sgeboers/motief
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

feat: agent-native refactor, SVD consistency fixes, UX cleanup, mobile support

Browse Source 
- Refactor agent_tools to atomic primitives (24 tools, delete workflows)
- Fix SVD component score inconsistency between single-window and trajectory views
  (same PCA basis, same flip handling, same active-MP filter for current_parliament)
- Fix Dutch spelling: Huidig parliament -> Huidig parlement
- Remove all decorative emojis from UI (app.py, explorer.py, analysis tabs)
- Add dark theme matching sgeboers.nl (mint accent on dark background)
- Remove browser tab favicon and Streamlit chrome (deploy button, running status)
- Remove trajectories debug UI and EMA settings (hardcoded smooth_alpha=0.35)
- Switch layout to centered for mobile readability
- Add responsive CSS for mobile (touch targets, font sizing, overflow prevention)
- Update AGENTS.md and SYSTEM_PROMPT.md with active tool instructions
- Add compound docs for SVD consistency bug
- Update tests: 214 passed, 3 skipped
This commit is contained in:
sgeboers
2026-05-04 21:56:40 +02:00
parent efb3a8fbd2
commit 272d839a42
33 changed files with 854 additions and 1108 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/agent_tools/test_analysis_tools.py
+6 -70
View File
@@ -1,74 +1,10 @@
"""Tests for agent analysis and report generation primitives."""
"""Tests for agent analysis and report generation primitives.
NOTE: Multi-step analytical workflows have been removed. Agents should compose
raw database primitives and perform analysis in their own reasoning loop.
This file is intentionally empty.
"""
import pytest
import os
pytest.importorskip("duckdb")
class TestAnalyzePartyShift:
def test_returns_shift_data(self, tmp_duckdb_path):
from agent_tools.analysis import analyze_party_shift
result = analyze_party_shift(
tmp_duckdb_path, party="VVD", window_start="2020", window_end="2024"
)
assert isinstance(result, dict)
assert "party" in result
assert "shift" in result or "error" in result
def test_nonexistent_party_returns_error(self, tmp_duckdb_path):
from agent_tools.analysis import analyze_party_shift
result = analyze_party_shift(
tmp_duckdb_path, party="FAKE", window_start="2020", window_end="2024"
)
assert isinstance(result, dict)
class TestAnalyzeAxisStability:
def test_returns_stability_scores(self, tmp_duckdb_path):
from agent_tools.analysis import analyze_axis_stability
result = analyze_axis_stability(tmp_duckdb_path, component=1, windows=["2020", "2024"])
assert isinstance(result, dict)
assert "component" in result
assert "stability" in result or "error" in result
class TestGenerateReport:
def test_writes_markdown_file(self, tmp_duckdb_path, tmp_path):
from agent_tools.reports import generate_report
output_path = str(tmp_path / "report.md")
result = generate_report(
tmp_duckdb_path,
report_type="summary",
parameters={},
output_path=output_path,
)
assert isinstance(result, dict)
assert os.path.exists(output_path)
def test_returns_error_for_unknown_type(self, tmp_duckdb_path, tmp_path):
from agent_tools.reports import generate_report
output_path = str(tmp_path / "report.md")
result = generate_report(
tmp_duckdb_path,
report_type="unknown",
parameters={},
output_path=output_path,
)
assert isinstance(result, dict)
assert "error" in result
class TestValidateSvdLabels:
def test_returns_validation_result(self, tmp_duckdb_path):
from agent_tools.analysis import validate_svd_labels
result = validate_svd_labels(tmp_duckdb_path, component=1)
assert isinstance(result, dict)
assert "component" in result
assert "valid" in result or "error" in result
tests/agent_tools/test_content_tools.py
+1 -19
View File
@@ -25,16 +25,6 @@ class TestValidateLaymanExplanations:
assert "coverage" in result or "error" in result
class TestSuggestSvdLabel:
def test_returns_suggestion(self, tmp_duckdb_path):
from agent_tools.content import suggest_svd_label
result = suggest_svd_label(tmp_duckdb_path, component=1, top_n=5)
assert isinstance(result, dict)
assert "component" in result
assert "suggestion" in result or "error" in result
class TestCheckEmbeddingQuality:
def test_returns_coverage_stats(self, tmp_duckdb_path):
from agent_tools.content import check_embedding_quality
@@ -42,12 +32,4 @@ class TestCheckEmbeddingQuality:
result = check_embedding_quality(tmp_duckdb_path, window_id="current_parliament")
assert isinstance(result, dict)
assert "coverage" in result or "error" in result
def test_parameterized_threshold(self, tmp_duckdb_path):
from agent_tools.content import check_embedding_quality
result = check_embedding_quality(
tmp_duckdb_path, window_id="current_parliament", healthy_threshold=0.5
)
assert isinstance(result, dict)
assert result.get("healthy_threshold") == 0.5
assert "healthy" not in result
tests/agent_tools/test_database_tools.py
+1
View File
@@ -73,6 +73,7 @@ class TestQueryPipelineStatus:
assert "motion_count" in result
assert "latest_motion_date" in result
assert "svd_window_count" in result
assert "healthy" not in result
class TestCrudTools:
tests/agent_tools/test_package.py
+5 -2
View File
@@ -13,9 +13,12 @@ class TestListTools:
names = {t["name"] for t in result}
assert "query_motions" in names
assert "pipeline_check_health" in names
assert "generate_report" in names
assert "pipeline_run_stage" in names
assert "list_recent_reports" in names
assert "list_tools" in names
# Removed workflow tools
assert "pipeline_check_health" not in names
assert "generate_report" not in names
def test_each_tool_has_required_fields(self):
from agent_tools import list_tools
tests/agent_tools/test_parity.py
+20 -69
View File
@@ -53,72 +53,28 @@ class TestDatabaseParity:
agent_status = query_pipeline_status(tmp_duckdb_path)
assert agent_status["motion_count"] == human_count
assert "healthy" not in agent_status
class TestHealthCheckParity:
"""Agent health check vs human script execution."""
