The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted chat data and has the capability to write files to the local system. Evidence chain: (1) Ingestion points: The skill captures context from chat conversations; (2) Boundary markers: The workflow mandates an 'acuerdo previo' (prior agreement) where the agent must propose a plan and wait for user confirmation before writing; (3) Capability inventory: The skill is restricted to writing markdown files to a specific local Obsidian vault; (4) Sanitization: The instructions include a 'Quality bar' item to redact secrets like tokens or passwords unless specifically requested.
[SAFE]: No malicious behaviors or obfuscation were detected. The hardcoded local path (C:\Users\oscar.rondon\Documents\DEV-OBSIDIAN) identifies a specific local vault location consistent with the author's identity and is used for organizational consistency.

obsidian-capture