issues-workflow

Pass

Audited by Gen Agent Trust Hub on Oct 8, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill frequently ingests untrusted data from external sources, specifically GitHub issue bodies, comments, and project metadata.
  • Ingestion points: Operations such as gh issue view, gh issue list, and GraphQL queries for project items (references/graphql-queries.md) fetch content that is attacker-controlled.
  • Boundary markers: The instructions do not define boundary markers or 'ignore embedded instructions' warnings when processing this external content.
  • Capability inventory: The skill has access to powerful tools including Bash, Write, and GitHub CLI operations (gh issue edit, gh project, gh api graphql) which could be misused if the agent follows instructions embedded in issues.
  • Sanitization: No sanitization or validation of the fetched issue content is prescribed before the agent acts upon it.
  • [DYNAMIC_EXECUTION]: The skill includes a 'Self-Evolving Skill' directive in SKILL.md which explicitly instructs the agent to 'fix this file immediately' if instructions are found to be incorrect or if workarounds are needed. This self-modification mechanism, while intended for optimization, creates a persistence vector where malicious instructions ingested via indirect prompt injection could be permanently written into the skill's own code.
  • [CREDENTIALS_UNSAFE]: The documentation in SKILL.md and references/graphql-queries.md explicitly references the location of sensitive authentication tokens (~/.claude/tools/bin/gh-token-for-repo). While this appears to be a standard path for the platform's GitHub integration, identifying and utilizing credential-retrieval scripts in skill instructions increases the risk of accidental exposure or misuse.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 8, 2026, 02:11 AM