The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of shell command execution across multiple components to orchestrate Git and GitHub operations. Specifically, resources/scripts/repo-analyzer.js utilizes execSync for repository analysis, and resources/scripts/release-orchestrator.py uses subprocess.run to execute Git and curl commands. While functional, this pattern presents a risk if command arguments are derived from untrusted external data.
[EXTERNAL_DOWNLOADS]: The skill requires and executes external dependencies from unverified sources. The setup instructions and examples (e.g., examples/example-1-repo-automation.md) direct the agent to install and run claude-flow@alpha and flow-nexus@latest via NPM/npx. These tools are not part of the established trusted vendor or service lists and represent an unverified dependency risk.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by processing untrusted data from GitHub.
Ingestion points: GitHub webhook payloads including Pull Request titles, descriptions, and issue comments (resources/scripts/webhook-handler.js).
Boundary markers: Absent. The skill does not implement delimiters or instructions to ignore embedded commands within the processed data.
Capability inventory: The agent has access to execSync, subprocess.run, file-writing tools, and authenticated GitHub API access via octokit.
Sanitization: Minimal. The skill uses simple regular expressions to detect slash commands in comments and keywords in issue text, which could be bypassed or manipulated by malicious actors to influence agent behavior.
[DYNAMIC_EXECUTION]: The skill implements logic to generate and execute scripts at runtime. For instance, example-1-repo-automation.sh and release-orchestrator.js are constructed and then executed to perform complex orchestration tasks. This dynamic generation increases the risk of command injection if the templates are populated with unsanitized data.
[DATA_EXPOSURE]: The Library-First Protocol defined in SKILL.md instructs the agent to search and extract code patterns and data from broad local system paths (D:\Projects\*). This instruction could lead to the unintended exposure of sensitive information from unrelated projects residing on the agent's host system.

github-integration