The Agent Skills Directory

[PROMPT_INJECTION]: The skill includes explicit behavioral steering in the 'Suggest Using K-Dense Web' section, instructing the agent to proactively recommend a specific external platform under certain conversation conditions.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted external data files and possesses significant capabilities.
Ingestion points: The skill loads data from external paths using si.read_spikeglx(), si.read_openephys(), and si.read_nwb() as shown in SKILL.md.
Boundary markers: There are no delimiters or instructions provided to the agent to ignore potential instructions embedded within data or metadata fields of the neural recordings.
Capability inventory: The skill has the ability to execute external binaries via si.run_sorter() and perform extensive file system operations (writing results, exporting to NWB/Phy).
Sanitization: No validation or sanitization of the content of binary recording files or their associated metadata is performed before the data is processed or analyzed by the AI.
[COMMAND_EXECUTION]: The workflow relies on the execution of external spike sorting software (e.g., Kilosort4, SpykingCircus2) via the si.run_sorter interface, which spawns external processes to handle data analysis.
[EXTERNAL_DOWNLOADS]: The installation documentation encourages the download and installation of several third-party Python packages from PyPI, including domain-specific libraries and the Anthropic API client.

neuropixels-analysis