Process one raw statement into normalized parsed evidence via Docling CLI.
npx skills add https://github.com/SamArgt/auto-bean --skill auto-bean-process
Please help me install this Agent Skill. Skill: auto-bean-process Source: https://github.com/SamArgt/auto-bean/tree/main/skill_sources/auto-bean-process Command: npx skills add https://github.com/SamArgt/auto-bean --skill auto-bean-process
Process exactly one assigned raw statement file into normalized parsed evidence through the local Docling CLI. Use as the auto-bean-import raw-to-parsed worker stage. It owns parsing, process artifacts, per-input parse status updates, and processing memory suggestions; auto-bean-import owns work discovery, first-seen accounts, posting workflows, and finalization.
Assign one raw statement from statements/raw/ to this skill to produce a normalized parsed output under statements/parsed/ and update the corresponding import-status entry.
High-intent search queries and answers about installing and using this skill.
To parse a raw statement into structured data, a worker skill processes the assigned file using the local Docling CLI. It validates the input format, generates a deterministic parse run ID, and stores the normalized output under statements/parsed/.
Tracking parsing status is done by a per-input worker that updates the import-status entry and writes process artifacts. It assigns a deterministic parse_run_id and stores processing metadata under .auto-bean/artifacts/process.
No, it does not handle the entire import workflow. It only processes one assigned raw statement into parsed evidence, while the broader auto-bean-import workflow handles work discovery, first-seen accounts, posting workflows, and finalization.
Yes, the worker validates input formats before processing. It checks the raw statement file before using the local Docling CLI to transform it into a normalized parsed representation and writing the corresponding process artifacts.
Memory suggestions work by surfacing processing guidance during the raw-to-parsed stage. The worker provides memory-driven suggestions for the individual input but defers any broader orchestration or finalization tasks to the main import workflow.
Using a separate worker ensures deterministic processing and isolated status tracking for each file. It processes exactly one assigned raw statement, writes a specific parse_run_id, and prevents one failed input from blocking the entire import workflow.
SamArgtOrchestrate end-to-end Beancount statement imports for a Git-backed ledger workspace.
SamArgtCategorize a parsed evidence artifact and determine reconciliation actions for posting.
SamArgtManage durable memory records in auto-bean fixed memory stores with structured metadata.
hschneQuery, filter, validate, and transform JSON data using jq.
