feat(motions): implement search_dokument_fulltext, analyze_g0v_by_department, and search_anforanden by Copilot · Pull Request #603 · Hack23/riksdagsmonitor

Copilot · 2026-02-26T21:01:10Z

Original prompt

This section details on the original issue you should resolve

<issue_title>Improve Motions: Add full-text analysis, government response tracking, and debate context</issue_title>
<issue_description>## 📋 Issue Type
Enhancement — Implement missing MCP tools for opposition motions: full-text analysis, government response tracking, and debate context

🎯 Objective

Implement the three TODO MCP tools in scripts/news-types/motions.ts: search_dokument_fulltext (full motion text analysis), analyze_g0v_by_department (government department response tracking), and search_anforanden (debate context and party positioning). These tools will transform motion articles from basic listings into strategic opposition intelligence.

📊 Current State

File: scripts/news-types/motions.ts (403 lines)
Implemented tools: get_motioner ✅ (only tool)
Missing tools: search_dokument_fulltext ❌, analyze_g0v_by_department ❌, search_anforanden ❌
REQUIRED_TOOLS: Only 1 tool (get_motioner) — reduced from original 4 to avoid validation failures
Impact: Motion articles list opposition proposals without explaining what they actually propose, how government responds, or how they're debated

🚀 Desired State

Full-text analysis extracts specific policy alternatives being proposed
Government department response tracking shows which proposals get engagement
Debate context captures party positioning and cross-party support signals
Articles explain opposition strategy (coordinated campaigns, emerging alliances)
Cross-party motion co-sponsorship analysis (which parties collaborate)
"Adoption potential" scoring based on historical patterns
Comparison with government propositions on same topics (alternative framing)

🔧 Implementation Approach

1. Add `search_dokument_fulltext` integration (~80 lines)

// Extract policy alternatives from motion full text
const fullText = await mcpClient.call('search_dokument_fulltext', {
  query: motion.titel,
  limit: 3
});

Parse specific policy alternatives proposed
Identify budget implications and funding sources
Compare with existing government policy
Use scripts/data-transformers/document-analysis.ts for opposition strategy detection

2. Add `analyze_g0v_by_department` integration (~60 lines)

// Track government department engagement with motion topics
const govResponse = await mcpClient.call('analyze_g0v_by_department', {
  dateFrom: motionDate,
  dateTo: currentDate
});

Map motions to responsible government departments
Track whether government has responded to motion topics
Identify areas where opposition is ahead of government agenda

3. Add `search_anforanden` integration (~60 lines)

// Find debate speeches related to this motion topic
const debates = await mcpClient.call('search_anforanden', {
  text: motion.titel,
  rm: currentRiksmote,
  limit: 10
});

Extract party positions from debate speeches
Identify cross-party support signals
Capture government response quotes
Detect coalition fault lines on motion topics

4. Update REQUIRED_TOOLS

Restore full 4-tool list: get_motioner, search_dokument_fulltext, analyze_g0v_by_department, search_anforanden

5. Enhanced article templates

Add "Policy Alternative" section explaining what's proposed
Add "Government Engagement" tracking
Add "Party Positioning" from debates
Add "Cross-Party Support" signals
Update all 14 language templates

🤖 Recommended Agent

code-quality-engineer — TypeScript implementation matching existing MCP integration patterns

✅ Acceptance Criteria

📚 References

File: scripts/news-types/motions.ts:57-60 (TODO comments)
File: scripts/news-types/motions.ts:189-196 (REQUIRED_TOOLS update comment)
MCP Tools: search_dokument_fulltext, analyze_g0v_by_department, search_anforanden
Document analysis: scripts/data-transformers/document-analysis.ts (opposition strategy detection)
Policy analysis: scripts/data-transformers/policy-analysis.ts
Pattern reference: scripts/news-types/weekly-review.ts (full enrichment pipeline example)
</issue_description>

<agent_i...

Fixes Improve Motions: Add full-text analysis, government response tracking, and debate context #601

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.

