firefrost-gaming/skill-seekers-reference
3
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

fix: remove arbitrary limits, fix hardcoded languages, and fix summarizer bugs

Browse Source 
Stage 1 quality improvements from the Arbitrary Limits & Dead Code audit:

Reference file truncation removed:
- codebase_scraper.py: remove code[:500] truncation at 5 locations — reference
  files now contain complete code blocks for copy-paste usability
- unified_skill_builder.py: remove issues[:20], releases[:10], body[:500],
  and code_snippet[:300] caps in reference files — full content preserved

Enhancement summarizer rewrite:
- enhance_skill_local.py: replace arbitrary [:5] code block cap with
  character-budget approach using target_ratio * content_chars
- Fix intro boundary bug: track code block state so intro never ends
  inside a code block, which was desynchronizing the parser
- Remove dead _target_lines variable (assigned but never used)
- Heading chunks now also respect the character budget

Hardcoded language fixes:
- unified_skill_builder.py: test examples use ex["language"] instead of
  always "python" for syntax highlighting
- how_to_guide_builder.py: add language field to HowToGuide dataclass,
  set from workflow at creation, used in AI enhancement prompt

Test fixes:
- test_enhance_skill_local.py: rename test to test_code_blocks_not_arbitrarily_capped,
  fix assertion to count actual blocks (```count // 2), use target_ratio=0.9

Documentation:
- Add Stage 1 plan, implementation summary, review, and corrected docs
- Update CHANGELOG.md with all changes

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
yusyus
2026-02-26 00:30:40 +03:00
parent b81d55fda0
commit b6d4dd8423
10 changed files with 1189 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
src/skill_seekers/cli/codebase_scraper.py
View File

@@ -419,7 +419,7 @@ def extract_markdown_structure(content: str) -> dict[str, Any]:
structure["code_blocks"].append(
{
"language": language,
"code": code[:500], # Truncate long code blocks
"code": code, # Full code - no truncation
"full_length": len(code),
}
)
@@ -486,7 +486,7 @@ def extract_rst_structure(content: str) -> dict[str, Any]:
"code_blocks": [
{
"language": cb.language or "text",
"code": cb.code[:500] if len(cb.code) > 500 else cb.code,
"code": cb.code, # Full code - no truncation
"full_length": len(cb.code),
"quality_score": cb.quality_score,
}
@@ -572,7 +572,7 @@ def extract_rst_structure(content: str) -> dict[str, Any]:
structure["code_blocks"].append(
{
"language": language,
"code": code[:500],
"code": code, # Full code - no truncation
"full_length": len(code),
}
)
@@ -717,7 +717,7 @@ def process_markdown_docs(
for h in parsed_doc.headings
],
"code_blocks": [
{"language": cb.language, "code": cb.code[:500]}
{"language": cb.language, "code": cb.code} # Full code
for cb in parsed_doc.code_blocks
],
"tables": len(parsed_doc.tables),
@@ -743,7 +743,7 @@ def process_markdown_docs(
for h in parsed_doc.headings
],
"code_blocks": [
{"language": cb.language, "code": cb.code[:500]}
{"language": cb.language, "code": cb.code} # Full code
for cb in parsed_doc.code_blocks
],
"tables": len(parsed_doc.tables),