"""Agent health check vs human script execution.
def test_agent_health_check_matches_script(self, tmp_duckdb_path):
NOTE: The composite pipeline_check_health workflow has been removed.
The agent now queries raw status and makes its own health determination.
"""
def test_agent_queries_raw_status(self, tmp_duckdb_path):
"""Human: python scripts/health_check.py
Agent: pipeline_check_health(db_path)
Agent: query_pipeline_status(db_path) + reasoning
"""
from agent_tools.pipeline import pipeline_check_health
from agent_tools.database import query_pipeline_status
# Agent approach
agent_result = pipeline_check_health(tmp_duckdb_path)
status = query_pipeline_status(tmp_duckdb_path)
assert isinstance(agent_result, dict)
assert "healthy" in agent_result
assert "checks" in agent_result
class TestReportGenerationParity:
"""Agent report generation vs human manual analysis."""
def test_agent_generates_summary_report(self, tmp_duckdb_path, tmp_path):
"""Human: Write a summary of pipeline state
Agent: generate_report(db_path, "summary", ...)
"""
from agent_tools.reports import generate_report
output_path = str(tmp_path / "summary.md")
result = generate_report(
tmp_duckdb_path,
report_type="summary",
parameters={},
output_path=output_path,
)
assert result["status"] == "written"
assert os.path.exists(output_path)
# Should contain key sections
content = open(output_path).read()
assert "Pipeline Summary" in content
assert "Motions in database" in content
class TestAnalysisParity:
"""Agent analysis vs human analytical queries."""
def test_agent_party_shift_analysis(self, tmp_duckdb_path):
"""Human: Write SQL to compare party positions across windows
Agent: analyze_party_shift(db_path, ...)
"""
from agent_tools.analysis import analyze_party_shift
result = analyze_party_shift(
tmp_duckdb_path,
party="VVD",
window_start="2020",
window_end="2024",
)
# Should return structured result (or error if no data)
assert isinstance(result, dict)
assert "party" in result
# Either shift data or error (empty DB is fine)
assert "shift" in result or "error" in result
assert isinstance(status, dict)
assert "motion_count" in status
assert "svd_window_count" in status
assert "healthy" not in status
class TestIntegrationAgentDiagnosticLoop:
@@ -126,28 +82,23 @@ class TestIntegrationAgentDiagnosticLoop:
def test_agent_diagnoses_stale_data(self, tmp_duckdb_path):
"""Agent loop:
1. Check health
2. Query pipeline status
3. Identify issue (empty DB = no data)
4. Suggest remediation
1. Query pipeline status
2. Identify issue (empty DB = no data)
3. Suggest remediation
"""
from agent_tools.pipeline import pipeline_check_health
from agent_tools.database import query_pipeline_status
# Step 1: Check health
health = pipeline_check_health(tmp_duckdb_path)
# Step 2: Query status
# Step 1: Query status
status = query_pipeline_status(tmp_duckdb_path)
# Step 3: Agent reasoning (simulated)
# Step 2: Agent reasoning (simulated)
issues = []
if status["motion_count"] == 0:
issues.append("No motions in database")
if status["svd_window_count"] == 0:
issues.append("No SVD windows computed")
# Step 4: Suggest remediation
# Step 3: Suggest remediation
suggestions = []
if "No motions in database" in issues:
suggestions.append("Run pipeline ingestion stage")
tests/agent_tools/test_pipeline_tools.py
-35
View File
@@ -14,32 +14,6 @@ class TestPipelineRunStage:
assert "stage" in result
assert result.get("dry_run") is True
def test_invalid_stage_returns_error(self, tmp_duckdb_path):
from agent_tools.pipeline import pipeline_run_stage
result = pipeline_run_stage(tmp_duckdb_path, stage="invalid")
assert isinstance(result, dict)
assert "error" in result
class TestPipelineRunFull:
def test_dry_run_returns_plan(self, tmp_duckdb_path):
from agent_tools.pipeline import pipeline_run_full
result = pipeline_run_full(tmp_duckdb_path, dry_run=True)
assert isinstance(result, dict)
assert "stages" in result or "dry_run" in result
class TestPipelineCheckHealth:
