Security audit
Security checks for vulnerabilities and agentic risk
This is a Markdown-only reference skill for using Atlassian's CLI, with high-impact Jira and admin commands disclosed and constrained by explicit confirmation guidance.
Install only if you intend to let Codex help operate an authenticated Atlassian CLI. Review bulk selectors, deletion/archive commands, admin user commands, and feedback attachments carefully before approving execution, and avoid exposing API tokens in commands, logs, or shell history.
In this markdown/manifest-style file, the description says to use the skill whenever the user wants to perform a wide range of Jira, admin, or automation tasks, but it does not define specific trigger phrases, exclusions, or scope limits. That breadth makes invocation conditions ambiguous and increases the chance of unintended activation for ordinary Atlassian-related conversation.
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
- `--from-file "items.txt"` - file with keys/IDs (comma/whitespace/newline separated)
Use `--ignore-errors` to continue past failures in bulk operations.
Use `--yes` / `-y` to skip confirmation prompts (useful for automation).
### Pagination
- `--limit N` - max items to return (defaults vary: 30-50)
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
| --from-json | string | JSON file with array of issue objects |
| --generate-json | | Output example JSON structure |
| --ignore-errors | | Continue past failures |
| --yes | | Skip confirmation |
Examples:
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
| --from-json | string | JSON file with array of issue objects |
| --generate-json | | Output example JSON structure |
| --ignore-errors | | Continue past failures |
| --yes | | Skip confirmation |
Examples:
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
| --from-json | string | JSON file with array of issue objects |
| --generate-json | | Output example JSON structure |
| --ignore-errors | | Continue past failures |
| --yes | | Skip confirmation |
Examples:
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
| --from-json | string | JSON file with array of issue objects |
| --generate-json | | Output example JSON structure |
| --ignore-errors | | Continue past failures |
| --yes | | Skip confirmation |
Examples:
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
| --from-json | string | JSON file with array of issue objects |
| --generate-json | | Output example JSON structure |
| --ignore-errors | | Continue past failures |
| --yes | | Skip confirmation |
Examples:
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
| --from-json | string | JSON file with array of issue objects |
| --generate-json | | Output example JSON structure |
| --ignore-errors | | Continue past failures |
| --yes | | Skip confirmation |
Examples:
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
| --from-json | string | JSON file with array of issue objects |
| --generate-json | | Output example JSON structure |
| --ignore-errors | | Continue past failures |
| --yes | | Skip confirmation |
Examples:
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
| --from-json | string | JSON file with array of issue objects |
| --generate-json | | Output example JSON structure |
| --ignore-errors | | Continue past failures |
| --yes | | Skip confirmation |
Examples:
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
| --from-json | string | JSON file with array of issue objects |
| --generate-json | | Output example JSON structure |
| --ignore-errors | | Continue past failures |
| --yes | | Skip confirmation |
Examples:
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
| --from-json | string | JSON file with array of issue objects |
| --generate-json | | Output example JSON structure |
| --ignore-errors | | Continue past failures |
| --yes | | Skip confirmation |
Examples:
The bulk edit/create documentation exposes a --yes confirmation bypass for mass-changing operations without instructing the user to preview or verify the affected set first. In an agent skill context, this can enable unintended large-scale Jira changes if an LLM constructs query/file-driven commands autonomously or from ambiguous user input.
Bulk assignment supports --jql, --filter, and --yes, allowing mass reassignment without a documented review step. In a skill used by an autonomous agent, this increases the risk of silently reassigning many issues to the wrong user, disrupting workflows and accountability.
Bulk transition with --jql or --filter and --yes can move many issues between workflow states without a mandated review. This is dangerous because mass status changes can trigger downstream automations, SLAs, notifications, or reporting errors across a project.
The clone command supports bulk selectors and cross-project or cross-site duplication, but lacks a warning that this may replicate many work items unexpectedly. In an agent setting, this can create widespread duplication, data sprawl, and accidental propagation of sensitive descriptions or metadata to other projects/sites.
Comment deletion is a destructive action and the documentation omits any warning or confirmation guidance. In Jira, comments may contain operational context, approvals, or audit-relevant discussion, so accidental deletion can remove important project history.
Link deletion supports bulk input from JSON/CSV and --yes, but there is no warning about irreversible removal of issue relationships. Removing links in bulk can break dependency tracking, duplicate detection, and workflow coordination across many issues.
The feedback command allows attaching arbitrary files and sending freeform details/email, but the documentation does not warn that attachments may contain sensitive data such as tokens, logs, customer data, or internal configuration. In an admin/Jira automation context, users may attach diagnostic files from systems containing credentials or regulated data, creating a realistic risk of unintended data disclosure.
Bulk archive can make many issues read-only and effectively remove them from normal workflows, yet the documentation does not warn users when using query selectors with --yes. While less severe than deletion, accidental archival can still interrupt active work and obscure records.
Watcher removal changes another user's notification/watch state, but the documentation provides no caution about its effect. Although not highly destructive, an agent could silently remove stakeholders from updates and reduce visibility into issue changes.
This markdown reference documents jira project archive, which changes project availability and can affect user workflows, but it does not include any caution or confirmation guidance. Nearby destructive or state-changing commands such as project delete and user deactivate explicitly warn the operator, so the omission is notable here.
The admin auth login section instructs users to provide an organization API key, which is sensitive credential material, but the warning about safe handling appears only after the examples. A clearer upfront warning that credentials are sensitive and should not be exposed in commands or logs would better satisfy user disclosure expectations for privacy-impacting operations.
No suspicious patterns detected.