The Agent Skills Directory

[COMMAND_EXECUTION]: The script md-to-confluence.py uses subprocess.run to execute the pandoc binary for converting markdown to HTML. While this is the intended functionality, it involves passing local file paths to a subprocess.
[COMMAND_EXECUTION]: The confluence-preflight.py script uses subprocess.run to call a sibling script (confluence-to-md.py) to fetch remote content for conflict detection.
[DYNAMIC_EXECUTION]: The SKILL.md file provides a Python one-liner in the 'Post-push verification' section for the agent to execute via python3 -c. This script dynamically accesses sensitive credentials from environment variables (CONFLUENCE_TOKEN) and performs network requests to verify page status.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests untrusted markdown data and remote Confluence content while having the capability to perform network operations and execute shell commands.
Ingestion points: md-to-confluence.py and confluence-preflight.py read local markdown files and fetch page content from the Confluence REST API.
Boundary markers: The skill does not use specific delimiters or instructions to prevent the agent from following directives embedded in the markdown files.
Capability inventory: The skill uses subprocess.run to execute external tools, performs network operations via http.client, and writes snapshot files to the local filesystem.
Sanitization: The skill performs regex-based transformations to convert HTML to Confluence storage format but does not sanitize content against prompt injection instructions.

md-to-confluence