The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill presents an attack surface where instructions embedded in processed data could influence the agent.
Ingestion points: The skill reads state files from ~/.openclaw/skill-state/ belonging to other skills (daily-review, long-running-task-management, task-handoff), as defined in SKILL.md and implemented in run.py.
Boundary markers: No boundary markers, delimiters, or 'ignore instructions' warnings are used when interpolating this external data into the briefing output in the format_briefing function within run.py.
Capability inventory: The skill performs local file reads (load_state) and writes (save_own_state) to manage its state, and outputs the briefing text to standard output for the agent to process.
Sanitization: There is no evidence of sanitization, escaping, or validation of the content retrieved from the source state files before it is presented to the agent.

morning-briefing