…search_anforanden in motions.ts Co-authored-by: pethers <1726836+pethers@users.noreply.github.com>

github-actions · 2026-02-27T02:13:38Z

🔍 Lighthouse Performance Audit

Category	Score	Status
Performance	85/100	🟡
Accessibility	95/100	🟢
Best Practices	90/100	🟢
SEO	95/100	🟢

📥 Download full Lighthouse report

Budget Compliance: Performance budgets enforced via budget.json

Copilot

Pull request overview

This PR implements three missing MCP tools in the motions article generator to transform basic motion listings into comprehensive opposition intelligence with full-text analysis, government response tracking, and debate context.

Changes:

Restored REQUIRED_TOOLS from 1 to 4 tools with graceful degradation on failures
Added three MCP enrichment calls: search_dokument_fulltext, analyze_g0v_by_department, and search_anforanden
Extended content generation with "Government Department Engagement" section and enhanced cross-references

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
`scripts/news-types/motions.ts`	Implemented three MCP tools with try-catch error handling, updated REQUIRED_TOOLS to 4, expanded crossReferences sources
`scripts/data-transformers/content-generators.ts`	Added Government Department Engagement section with HTML-escaped output
`scripts/data-transformers/types.ts`	Added `govDeptData` field to ArticleContentData interface
`scripts/types/content.ts`	Added `govEngagement` field to ContentLabelSet interface
`scripts/data-transformers/constants.ts`	Added `govEngagement` label translations for all 14 languages
`tests/news-types/motions.test.ts`	Extended mock client with `request` and `searchSpeeches`, added 8 new tests for tool calls and graceful degradation

Comments suppressed due to low confidence (1)

scripts/news-types/motions.ts:333

The generateSources function in scripts/data-transformers/metadata.ts doesn't have mappings for the three new tools: search_dokument_fulltext, analyze_g0v_by_department, and search_anforanden. While the tools are correctly called and their results used in article generation, the article footer's "Data Sources" section won't include human-readable names for these enrichment sources. Consider adding mappings like: "Full-text Document Analysis", "Government Department Tracking", and "Parliamentary Debate Speeches" to improve transparency about data sources used in the intelligence analysis.

      const sources: string[] = generateSources([
        'get_motioner',
        'search_dokument_fulltext',
        'analyze_g0v_by_department',
        'search_anforanden',
      ]);

Copilot · 2026-02-27T02:17:13Z

tests/news-types/motions.test.ts

+      mockClientInstance.searchSpeeches.mockRejectedValueOnce(new Error('Tool unavailable'));
+      const result = await motionsModule.generateMotions({ languages: ['en'] });
+      expect(result.success).toBe(true);
+    });


Missing test coverage for analyze_g0v_by_department graceful degradation. The implementation calls client.request() twice (for search_dokument_fulltext at line 257 and analyze_g0v_by_department at line 279), but there's no test verifying that failure of the second tool call is handled gracefully. The test at line 160 only covers the first request call failure using mockRejectedValueOnce. Add a test that specifically verifies analyze_g0v_by_department failure doesn't break article generation.

Suggested change

});

});

it('should degrade gracefully when analyze_g0v_by_department fails', async () => {

const originalRequestImpl = mockClientInstance.request.getMockImplementation();

// First call (search_dokument_fulltext) uses the existing implementation

mockClientInstance.request.mockImplementationOnce(

// eslint-disable-next-line @typescript-eslint/no-explicit-any

(...args: any[]) => originalRequestImpl && originalRequestImpl(...args)

);

// Second call (analyze_g0v_by_department) fails

mockClientInstance.request.mockRejectedValueOnce(new Error('Tool unavailable'));

const result = await motionsModule.generateMotions({ languages: ['en'] });

expect(result.success).toBe(true);

});

Added a dedicated test 'should degrade gracefully when analyze_g0v_by_department fails' in commit 1ef7492. It uses mockResolvedValueOnce({}) for the first request call (search_dokument_fulltext) and mockRejectedValueOnce(new Error(...)) for the second (analyze_g0v_by_department), confirming the article still generates successfully when that tool fails.

