release-openclaw-nightly
Warn
Audited by Gen Agent Trust Hub on Jul 23, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
OPENCLAW_ALLOW_ROOT=1environment variable when running theopenclaw cron runcommand. This flag enables root-level privileges for the automated task, which represents a privilege escalation risk if the task or environment is compromised.\n- [PROMPT_INJECTION]: The skill implements logic to process and act upon commands received via Discord mentions (e.g.,@Tideclaw run alpha now). This creates an indirect prompt injection attack surface where users in specific Discord channels can trigger significant repository actions.\n - Ingestion points: Discord messages/mentions in the
#releasesand#maintainerschannels.\n - Boundary markers: None identified; the agent is instructed to act directly on the mention text.\n
- Capability inventory: Git write access (branch creation, deletion, and pushing), GitHub workflow dispatch via the
/usr/local/bin/gh-tideclaw-writewrapper, and automated task execution viaopenclaw cron run.\n - Sanitization: No explicit sanitization or validation of the Discord input is specified before performing sensitive operations.
Audit Metadata