aspice-check 0.2.0

Orchestrator composing confluence-ai and aspice-eval into a pipeline CLI and MCP server

aspice-check

Orchestrator composing confluence-ai and aspice-eval into a pipeline CLI and MCP server.

Pipeline CLI — aspice-analyze

Run a full ASPICE gap analysis pipeline on a Confluence SDP page: export → evaluate → publish.

Usage

aspice-analyze <PAGE_URL> --target-level <1-5> --groups <GROUPS> [OPTIONS]

Examples

# Full pipeline: export, evaluate at level 2, publish report back to Confluence
aspice-analyze \
  "https://acme.atlassian.net/wiki/spaces/ENG/pages/12345/My-SDP" \
  --target-level 2 \
  --groups SWE,MAN \
  --email [e-mail] \
  --api-token YOUR_TOKEN \
  --region us-east-1

# Evaluate without publishing
aspice-analyze \
  "https://acme.atlassian.net/wiki/spaces/ENG/pages/12345/My-SDP" \
  --target-level 1 \
  --groups SWE \
  --no-publish

# Use OpenAI instead of Bedrock
aspice-analyze \
  "https://acme.atlassian.net/wiki/spaces/ENG/pages/12345/My-SDP" \
  --target-level 3 \
  --groups SWE,SYS,MAN,SUP \
  --provider openai \
  --model gpt-4o

Options

Option	Env Variable	Description
PAGE_URL	—	Confluence Cloud page URL (required)
--target-level	—	ASPICE capability level 1–5 (required)
--groups	—	Comma-separated process groups, e.g. SWE,MAN (required)
--email	CONFLUENCE_EMAIL	Confluence account email
--api-token	CONFLUENCE_API_TOKEN	Confluence API token
--provider	ASPICE_EVAL_PROVIDER	AI provider: bedrock, openai, anthropic (default: bedrock)
--model	—	AI model name (default depends on provider)
--region	AWS_DEFAULT_REGION	AWS region (required for Bedrock)
--report-title	—	Custom title for the published report page
--output-dir	—	Local directory for intermediate artifacts
--no-publish	—	Skip publishing report to Confluence
--verbose	—	Enable DEBUG-level logging
--quiet	—	Suppress progress messages

Exit Codes

Code	Meaning
0	Success
1	Parameter validation error
2	Export stage failure
3	Evaluation stage failure
4	Publish stage failure

MCP Server — aspice-mcp

An MCP (Model Context Protocol) server exposing evaluation and Confluence tools to AI assistants. It uses stdio transport (JSON-RPC 2.0) and works with any MCP-compatible client.

Configuration

The server is configured in your MCP client's config file. The exact location depends on the client:

Client	Config file
Claude Desktop (macOS)	~/Library/Application Support/Claude/claude_desktop_config.json
Claude Desktop (Windows)	%APPDATA%\Claude\claude_desktop_config.json
Kiro	.kiro/settings/mcp.json (workspace) or ~/.kiro/settings/mcp.json (global)
VS Code (Copilot)	.vscode/mcp.json

Add the following to your config:

{
  "mcpServers": {
    "aspice": {
      "command": "aspice-mcp",
      "args": [],
      "env": {
        "CONFLUENCE_EMAIL": "[e-mail]",
        "CONFLUENCE_API_TOKEN": "your-token",
        "AWS_DEFAULT_REGION": "us-east-1"
      }
    }
  }
}

The env block passes credentials to the server process. Tools that need Confluence access (export_page) or AI providers (evaluate_sdp, describe_image) will use these values. You can omit credentials here and pass them per-tool-call instead.

If aspice-mcp is not on your PATH (e.g. installed in a virtualenv), use the full path:

{
  "mcpServers": {
    "aspice": {
      "command": "/path/to/venv/bin/aspice-mcp",
      "args": []
    }
  }
}

Starting Manually (for testing)

aspice-mcp

The server reads JSON-RPC requests from stdin and writes responses to stdout. Logs go to stderr.

Tool Inventory

Tool	Description
evaluate_sdp	Evaluate an SDP document against ASPICE knowledge base criteria
validate_kb	Validate a knowledge base directory for schema compliance and completeness
list_standards	List available knowledge base standards and their process groups
export_page	Export a Confluence Cloud page to Markdown with AI image descriptions
describe_image	Generate an AI description of an image file

Tool Parameters

evaluate_sdp — provider (required), model (required), sdp_path, sdp_content, target_level (1–5, default 3), process_groups, standard

validate_kb — kb_path (required), standard

list_standards — kb_path (optional, uses bundled KB if omitted)

export_page — page_url (required), output_dir (required), email (required), api_token (required), ai_provider, ai_model, output_format

describe_image — image_path (required), provider (required), model (required), is_gliffy, page_title

Example Tool Call (JSON-RPC)

{"jsonrpc": "2.0", "id": 1, "method": "tools/call", "params": {"name": "validate_kb", "arguments": {"kb_path": "/path/to/knowledge_base", "standard": "aspice"}}}

Response:

{"jsonrpc": "2.0", "id": 1, "result": {"content": [{"type": "text", "text": "{\"is_valid\": true, ...}"}]}}

Error Handling

Invalid parameters return a structured error with code -32602:

{"jsonrpc": "2.0", "id": 1, "error": {"code": -32602, "message": "Invalid params", "data": {"tool": "evaluate_sdp", "parameter": "provider", "actual_value": "gpt", "valid_values": ["bedrock", "openai", "anthropic"], "suggestion": "Use one of: bedrock, openai, anthropic"}}}

Installation

From the monorepo (development)

Install all three packages in editable mode from the repo root:

pip install -e ./confluence-ai
pip install -e ./aspice-eval
pip install -e ./aspice-check

This registers the aspice-analyze and aspice-mcp commands in your environment. Verify:

which aspice-mcp
# → /path/to/venv/bin/aspice-mcp

From PyPI (once published)

pip install aspice-check

This pulls in confluence-ai and aspice-eval automatically.

Making aspice-mcp available to MCP clients

MCP clients launch the server as a subprocess, so the aspice-mcp command must be resolvable from the client's environment. Two options:

1. Use the absolute path in your MCP config (works regardless of PATH):

   { "command": "/path/to/venv/bin/aspice-mcp" }
   

2. Activate the venv before launching the client, or install into the system Python so aspice-mcp is on the global PATH.

To find the path after installing:

which aspice-mcp

Requires Python 3.10+.

License

MIT