The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides an interface to persist user-supplied facts in a knowledge base, which establishes a potential vulnerability surface for indirect prompt injection if malicious instructions are stored and subsequently retrieved. \n
Ingestion points: User input is accepted via the title and content arguments of the teamkb_propose tool call (SKILL.md). \n
Boundary markers: The instructions do not specify any delimiters or boundary markers to wrap the ingested fact content. \n
Capability inventory: The skill possesses write capabilities through the teamkb_propose and teamkb_transition tools, and status monitoring via teamkb_status (SKILL.md). \n
Sanitization: The instructions include a specific requirement to strip secrets, tokens, credentials, and private keys from the content before persisting it (SKILL.md). \n- [EXTERNAL_DOWNLOADS]: The skill links to official documentation and source code repositories for the system it manages. \n
Source: The skill references the jeremylongshore/bobs-big-brain-registrar repository on GitHub for build and configuration guidance (SKILL.md).

brain-save