milestones
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill extensively uses the GitHub CLI (
gh api,gh milestone,gh issue) to read and modify repository metadata. It also invokes secondary automated skills such asfix-it,test-it, andmerge-itto perform implementation work. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from GitHub issue titles, descriptions, and milestone metadata which could contain malicious instructions.
- Ingestion points: Data is pulled from GitHub via
gh issue list --milestoneand milestone description lookups. - Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are defined for the ingested issue or milestone content.
- Capability inventory: The skill possesses significant capabilities, including GitHub API mutations (create, update, close milestones), file system writes (repository documentation and
TASKS.md), and the ability to execute other automated agents. - Sanitization: There is no evidence of explicit sanitization, escaping, or validation of the external content before it is processed.
- [DYNAMIC_EXECUTION]: The skill implements a "fan out" mechanism where it spawns multiple subagents to handle independent issues in parallel. It dynamically selects model tiers (e.g.,
haiku,sonnet,opus) for these sub-tasks, increasing the complexity of the execution environment. - [DATA_EXPOSURE]: The skill accesses local repository files such as
AGENTS.md,TASKS.md, and internal handbook documentation to guide its operations and establish context.
Audit Metadata