claude-ctx 0.5.0rc1__tar.gz

claude_ctx-0.5.0rc1/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Steve Solun
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

claude_ctx-0.5.0rc1/PKG-INFO

@@ -0,0 +1,68 @@
+Metadata-Version: 2.4
+Name: claude-ctx
+Version: 0.5.0rc1
+Summary: Skill and agent recommendation system for Claude Code — knowledge graph, wiki, and intake quality gates
+Author: Steve Solun
+License: MIT
+Project-URL: Homepage, https://stevesolun.github.io/ctx/
+Project-URL: Repository, https://github.com/stevesolun/ctx
+Project-URL: Documentation, https://stevesolun.github.io/ctx/
+Project-URL: Changelog, https://github.com/stevesolun/ctx/blob/main/CHANGELOG.md
+Requires-Python: >=3.11
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: networkx<4,>=3
+Requires-Dist: numpy<3,>=1.24
+Requires-Dist: plotly<7,>=5
+Requires-Dist: pyyaml<7,>=6
+Provides-Extra: dev
+Requires-Dist: pytest>=8; extra == "dev"
+Requires-Dist: pytest-cov>=5; extra == "dev"
+Requires-Dist: mypy>=1.8; extra == "dev"
+Requires-Dist: ruff>=0.4; extra == "dev"
+Provides-Extra: embeddings
+Requires-Dist: sentence-transformers<4,>=2; extra == "embeddings"
+Requires-Dist: torch<3,>=2; extra == "embeddings"
+Dynamic: license-file
+
+# ctx — Skill & Agent Recommendation for Claude Code
+
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)
+[![Python 3.11+](https://img.shields.io/badge/Python-3.11+-green.svg)](https://python.org)
+[![PyPI](https://img.shields.io/pypi/v/claude-ctx.svg)](https://pypi.org/project/claude-ctx/)
+[![Tests](https://img.shields.io/badge/Tests-1319_passing-brightgreen.svg)](#)
+[![Docs](https://img.shields.io/badge/docs-MkDocs_Material-blue.svg)](https://stevesolun.github.io/ctx/)
+
+Watches what you develop, walks a knowledge graph of **1,768 skills and 443 agents**, and recommends the right ones on the fly — you decide what to load and unload. Powered by a Karpathy LLM wiki with persistent memory that gets smarter every session.
+
+## Why it exists
+
+- **Discovery** — with 1,700+ skills and 400+ agents, you can't possibly know which exist or which apply to your current repo.
+- **Context budget** — loading everything wastes tokens and degrades quality. You need the right 10–15 per session.
+- **Skill rot** — skills you installed months ago and never used are cluttering context. Stale ones should be flagged automatically.
+
+## Install
+
+```bash
+pip install claude-ctx
+```
+
+That's it. Optional extras: `pip install "claude-ctx[embeddings]"` for the semantic backend, `pip install "claude-ctx[dev]"` for the test toolchain.
+
+## Use
+
+After install, the `ctx` hooks integrate automatically with Claude Code's `PostToolUse` + `Stop` events. Typical flow:
+
+```bash
+ctx-scan-repo              # scan current repo, surface recommended skills/agents
+ctx-skill-quality list     # see the four-signal quality score for every skill
+ctx-skill-quality explain python-patterns   # drill into a single skill
+ctx-skill-health dashboard # structural health + drift detection
+ctx-toolbox run review     # run a council of skills + agents on the current diff
+```
+
+Full docs, architecture, and every module: **<https://stevesolun.github.io/ctx/>**
+
+## License
+
+MIT — see [LICENSE](LICENSE).