Copilot · 2026-02-27T02:17:13Z

scripts/news-types/motions.ts

+    try {
+      const debateQuery = motions[0]?.titel || motions[0]?.title || '';
+      if (debateQuery) {
+        const speeches = await client.searchSpeeches({ text: debateQuery, rm: '2025/26', limit: 10 }) as Array<Record<string, unknown>>;


Hardcoded riksmöte value '2025/26' will become outdated when the parliamentary year changes (typically September/October). The Swedish riksmöte follows the pattern YYYY/YY where the year changes each September. Consider extracting a utility function to calculate the current riksmöte dynamically based on the current date, or at minimum document when this value needs to be updated. This same hardcoded value appears in monthly-review.ts and weekly-review.ts, indicating a broader maintainability concern across the codebase.

Fixed in commit 1ef7492. Added getCurrentRiksmote(date) utility that computes the session string dynamically (month >= 8 ? year : year - 1, since getMonth() is 0-based so September = 8). The search_anforanden call now uses getCurrentRiksmote() instead of the hardcoded '2025/26'.

pethers · 2026-02-27T06:39:27Z

@copilot apply changes based on the comments in this thread

…öte calculation Co-authored-by: pethers <1726836+pethers@users.noreply.github.com>

Copilot

Pull request overview

Copilot reviewed 6 out of 6 changed files in this pull request and generated 6 comments.

Copilot · 2026-02-27T07:01:35Z

scripts/news-types/motions.ts

+        });
+      }
+    } catch (err) {
+      console.warn('  ⚠ search_dokument_fulltext unavailable:', (err as Error).message);


In the search_dokument_fulltext catch block, the tool failure is swallowed without recording the tool in mcpCalls. This can cause cross-reference validation to report missing required tools, and the article can still list the tool in sources even though it didn’t produce data. Record the attempted call in mcpCalls with an empty result (and optionally an error marker) when the request fails, or derive sources from the actual mcpCalls list.

Suggested change

console.warn(' ⚠ search_dokument_fulltext unavailable:', (err as Error).message);

console.warn(' ⚠ search_dokument_fulltext unavailable:', (err as Error).message);

// Record attempted tool call with empty result so validation and sources stay consistent

mcpCalls.push({ tool: 'search_dokument_fulltext', result: [] });

Fixed in ca4ea6f. The catch block now pushes { tool: 'search_dokument_fulltext', result: [] } to mcpCalls so validation and source attribution stay consistent even on failure.

Copilot · 2026-02-27T07:01:35Z

scripts/news-types/motions.ts

+      mcpCalls.push({ tool: 'analyze_g0v_by_department', result: govDeptData });
+      console.log(`  🏛 Gov dept analysis: ${govDeptData.length} departments`);
+    } catch (err) {
+      console.warn('  ⚠ analyze_g0v_by_department unavailable:', (err as Error).message);


analyze_g0v_by_department failures currently only log a warning and don’t add an entry to mcpCalls. Since this tool is in REQUIRED_TOOLS, missing the record can fail cross-reference validation even though generation returns success. Add a fallback mcpCalls entry (empty array/result + optional error info) in the catch path so downstream validation and reporting remain consistent.

Suggested change

console.warn(' ⚠ analyze_g0v_by_department unavailable:', (err as Error).message);

const error = err as Error;

console.warn(' ⚠ analyze_g0v_by_department unavailable:', error.message);

// Fallback MCP call record to keep REQUIRED_TOOLS cross-reference consistent

mcpCalls.push({

tool: 'analyze_g0v_by_department',

result: [],

error: {

message: error.message,

name: error.name,

},

} as MCPCallRecord);

Fixed in ca4ea6f. The catch block now pushes { tool: 'analyze_g0v_by_department', result: [] } to keep mcpCalls consistent with REQUIRED_TOOLS.

Copilot · 2026-02-27T07:01:35Z

scripts/news-types/motions.ts

+        }
+      }
+    } catch (err) {
+      console.warn('  ⚠ search_anforanden unavailable:', (err as Error).message);


search_anforanden failures are not recorded in mcpCalls, but the tool remains required and listed as a source. To keep validation and source attribution accurate during graceful degradation, add an mcpCalls entry in the catch path (e.g., empty speeches array and/or captured error message).

