The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/run.py executes the seren CLI tool to resolve database connectivity details. The execution uses controlled arguments derived from configuration.
[DATA_EXFILTRATION]: The skill communicates with a user-defined PostgreSQL database to retrieve transaction history. This behavior is consistent with the skill's primary function and does not involve unauthorized data transfer.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes raw transaction descriptions from bank data and incorporates them into reports. Evidence chain: (1) Ingestion points: scripts/run.py retrieves transaction descriptions from the wf_transactions table. (2) Boundary markers: The generated Markdown reports do not use explicit delimiters or warnings to isolate untrusted data. (3) Capability inventory: The skill can write to the local filesystem and the configured database. (4) Sanitization: Basic payee normalization is performed in scripts/recurring_detector.py using regular expressions.

recurring-transactions