01
Document analysis · Apache-2.0 code
Bias Detection — DOXA
Examine framing and attribution across a document collection. Compare declared analytical criteria and review the passages behind each interpretation.
A local research workspace. Saved results can be inspected without new inference; new analysis needs a configured local model or your own provider. Model readings require human review.
Local setup and examples
Python 3.11+ and Node 22+, or the documented Docker setup. Import sources, inspect recorded results and export your work. Provider calls are optional and separate.
Setup, persistence and deployment ↗
Recorded examples and their limits ↗
02
Policy data · Apache-2.0 code
Knowledge Pipeline
Turn exact source passages and authored annotations into a graph, a bounded context capsule and an evidence-review sheet.
The Python reference needs no dependencies, credentials or model calls. It preserves your declared relationships and qualifications; it does not certify an interpretation or automatically extract meaning.
Run the source-to-context example
Python 3.11+. From a clean clone:
git clone https://github.com/sargonxg/TACITUS-Knowledge-Pipeline-open.git
cd TACITUS-Knowledge-Pipeline-open/reference-pipeline
PYTHONPATH=src python -m tacitus_pipeline build \
--input examples/assessment/collection.json \
--ontology ontology/policy.json \
--layer layers/assessment-evidence.json \
--question "Map assessment evidence and review roles." \
--output build/assessment
Open build/assessment/brief.md, graph.json and capsule.json. The example uses historical EU source editions; check current applicability separately.
Full setup, tests and output guide ↗
03
Scenario analysis · Redevelopment planned
Windtunnel — Scenario analysis
A private historical prototype exploring scenario analysis for policy and political questions.
Redevelopment planned. The historical prototype is private and is not available as a public app, code release or runnable download.
Discuss scenario-analysis needs ↗