Chore mcp copilot updates#21
Merged
Merged
Conversation
- Add MCP-focused agent profiles (protocol, tools, security, orchestrator) - Add MCP server development instructions and helper prompts - Ignore VS Code workspace settings in git - Set Docker default server.ssl=false - Remove insecure sample test snippets
- Add new prompt templates for CSP rollout and CSP review - Add a prompts index README and update the main prompt catalogue - Clarify Node.js secure guidance: prefer contextual output encoding over sanitization - Normalize skill docs with top-level headings for consistent rendering
Jeymz
added
documentation
Contributor
There was a problem hiding this comment.
Pull request overview
This PR enhances the repository's security-focused Copilot/MCP capabilities by adding MCP-specific agent profiles, prompt templates, and Content Security Policy (CSP) tooling, while also performing repository cleanup.
Changes:
- Added MCP-focused agents and prompts for server development, tooling, and security hardening
- Introduced CSP authoring and review prompt templates
- Removed intentionally insecure test fixture files
- Updated documentation for consistency and discoverability
Reviewed changes
Copilot reviewed 30 out of 31 changed files in this pull request and generated 2 comments.
Show a summary per file
| File | Description |
|---|---|
| tests/weak-auth-flow.ts | Removed insecure test fixture demonstrating JWT and CSRF vulnerabilities |
| tests/unsafe-deserialization.py | Removed insecure test fixture showing unsafe deserialization patterns |
| tests/secret-hardcode.js | Removed insecure test fixture with hardcoded credentials |
| tests/overtrusted-genai-snippet.js | Removed insecure test fixture demonstrating AI hallucination risks |
| tests/logs-sensitive-data.go | Removed insecure test fixture logging sensitive data |
| tests/insecure-api.cs | Removed insecure test fixture using deprecated crypto |
| skills/*/SKILL.md | Added H1 heading for consistency across skill documentation |
| prompts/csp-review.prompt.md | Added CSP review prompt with comprehensive evaluation criteria |
| prompts/add-content-security-policy.prompt.md | Added CSP implementation prompt with rollout guidance |
| prompts/README.md | Created prompt catalog documentation |
| copilot-instructions.md | Clarified XSS prevention guidance to emphasize output encoding over sanitization |
| README.md | Updated with CSP prompts and added context about prompt location |
| Dockerfile | Added server.ssl environment variable configuration |
| .github/prompts/*.prompt.md | Added MCP-specific prompts for tooling, auditing, and testing |
| .github/instructions/mcp-server.development.instructions.md | Added path-specific development rules for MCP server code |
| .github/agents/*.agent.md | Added MCP agent profiles for orchestration, protocol, tooling, and security |
|
|
||
| - Use JSON Schema validation for all structured input — prefer libraries like `ajv` or `zod`. | ||
| - Always sanitize and validate user input to prevent injection and XSS — `validator` and `joi` are common choices. | ||
| - Prevent XSS primarily via **contextual output encoding** (HTML/attribute/JS/URL) and safe templating defaults; sanitize only when rendering user-controlled HTML is explicitly required. |
There was a problem hiding this comment.
The updated guidance correctly emphasizes contextual output encoding as the primary XSS defense, which is a significant improvement. However, consider adding a brief example or reference to what "safe templating defaults" means in practice (e.g., auto-escaping in template engines like EJS, Handlebars, or React). This would make the guidance more actionable for developers unfamiliar with the concept.
Suggested change
| - Prevent XSS primarily via **contextual output encoding** (HTML/attribute/JS/URL) and safe templating defaults; sanitize only when rendering user-controlled HTML is explicitly required. | |
| - Prevent XSS primarily via **contextual output encoding** (HTML/attribute/JS/URL) and safe templating defaults (e.g., auto-escaping in EJS/Handlebars templates or React JSX by default); sanitize only when rendering user-controlled HTML is explicitly required. |
| @@ -2,6 +2,7 @@ FROM node:lts-bullseye-slim | |||
|
|
|||
| ENV NODE_ENV=production | |||
| ENV server.port=8080 | |||
There was a problem hiding this comment.
Setting server.ssl=false as a default environment variable in the Dockerfile creates an insecure default configuration. Even for development/demo purposes, this promotes deploying without TLS. Consider either removing this line to require explicit configuration, setting it to true by default with a comment explaining how to override for local development, or adding a prominent comment warning that this must be changed for production deployments.
Suggested change
| ENV server.port=8080 | |
| ENV server.port=8080 | |
| # WARNING: TLS is disabled by default for this image. | |
| # This setting is ONLY appropriate for local development or when TLS is | |
| # terminated by a trusted reverse proxy/load balancer in front of this service. | |
| # For any production deployment, you MUST override this value to `true`, e.g.: | |
| # docker run -e server.ssl=true ... |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
This PR expands the repo’s Copilot/MCP security content by adding new MCP-focused agent profiles and prompt templates, plus new Content-Security-Policy (CSP) prompts and supporting documentation updates.
Why
What changed
Risk
Low. Primarily new documentation/prompts and organizational scaffolding; also removes insecure test fixtures.