The Agent Skills Directory

[DATA_EXFILTRATION]: The utility script scripts/update-comment-wiki.mjs is designed to access sensitive local files for Jira authentication. It attempts to read client certificates and keys from ~/.config/jira-certs/client.crt and ~/.config/jira-certs/client.key, and it imports a token resolver that reads ~/.config/jira-certs/jira-pat.txt. These credentials are used to authorize REST API calls to the user-provided JIRA_URL.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the PR environment (such as Root cause, Fix summary, and Verification notes) and interpolates it into a Jira Wiki-markup comment. This creates a surface where malicious instructions embedded in a PR could be written into a Jira ticket and potentially influence other agents or automated systems that process that ticket.
Ingestion points: Data enters via the "Host field map" defined in SKILL.md.
Boundary markers: The skill uses a structured Wiki-markup skeleton to delimit the content.
Capability inventory: The skill can transition Jira issue states and perform network writes (POST/PUT) to the Jira API.
Sanitization: The skill implements basic heading normalization (rewriting h1. and h2. headings) to maintain comment structure, but it does not filter for potential prompt injection content.
[COMMAND_EXECUTION]: The documentation in SKILL.md includes an npx command hint for manual skill installation. While this involves command execution, it is presented as a standard setup instruction for the user and is not an automated capability of the skill itself.

jira-status-writeback