29
src/skill_seekers/cli/enhance_skill_local.py
View File

@@ -306,10 +306,19 @@ class LocalSkillEnhancer:
Summarized content
"""
lines = content.split("\n")
_target_lines = int(len(lines) * target_ratio)
# Priority 1: Keep introduction (first 20%)
intro_lines = int(len(lines) * 0.2)
# Ensure intro doesn't cut inside a code block
in_block = False
safe_end = 0
for i in range(intro_lines):
if lines[i].strip().startswith("```"):
in_block = not in_block
if not in_block:
safe_end = i + 1
intro_lines = safe_end
result_lines = lines[:intro_lines]
# Priority 2: Extract code blocks
@@ -334,13 +343,21 @@ class LocalSkillEnhancer:
elif in_code_block:
current_block.append(line)
# Combine: intro + code blocks + headings
# Combine: intro + code blocks + headings with token budget
result = result_lines.copy()
# Budget is target_ratio of original content length
content_chars = len(content)
max_chars = int(content_chars * target_ratio)
current_chars = sum(len(line) for line in result)
# Add code blocks first (prioritize code examples)
for _idx, block in code_blocks[:5]: # Max 5 code blocks
# Priority 2: Add code blocks first (prioritize code examples) - no arbitrary limit
for _idx, block in code_blocks:
block_chars = sum(len(line) for line in block) + 1 # +1 for blank line
if current_chars + block_chars > max_chars:
break
result.append("") # Add blank line before code block
result.extend(block)
current_chars += block_chars
# Priority 3: Keep headings with first paragraph
i = intro_lines
@@ -350,8 +367,12 @@ class LocalSkillEnhancer:
if line.startswith("#"):
# Found heading - keep it and next 3 lines
chunk = lines[i : min(i + 4, len(lines))]
chunk_chars = sum(len(l) for l in chunk)
if current_chars + chunk_chars > max_chars:
break
result.extend(chunk)
headings_added += 1
current_chars += chunk_chars
i += 4
else:
i += 1

4
src/skill_seekers/cli/how_to_guide_builder.py
View File

@@ -105,6 +105,7 @@ class HowToGuide:
tags: list[str] = field(default_factory=list)
estimated_time: str = "10 minutes"
source_files: list[str] = field(default_factory=list)
language: str = "python" # Source file language
# Optional AI enhancement (basic)
common_pitfalls: list[str] = field(default_factory=list)
@@ -966,6 +967,7 @@ class HowToGuideBuilder:
tags=tags,
estimated_time=metadata.get("estimated_time", "10 minutes"),
source_files=source_files,
language=primary_workflow.get("language", "python"),
)
# Add AI enhancements if enhancer is available
@@ -1015,7 +1017,7 @@ class HowToGuideBuilder:
guide_data = {
"title": guide.title,
"steps": [{"description": step.description, "code": step.code} for step in guide.steps],
"language": "python", # TODO: Detect from code
"language": guide.language,
"prerequisites": guide.prerequisites,
"description": guide.overview,
}

9
src/skill_seekers/cli/unified_skill_builder.py
View File

@@ -907,7 +907,7 @@ This skill combines knowledge from multiple sources:
f.write(f"# GitHub Issues: {repo}\n\n")
f.write(f"{len(github_data['issues'])} recent issues.\n\n")
for issue in github_data["issues"][:20]:
for issue in github_data["issues"]: # All issues, no arbitrary limit
f.write(f"## #{issue['number']}: {issue['title']}\n\n")
f.write(f"**State**: {issue['state']}\n")
if issue.get("labels"):
@@ -920,11 +920,11 @@ This skill combines knowledge from multiple sources:
with open(releases_path, "w", encoding="utf-8") as f:
f.write(f"# Releases: {repo}\n\n")
for release in github_data["releases"][:10]:
for release in github_data["releases"]: # All releases, no arbitrary limit
f.write(f"## {release['tag_name']}: {release.get('name', 'N/A')}\n\n")
f.write(f"**Published**: {release.get('published_at', 'N/A')[:10]}\n\n")
if release.get("body"):
f.write(release["body"][:500])
f.write(release["body"]) # Full release notes
f.write("\n\n")
# Create index for this repo
@@ -1295,7 +1295,8 @@ This skill combines knowledge from multiple sources:
f.write(f"- **Confidence**: {ex.get('confidence', 0):.2f}\n")
f.write(f"- **File**: `{ex.get('file_path', 'N/A')}`\n")
if ex.get("code_snippet"):
f.write(f"\n```python\n{ex['code_snippet'][:300]}\n```\n")
lang = ex.get("language", "text")
f.write(f"\n```{lang}\n{ex['code_snippet']}\n```\n") # Full code, no truncation
f.write("\n")
logger.info(f" ✓ Test examples: {total} total, {high_value} high-value")