Suggested change

console.warn(' ⚠ search_anforanden unavailable:', (err as Error).message);

console.warn(' ⚠ search_anforanden unavailable:', (err as Error).message);

mcpCalls.push({ tool: 'search_anforanden', result: [] });

Fixed in ca4ea6f. The catch block now pushes { tool: 'search_anforanden', result: [] } to mcpCalls.

Copilot · 2026-02-27T07:01:36Z

scripts/news-types/motions.ts

+        ftDocs.forEach((ftDoc, i) => {
+          const m = motions[i] as Record<string, unknown>;
+          if (m && !m['fullText']) {
+            m['fullText'] = ftDoc.fullText || ftDoc.summary || '';
+          }
+        });


search_dokument_fulltext is queried once using only the first motion’s title, but the results are then assigned to motions by index and written into each motion’s fullText field. This can easily attach unrelated text to other motions and pollute downstream summaries/policy analysis. Consider keeping these documents as a separate related-docs collection, or do per-motion lookup / match on dok_id (when available) rather than index-based assignment.

Suggested change

ftDocs.forEach((ftDoc, i) => {

const m = motions[i] as Record<string, unknown>;

if (m && !m['fullText']) {

m['fullText'] = ftDoc.fullText || ftDoc.summary || '';

}

});

const primaryMotion = motions[0] as Record<string, unknown> | undefined;

if (primaryMotion && ftDocs.length > 0) {

// Attach the most relevant full text only to the motion that was used for the query

if (!primaryMotion['fullText']) {

const bestDoc = ftDocs[0] as Record<string, unknown>;

primaryMotion['fullText'] = (bestDoc['fullText'] as string) || (bestDoc['summary'] as string) || '';

}

// Optionally expose all related documents for downstream "Policy Alternative" rendering

if (!primaryMotion['policyAlternativeDocs']) {

primaryMotion['policyAlternativeDocs'] = ftDocs;

}

}

Fixed in ca4ea6f. Replaced the forEach index loop with a targeted assignment: only motions[0] (the motion used for the query) receives fullText from ftDocs[0]. All related docs are exposed on primaryMotion['policyAlternativeDocs'] when multiple results are returned.

Copilot · 2026-02-27T07:01:36Z

scripts/data-transformers/content-generators.ts

+      if (deptName) {
+        content += deptCount


deptCount is checked using a truthy test, so a legitimate value of 0 will be treated as “missing” and won’t render in parentheses. Prefer checking for null/undefined (e.g., deptCount !== null && deptCount !== undefined) so 0 still displays correctly.

Suggested change

if (deptName) {

content += deptCount

const hasDeptCount = deptCount !== null && deptCount !== undefined;

if (deptName) {

content += hasDeptCount

Fixed in ca4ea6f. Changed to const hasDeptCount = deptCount !== null && deptCount !== undefined so a value of 0 renders correctly in parentheses.

Copilot · 2026-02-27T07:01:36Z

tests/news-types/motions.test.ts

+    it('should degrade gracefully when search_dokument_fulltext fails', async () => {
+      mockClientInstance.request.mockRejectedValueOnce(new Error('Tool unavailable'));
+      const result = await motionsModule.generateMotions({ languages: ['en'] });
+      expect(result.success).toBe(true);
+    });
+
+    it('should degrade gracefully when analyze_g0v_by_department fails', async () => {
+      // First request call (search_dokument_fulltext) succeeds; second (analyze_g0v_by_department) fails
+      mockClientInstance.request
+        .mockResolvedValueOnce({})
+        .mockRejectedValueOnce(new Error('Tool unavailable'));
+      const result = await motionsModule.generateMotions({ languages: ['en'] });
+      expect(result.success).toBe(true);
+    });


Current tests assert generation succeeds when enrichment tools fail, but they don’t verify that the failed tool is still recorded in mcpCalls (which downstream cross-reference validation relies on). Add assertions in the graceful-degradation tests to ensure mcpCalls includes the failed tool name with an empty result/error marker.

