The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources (Jira, Linear, GitHub Issues) to automate various parts of the development workflow, including branch naming and pull request content.
Ingestion points: External data is ingested through tracker API calls and CLI commands in references/jira.md, references/linear.md, and references/github-issues.md (e.g., gh issue view, mcp__claude_ai_Linear__get_issue, and Jira REST API responses).
Boundary markers: The skill does not define or use explicit delimiters or instructions to ignore embedded prompts when processing the content of ticket titles or descriptions.
Capability inventory: The skill possesses the capability to execute shell commands for git operations, use the GitHub CLI (gh) for PR and issue management, and perform file writes in the local .dev-flow/ directory.
Sanitization: While the skill applies kebab-case formatting to branch names, it largely passes external ticket content directly into pull request summaries and titles without rigorous sanitization.
Specialist Skill Selection: In Step 5, the skill dynamically suggests and loads other skills based on the ticket context, which is a potential vector where malicious ticket content could influence the agent's environment, though this is mitigated by the fact that it only selects from currently available skills.

dev-flow