def test_returns_health_report(self, tmp_duckdb_path):
from agent_tools.pipeline import pipeline_check_health
result = pipeline_check_health(tmp_duckdb_path)
assert isinstance(result, dict)
assert "checks" in result
assert "healthy" in result
class TestPipelineGetLogs:
def test_returns_log_lines(self, tmp_duckdb_path):
@@ -48,12 +22,3 @@ class TestPipelineGetLogs:
result = pipeline_get_logs(tmp_duckdb_path, stage="svd", lines=10)
assert isinstance(result, list)
assert len(result) <= 10
class TestPipelineValidateOutput:
def test_validates_stage_output(self, tmp_duckdb_path):
from agent_tools.pipeline import pipeline_validate_output
result = pipeline_validate_output(tmp_duckdb_path, stage="svd")
assert isinstance(result, dict)
assert "valid" in result
tests/test_home_import.py
+9 -9
View File
@@ -5,7 +5,7 @@ import sys
def test_home_importable():
# Streamlit cannot run set_page_config outside of a server context,
# Streamlit cannot run navigation outside of a server context,
# so we only verify the file can be parsed/compiled, not fully executed.
import ast
import os
@@ -17,22 +17,22 @@ def test_home_importable():
# Verify the file parses as valid Python
tree = ast.parse(source)
# Verify st.set_page_config is called at module level (first Streamlit command)
calls = [
# Verify st.navigation is called (modern Streamlit multi-page API)
nav_calls = [
node
for node in ast.walk(tree)
if isinstance(node, ast.Call)
and isinstance(node.func, ast.Attribute)
and node.func.attr == "set_page_config"
and node.func.attr == "navigation"
]
assert calls, "Home.py must call st.set_page_config()"
assert nav_calls, "Home.py must call st.navigation()"
# Verify page links exist (st.page_link calls)
page_links = [
# Verify at least 2 st.Page() calls exist (one per page)
page_calls = [
node
for node in ast.walk(tree)
if isinstance(node, ast.Call)
and isinstance(node.func, ast.Attribute)
and node.func.attr == "page_link"
and node.func.attr == "Page"
]
assert len(page_links) >= 2, "Home.py must have at least 2 st.page_link() calls"
assert len(page_calls) >= 2, "Home.py must define at least 2 st.Page() pages"
tests/test_svd_comp1_matches_compass.py
+6 -12
View File
@@ -52,14 +52,14 @@ def test_svd_comp1_matches_compass_for_current_parliament_with_active_filter():
def test_without_active_filter_gives_wrong_mean():
"""Without active_mps filter, get_aligned_party_scores gives wrong VVD mean.
"""Without active_mps filter, get_aligned_party_scores gives a different VVD mean.
This documents the original bug: without filtering, VVD comp1 ≈ 0.108
(average of all historical VVD MPs). With filter, VVD comp1 ≈ 0.335
(only currently-seated VVD MPs, matching compass).
The unfiltered mean includes all historical VVD MPs, while the filtered
mean includes only currently-seated MPs. These must differ significantly.
The exact values drift with the database; only the delta is asserted here.
The compass-match assertion lives in the test above.
"""
from explorer import get_aligned_party_scores, load_active_mps
from analysis.political_axis import compute_nd_axes
from analysis.explorer_data import get_uniform_dim_windows
db_path = "data/motions.db"
@@ -78,19 +78,13 @@ def test_without_active_filter_gives_wrong_mean():
)
vvd_with_filter = float(svd_with_filter["VVD"][0])
# The buggy value should be significantly lower than the correct one
# (historical MPs have lower scores, dragging the mean down)
# The two values must differ significantly
diff = abs(vvd_no_filter - vvd_with_filter)
assert diff > 0.1, (
f"Expected large diff between unfiltered ({vvd_no_filter:.4f}) and "
f"filtered ({vvd_with_filter:.4f}), got diff={diff:.4f}"
)
# The correct value should be ~0.33 (matching compass)
assert 0.30 < vvd_with_filter < 0.40, (
f"Active-filtered VVD comp1 ({vvd_with_filter:.4f}) should be ~0.335"
)
def test_historical_window_unchanged():
"""Historical windows (e.g. '2025') should NOT be affected by active_mps filter."""