Fixed in ca4ea6f. All three graceful-degradation tests now assert result.mcpCalls?.map(c => c.tool) contains the failed tool name, verifying the empty-result fallback entry is recorded.

pethers · 2026-02-27T07:13:06Z

@copilot apply changes based on the comments in this thread

…ignment, fix deptCount null check Co-authored-by: pethers <1726836+pethers@users.noreply.github.com>

Signed-off-by: James Pether Sörling <pethers@users.noreply.github.com>

github-actions · 2026-02-27T08:28:03Z

🔍 Lighthouse Performance Audit

Category	Score	Status
Performance	85/100	🟡
Accessibility	95/100	🟢
Best Practices	90/100	🟢
SEO	95/100	🟢

📥 Download full Lighthouse report

Budget Compliance: Performance budgets enforced via budget.json

Copilot

Pull request overview

Copilot reviewed 6 out of 6 changed files in this pull request and generated no new comments.

Comments suppressed due to low confidence (1)

scripts/data-transformers/types.ts:101

The ArticleContentData interface now has two fields for essentially the same data from analyze_g0v_by_department:

govDeptData (line 97): An array of department records, used by motions.ts
departmentAnalysis (line 101): A single object containing a 'departments' property, used by propositions.ts

This creates an inconsistent API where different news types use different field names and shapes for the same MCP tool response. Consider standardizing on one field name and shape across all news types. For example, consolidate to use govDeptData as the direct array (since it's simpler), and update propositions.ts to extract the array from the response before passing it to generateArticleContent.

  /** Government department analysis from analyze_g0v_by_department */
  govDeptData?: Record<string, unknown>[];
  /** Full-text search results for policy substance extraction */
  fullTextResults?: unknown[];
  /** Government department analysis from analyze_g0v_by_department */
  departmentAnalysis?: Record<string, unknown>;

Copilot AI assigned Copilot and pethers Feb 26, 2026

Copilot started work on behalf of pethers February 26, 2026 21:01 View session

Copilot AI mentioned this pull request Feb 26, 2026

Create 8 issues to improve/extend all article types with MCP tool integrations #594

Draft

feat: implement search_dokument_fulltext, analyze_g0v_by_department, …

bb32fe5

…search_anforanden in motions.ts Co-authored-by: pethers <1726836+pethers@users.noreply.github.com>

Copilot AI changed the title ~~[WIP] Improve motions with full-text analysis and response tracking~~ feat(motions): implement search_dokument_fulltext, analyze_g0v_by_department, and search_anforanden Feb 26, 2026

Copilot finished work on behalf of pethers February 26, 2026 21:20

pethers requested a review from Copilot February 27, 2026 02:12

github-actions bot added testing Test coverage refactor Code refactoring size-m Medium change (50-250 lines) labels Feb 27, 2026

Copilot started reviewing on behalf of pethers February 27, 2026 02:13 View session

Copilot AI reviewed Feb 27, 2026

View reviewed changes

Copilot started work on behalf of pethers February 27, 2026 06:39 View session

fix: add analyze_g0v_by_department degradation test and dynamic riksm…

1ef7492

…öte calculation Co-authored-by: pethers <1726836+pethers@users.noreply.github.com>

Copilot finished work on behalf of pethers February 27, 2026 06:44

pethers requested a review from Copilot February 27, 2026 06:57

Copilot started reviewing on behalf of pethers February 27, 2026 06:57 View session

Copilot AI reviewed Feb 27, 2026

View reviewed changes

Copilot started work on behalf of pethers February 27, 2026 07:26 View session

fix: record failed MCP tool calls in mcpCalls, fix fullText index ass…

ca4ea6f

…ignment, fix deptCount null check Co-authored-by: pethers <1726836+pethers@users.noreply.github.com>

