The Agent Skills Directory

[COMMAND_EXECUTION]: The skill interpolates the <issue> argument directly into a shell command (gh issue view <issue>). If this argument is not properly sanitized before execution, it could allow for command injection if an attacker provides a malicious string instead of a valid issue number.
[INDIRECT_PROMPT_INJECTION]: The skill ingests issue titles and bodies from GitHub, which are then used as context for an LLM (Jev or Claude) to determine a difficulty tier. This tier influences subsequent agent behavior, such as model selection or skipping tasks. A maliciously crafted GitHub issue could include instructions to manipulate the triage outcome.
Ingestion points: GitHub issue title, body, and labels retrieved via gh issue view in SKILL.md.
Boundary markers: None. The issue content is processed and summarized without explicit delimiters or instructions to ignore embedded commands.
Capability inventory: The skill utilizes Bash, Write, Read, Glob, and Grep tools. It executes subprocesses including gh, git grep, and curl.
Sanitization: No sanitization or safety filtering is performed on the external content retrieved from GitHub.
[EXTERNAL_DOWNLOADS]: The skill performs outbound network requests to api.typesafe.ai using curl to transmit issue state and receive triage scores. While this is the intended functionality for the TypeSafe AI integration, it involves sending data to an external endpoint.

triage