The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous shell commands for the agent to execute, including complex pipelines using curl, jq, and the GitHub CLI (gh api) to interact with remote services and local files.
[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install external tools such as the zizmor Python package and the rhysd/actionlint GitHub extension. It also advises downloading the jenkins-cli.jar from target servers.
[REMOTE_CODE_EXECUTION]: The skill instructs the agent to execute a downloaded JAR file (jenkins-cli.jar) from an untrusted target server, which could lead to code execution on the agent's host if the file is malicious.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The skill reads data from potentially untrusted external sources including public workflow logs, build artifacts, and Terraform state files via gh api and curl commands in SKILL.md.
Boundary markers: No specific boundary markers or instructions to ignore embedded commands are present when processing the content of logs or artifacts.
Capability inventory: The skill possesses capabilities for network requests (curl), file manipulation, and binary execution (java, bash).
Sanitization: There is no evidence of sanitization or validation of the content ingested from logs or state files before it is processed by the agent.

hunt-cicd