The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the user's Heptabase database which could contain adversarial instructions.
Ingestion points: Multiple commands documented in SKILL.md and reference files allow the agent to ingest external data, specifically heptabase note read, heptabase journal read, heptabase pdf read, heptabase audio read, and heptabase video read.
Boundary markers: Absent. The instructions do not specify the use of delimiters or provide the agent with a strategy to distinguish between system instructions and data ingested from cards.
Capability inventory: The skill possesses extensive write and delete capabilities via Bash(heptabase *), including note create/save, journal create/save, and various whiteboard mutation commands.
Sanitization: No sanitization or validation of the ingested content is performed before it is processed by the agent.

heptabase-cli