wiki-stage-commit
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to read and process files from the
_staging/directory, which are described as LLM-written content. This creates an attack surface where instructions embedded in these files could attempt to influence the agent's review or merge process.\n - Ingestion points: The skill reads content from
_staging/and the target wiki pages in SKILL.md Step 2 and Step 3.\n - Boundary markers: Absent. The instructions do not provide explicit delimiters or warnings to ignore instructions found within the processed files.\n
- Capability inventory: The agent has the ability to execute
obsidian-wikiCLI commands, read/write files in the local vault, and perform manual text merges.\n - Sanitization: Absent. Content is displayed to the user and merged into the wiki without sanitization or escaping.\n- [COMMAND_EXECUTION]: The skill relies on executing shell commands via the
obsidian-wikiCLI to perform its functions.\n - Evidence: The skill uses commands such as
obsidian-wiki staging list --json,obsidian-wiki staging promote,obsidian-wiki staging discard, andobsidian-wiki memory syncto manage the vault state.
Audit Metadata