dot-pr-spec-summary
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from GitHub PRs, issues, and external URLs, which could contain instructions designed to manipulate the agent's behavior during summarization.
- Ingestion points: Data is ingested from GitHub PR titles, bodies, and diffs via the
ghCLI, issue descriptions, local files, and external web content (SKILL.md). - Boundary markers: Absent; no specific delimiters or instructions to ignore embedded commands are mandated for the fetched data.
- Capability inventory: The skill uses
ghfor data retrieval,WebFetchfor web access, and performs file system writes for report generation (SKILL.md). - Sanitization: Absent; the skill does not specify validation or filtering of the content retrieved from external sources.
- [COMMAND_EXECUTION]: The workflow involves instructions for the agent to execute a background shell command (
python3 -m http.server) for verifying the rendering of generated reports. - Running a local web server, even for verification, introduces a risk of exposing the working directory if the process is not strictly bound to the loopback interface or if the process remains active after use.
- [EXTERNAL_DOWNLOADS]: Generated HTML reports are configured to fetch the Mermaid diagramming library from a well-known content delivery network.
- The skill references
https://cdn.jsdelivr.net/npm/mermaid@11/dist/mermaid.esm.min.mjs(SKILL.md), which is an established service for diagram rendering.
Audit Metadata