claude_ctx-0.5.0rc1/README.md

@@ -0,0 +1,41 @@
+# ctx — Skill & Agent Recommendation for Claude Code
+
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)
+[![Python 3.11+](https://img.shields.io/badge/Python-3.11+-green.svg)](https://python.org)
+[![PyPI](https://img.shields.io/pypi/v/claude-ctx.svg)](https://pypi.org/project/claude-ctx/)
+[![Tests](https://img.shields.io/badge/Tests-1319_passing-brightgreen.svg)](#)
+[![Docs](https://img.shields.io/badge/docs-MkDocs_Material-blue.svg)](https://stevesolun.github.io/ctx/)
+
+Watches what you develop, walks a knowledge graph of **1,768 skills and 443 agents**, and recommends the right ones on the fly — you decide what to load and unload. Powered by a Karpathy LLM wiki with persistent memory that gets smarter every session.
+
+## Why it exists
+
+- **Discovery** — with 1,700+ skills and 400+ agents, you can't possibly know which exist or which apply to your current repo.
+- **Context budget** — loading everything wastes tokens and degrades quality. You need the right 10–15 per session.
+- **Skill rot** — skills you installed months ago and never used are cluttering context. Stale ones should be flagged automatically.
+
+## Install
+
+```bash
+pip install claude-ctx
+```
+
+That's it. Optional extras: `pip install "claude-ctx[embeddings]"` for the semantic backend, `pip install "claude-ctx[dev]"` for the test toolchain.
+
+## Use
+
+After install, the `ctx` hooks integrate automatically with Claude Code's `PostToolUse` + `Stop` events. Typical flow:
+
+```bash
+ctx-scan-repo              # scan current repo, surface recommended skills/agents
+ctx-skill-quality list     # see the four-signal quality score for every skill
+ctx-skill-quality explain python-patterns   # drill into a single skill
+ctx-skill-health dashboard # structural health + drift detection
+ctx-toolbox run review     # run a council of skills + agents on the current diff
+```
+
+Full docs, architecture, and every module: **<https://stevesolun.github.io/ctx/>**
+
+## License
+
+MIT — see [LICENSE](LICENSE).

claude_ctx-0.5.0rc1/pyproject.toml

@@ -0,0 +1,59 @@
+[build-system]
+requires = ["setuptools>=42"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "claude-ctx"
+version = "0.5.0-rc1"
+description = "Skill and agent recommendation system for Claude Code — knowledge graph, wiki, and intake quality gates"
+authors = [{ name = "Steve Solun" }]
+license = { text = "MIT" }
+readme = "README.md"
+requires-python = ">=3.11"
+
+dependencies = [
+    "networkx>=3,<4",
+    "numpy>=1.24,<3",
+    "plotly>=5,<7",
+    "pyyaml>=6,<7",
+]
+
+[project.urls]
+Homepage = "https://stevesolun.github.io/ctx/"
+Repository = "https://github.com/stevesolun/ctx"
+Documentation = "https://stevesolun.github.io/ctx/"
+Changelog = "https://github.com/stevesolun/ctx/blob/main/CHANGELOG.md"
+
+[project.scripts]
+ctx-scan-repo = "scan_repo:main"
+ctx-skill-quality = "skill_quality:main"
+ctx-skill-health = "skill_health:main"
+ctx-toolbox = "toolbox:main"
+
+[project.optional-dependencies]
+dev = [
+    "pytest>=8",
+    "pytest-cov>=5",
+    "mypy>=1.8",
+    "ruff>=0.4",
+]
+embeddings = [
+    "sentence-transformers>=2,<4",
+    "torch>=2,<3",
+]
+
+[tool.setuptools]
+package-dir = {"" = "src"}
+
+[tool.setuptools.packages.find]
+where = ["src"]
+
+[tool.pytest.ini_options]
+testpaths = ["src/tests"]
+markers = [
+    "integration: marks tests that require external models or network (deselect with -m 'not integration')",
+]
+
+[tool.ruff]
+line-length = 100
+target-version = "py311"

