The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes several Python scripts (scripts/setup_jira_env.py, scripts/check_jira_env.py, scripts/run_with_jira_env.py) that utilize subprocess.run to interact with the local environment.
setup_jira_env.py and check_jira_env.py use git commands (git rev-parse, git ls-files, git check-ignore) to ensure that sensitive credential files (.krt/env/jira-scribe.env) are correctly initialized and excluded from version control.
run_with_jira_env.py acts as a wrapper that loads Jira credentials from the local environment file and executes a provided command (typically curl for Jira API calls) as a child process. This pattern is designed to prevent sensitive tokens from being passed as plaintext arguments or stored in shell history.
[CREDENTIALS_SAFE]: The skill adopts a defensive posture regarding authentication. It explicitly forbids asking the user for credentials or printing tokens. It mandates the use of a project-local environment file that must be verified as Git-ignored before use. The references/jira-api.md file provides clear instructions on secure authentication using Bearer tokens.
[INDIRECT_PROMPT_INJECTION]: The skill addresses potential injection from processed Jira data by implementing a text validator (scripts/check_jira_text.py). This script checks for Spanish semantic presence and blocks operational patterns like commit prefixes, PR chatter, and internal IDs in autonomous mutations, reducing the risk of the agent misinterpreting external data as instructions.
[SAFE]: The skill requires explicit human confirmation for all Jira mutations (creation, transitions, backlinks) unless an audited autonomous ledger is in use. It also implements a 'two-endpoint strategy' for credential verification to handle common Jira Server API quirks gracefully without compromising security.

krt-jira-scribe