,
VALIDATION_GUIDE.md, WORKFLOW_GUIDE.md
- n8n-node-configuration: Updated SKILL.md, DEPENDENCIES.md
- n8n-validation-expert: Updated SKILL.md, ERROR_CATALOG.md, FALSE_POSITIVES.md
- n8n-workflow-patterns: Updated SKILL.md, README.md
- README.md, CLAUDE.md: Modernized documentation
Conceived by Romuald Członkowski - www.aiadvisors.pl/en
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com >
2026-01-08 15:37:57 +01:00
