The Agent Skills Directory

[COMMAND_EXECUTION]: Utilizes standard Git commands (git fetch, git pull, git push, git stash) to synchronize the notes repository with a remote host. This is the primary mechanism for the skill's advertised functionality.
[COMMAND_EXECUTION]: Executes a local bash script (scripts/redact_check.sh) to perform pre-commit scanning for potential secrets such as AWS keys, private keys, and API tokens.
[DATA_EXFILTRATION]: Automatically pushes note content, summaries, and attachments to a remote repository. While this involves transferring data externally, it is the primary intended function of the skill and is protected by a best-effort automated redaction process.
[PROMPT_INJECTION]: The skill ingests untrusted user input from various sources (chat, voice transcripts, file attachments) and uses it to generate notes and task lists. This presents a surface for indirect prompt injection.
Ingestion points: User messages, voice transcriptions, and file contents processed in SKILL.md (Step 1).
Boundary markers: The note template in assets/note-template.md does not use explicit delimiters or "ignore embedded instructions" warnings to isolate user-provided content.
Capability inventory: Includes file system modification, shell command execution (Git and grep-based scripts), and network operations via Git protocol.
Sanitization: The skill implements credential redaction via scripts/redact_check.sh but does not perform specific sanitization to prevent LLM instruction injection from processed data.

note-taker