The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests conversation history to synthesize wiki notes, creating a potential vector for indirect prompt injection.
Ingestion points: Current conversation history is processed in Step 1 and Quick Mode of SKILL.md.
Boundary markers: Absent; there are no instructions to the agent to treat conversation content as untrusted or to use specific delimiters to prevent the agent from following instructions embedded in the chat.
Capability inventory: The skill performs file writing to the local filesystem (OBSIDIAN_VAULT_PATH) and executes shell commands (qmd) as documented in SKILL.md.
Sanitization: Absent; no sanitization, escaping, or validation of the conversation text is performed before it is rewritten into declarative wiki notes.
[COMMAND_EXECUTION]: The skill triggers shell commands to update and verify a search index after writing vault files.
Evidence: SKILL.md contains instructions to run ${QMD_CLI:-qmd} update and ${QMD_CLI:-qmd} embed.
[DYNAMIC_EXECUTION]: The skill allows the execution path for the indexing tool to be determined dynamically via environment variables.
Evidence: The use of ${QMD_CLI:-qmd} in SKILL.md allows the shell to execute a command defined by the QMD_CLI environment variable, which could lead to arbitrary command execution if the environment is compromised.

wiki-capture