Tax questions get vague answers because the underlying rules are dense and change by year and jurisdiction. This assistant grounds every reply in the actual code.
sonnet1 weekQdrantOpenAILangChainPostgreSQL
Claude
62ROI
81Scale
$98073Saved
ROI for
README.md
Why this subagent
Tax questions get vague answers because the underlying rules are dense and change by year and jurisdiction. This assistant grounds every reply in the actual code.
Provisions are embedded into Qdrant with their section, year, and jurisdiction attached. A question pulls the relevant passages, and the answer quotes exact section numbers rather than paraphrasing from memory. When a provision looks like it may have been amended, it flags that instead of guessing.
How it runs
Used at step 01 to kick off the pipeline.
Write
Used at step 01 to kick off the pipeline.
WebFetch
Used at step 01 to kick off the pipeline.
WebSearch
Used at step 01 to kick off the pipeline.
Store passages in Qdrant with metadata for the code section, year, and jurisdiction.
pending
Retrieve the matching provisions for a question and answer strictly from their text.
pending
Quote the exact section numbers and flag when a provision may have been amended.
pending
Sample output
json
// Sample output
// (generated when the pipeline finishes)
Answer tax questions only from retrieved code passages, quoting section numbers and never inventing a provision.
Unlock the rest
The full agent definition, install snippet, and starter task are gated for community members.
Members get the full `.md` agent file, the npm / pnpm install one-liners, a starter prompt that we've tuned against real runs, and the open-source repo when this automation ships there. One email, magic link, done.