Copilot finished work on behalf of pethers February 27, 2026 07:31

Merge branch 'main' into copilot/add-full-text-analysis-tools

81cbefa

Signed-off-by: James Pether Sörling <pethers@users.noreply.github.com>

pethers requested a review from Copilot February 27, 2026 08:26

Copilot started reviewing on behalf of pethers February 27, 2026 08:27 View session

Copilot AI reviewed Feb 27, 2026

View reviewed changes

pethers marked this pull request as ready for review February 27, 2026 10:15

pethers approved these changes Feb 27, 2026

View reviewed changes

pethers merged commit ab4eb79 into main Feb 27, 2026
19 checks passed

pethers deleted the copilot/add-full-text-analysis-tools branch February 27, 2026 10:15

-    });
+    });
+    it('should degrade gracefully when analyze_g0v_by_department fails', async () => {
+      const originalRequestImpl = mockClientInstance.request.getMockImplementation();
+      // First call (search_dokument_fulltext) uses the existing implementation
+      mockClientInstance.request.mockImplementationOnce(
+        // eslint-disable-next-line @typescript-eslint/no-explicit-any
+        (...args: any[]) => originalRequestImpl && originalRequestImpl(...args)
+      );
+      // Second call (analyze_g0v_by_department) fails
+      mockClientInstance.request.mockRejectedValueOnce(new Error('Tool unavailable'));
+      const result = await motionsModule.generateMotions({ languages: ['en'] });
+      expect(result.success).toBe(true);
+    });

-      console.warn('  ⚠ analyze_g0v_by_department unavailable:', (err as Error).message);
+      const error = err as Error;
+      console.warn('  ⚠ analyze_g0v_by_department unavailable:', error.message);
+      // Fallback MCP call record to keep REQUIRED_TOOLS cross-reference consistent
+      mcpCalls.push({
+        tool: 'analyze_g0v_by_department',
+        result: [],
+        error: {
+          message: error.message,
+          name: error.name,
+        },
+      } as MCPCallRecord);

	console.warn(' ⚠ search_anforanden unavailable:', (err as Error).message);
	console.warn(' ⚠ search_anforanden unavailable:', (err as Error).message);
	mcpCalls.push({ tool: 'search_anforanden', result: [] });

-        ftDocs.forEach((ftDoc, i) => {
-          const m = motions[i] as Record<string, unknown>;
-          if (m && !m['fullText']) {
-            m['fullText'] = ftDoc.fullText || ftDoc.summary || '';
-          }
-        });
+        const primaryMotion = motions[0] as Record<string, unknown> | undefined;
+        if (primaryMotion && ftDocs.length > 0) {
+          // Attach the most relevant full text only to the motion that was used for the query
+          if (!primaryMotion['fullText']) {
+            const bestDoc = ftDocs[0] as Record<string, unknown>;
+            primaryMotion['fullText'] = (bestDoc['fullText'] as string) || (bestDoc['summary'] as string) || '';
+          }
+          // Optionally expose all related documents for downstream "Policy Alternative" rendering
+          if (!primaryMotion['policyAlternativeDocs']) {
+            primaryMotion['policyAlternativeDocs'] = ftDocs;
+          }
+        }

Conversation

Copilot AI commented Feb 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🎯 Objective

📊 Current State

🚀 Desired State

🔧 Implementation Approach

1. Add search_dokument_fulltext integration (~80 lines)

2. Add analyze_g0v_by_department integration (~60 lines)

3. Add search_anforanden integration (~60 lines)

4. Update REQUIRED_TOOLS

5. Enhanced article templates

🤖 Recommended Agent

✅ Acceptance Criteria

📚 References

Uh oh!

github-actions bot commented Feb 27, 2026

🔍 Lighthouse Performance Audit

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

pethers commented Feb 27, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 27, 2026

Choose a reason for hiding this comment

Uh oh!

pethers commented Feb 27, 2026

Uh oh!

Copilot AI commented Feb 26, 2026 •

edited

Loading

1. Add `search_dokument_fulltext` integration (~80 lines)

2. Add `analyze_g0v_by_department` integration (~60 lines)

3. Add `search_anforanden` integration (~60 lines)