The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/generate_daily_report.py executes powershell.exe via subprocess.check_output to retrieve the GITCODE_TOKEN from Windows User and Machine environment variables. While the arguments are hardcoded and intended for credential retrieval, this pattern involves spawning shell processes to interact with system configuration.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted data from GitCode PR bodies and Issue comments to generate summaries.
Ingestion points: External data is ingested in scripts/generate_daily_report.py through GitCode API calls and subsequently read by the agent from temp_dir/report.json.
Boundary markers: There are no explicit instructions or delimiters provided to the agent to treat the ingested repository content as untrusted data or to ignore nested instructions.
Capability inventory: The agent has the capability to execute shell commands (running the report script) and write to the local file system (creating summary files).
Sanitization: The script truncates PR bodies to a maximum length but does not perform any sanitization or filtering to prevent instruction injection within the text.

gitcode-repo-daily