claude_ctx-0.5.0rc1/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

claude_ctx-0.5.0rc1/src/claude_ctx.egg-info/PKG-INFO

@@ -0,0 +1,68 @@
+Metadata-Version: 2.4
+Name: claude-ctx
+Version: 0.5.0rc1
+Summary: Skill and agent recommendation system for Claude Code — knowledge graph, wiki, and intake quality gates
+Author: Steve Solun
+License: MIT
+Project-URL: Homepage, https://stevesolun.github.io/ctx/
+Project-URL: Repository, https://github.com/stevesolun/ctx
+Project-URL: Documentation, https://stevesolun.github.io/ctx/
+Project-URL: Changelog, https://github.com/stevesolun/ctx/blob/main/CHANGELOG.md
+Requires-Python: >=3.11
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: networkx<4,>=3
+Requires-Dist: numpy<3,>=1.24
+Requires-Dist: plotly<7,>=5
+Requires-Dist: pyyaml<7,>=6
+Provides-Extra: dev
+Requires-Dist: pytest>=8; extra == "dev"
+Requires-Dist: pytest-cov>=5; extra == "dev"
+Requires-Dist: mypy>=1.8; extra == "dev"
+Requires-Dist: ruff>=0.4; extra == "dev"
+Provides-Extra: embeddings
+Requires-Dist: sentence-transformers<4,>=2; extra == "embeddings"
+Requires-Dist: torch<3,>=2; extra == "embeddings"
+Dynamic: license-file
+
+# ctx — Skill & Agent Recommendation for Claude Code
+
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)
+[![Python 3.11+](https://img.shields.io/badge/Python-3.11+-green.svg)](https://python.org)
+[![PyPI](https://img.shields.io/pypi/v/claude-ctx.svg)](https://pypi.org/project/claude-ctx/)
+[![Tests](https://img.shields.io/badge/Tests-1319_passing-brightgreen.svg)](#)
+[![Docs](https://img.shields.io/badge/docs-MkDocs_Material-blue.svg)](https://stevesolun.github.io/ctx/)
+
+Watches what you develop, walks a knowledge graph of **1,768 skills and 443 agents**, and recommends the right ones on the fly — you decide what to load and unload. Powered by a Karpathy LLM wiki with persistent memory that gets smarter every session.
+
+## Why it exists
+
+- **Discovery** — with 1,700+ skills and 400+ agents, you can't possibly know which exist or which apply to your current repo.
+- **Context budget** — loading everything wastes tokens and degrades quality. You need the right 10–15 per session.
+- **Skill rot** — skills you installed months ago and never used are cluttering context. Stale ones should be flagged automatically.
+
+## Install
+
+```bash
+pip install claude-ctx
+```
+
+That's it. Optional extras: `pip install "claude-ctx[embeddings]"` for the semantic backend, `pip install "claude-ctx[dev]"` for the test toolchain.
+
+## Use
+
+After install, the `ctx` hooks integrate automatically with Claude Code's `PostToolUse` + `Stop` events. Typical flow:
+
+```bash
+ctx-scan-repo              # scan current repo, surface recommended skills/agents
+ctx-skill-quality list     # see the four-signal quality score for every skill
+ctx-skill-quality explain python-patterns   # drill into a single skill
+ctx-skill-health dashboard # structural health + drift detection
+ctx-toolbox run review     # run a council of skills + agents on the current diff
+```
+
+Full docs, architecture, and every module: **<https://stevesolun.github.io/ctx/>**
+
+## License
+
+MIT — see [LICENSE](LICENSE).

claude_ctx-0.5.0rc1/src/claude_ctx.egg-info/SOURCES.txt

@@ -0,0 +1,70 @@
+LICENSE
+README.md
+pyproject.toml
+src/claude_ctx.egg-info/PKG-INFO
+src/claude_ctx.egg-info/SOURCES.txt
+src/claude_ctx.egg-info/dependency_links.txt
+src/claude_ctx.egg-info/entry_points.txt
+src/claude_ctx.egg-info/requires.txt
+src/claude_ctx.egg-info/top_level.txt
+src/tests/__init__.py
+src/tests/_wiki_helpers.py
+src/tests/conftest.py
+src/tests/test_backup_config.py
+src/tests/test_backup_mirror.py
+src/tests/test_backup_retention.py
+src/tests/test_backup_watchdog.py
+src/tests/test_batch_convert.py
+src/tests/test_behavior_miner.py
+src/tests/test_config.py
+src/tests/test_context_monitor.py
+src/tests/test_corpus_cache.py
+src/tests/test_cosine_ranker.py
+src/tests/test_council_runner.py
+src/tests/test_ctx_lifecycle.py
+src/tests/test_ctx_lifecycle_auto.py
+src/tests/test_embedding_backend.py
+src/tests/test_flatten_agents.py
+src/tests/test_fs_utils.py
+src/tests/test_graph_json_integrity.py
+src/tests/test_inject_hooks_security.py
+src/tests/test_intake_gate.py
+src/tests/test_intake_pipeline.py
+src/tests/test_intent_interview.py
+src/tests/test_kpi_dashboard.py
+src/tests/test_link_conversions.py
+src/tests/test_lint.py
+src/tests/test_memory_anchor.py
+src/tests/test_orchestrator.py
+src/tests/test_quality_hook.py
+src/tests/test_query.py
+src/tests/test_resolve_skills.py
+src/tests/test_scan_repo.py
+src/tests/test_security_hardening.py
+src/tests/test_similarity_precision_recall.py
+src/tests/test_skill_add.py
+src/tests/test_skill_add_detector.py
+src/tests/test_skill_add_detector_md_escape.py
+src/tests/test_skill_add_yaml_escape.py
+src/tests/test_skill_category.py
+src/tests/test_skill_health.py
+src/tests/test_skill_loader.py
+src/tests/test_skill_loader_telemetry.py
+src/tests/test_skill_quality.py
+src/tests/test_skill_quality_bench.py
+src/tests/test_skill_quality_list.py
+src/tests/test_skill_quality_nested_agents.py
+src/tests/test_skill_telemetry.py
+src/tests/test_skill_unload.py
+src/tests/test_snapshot_if_changed.py
+src/tests/test_toolbox_cli.py
+src/tests/test_toolbox_config.py
+src/tests/test_toolbox_hooks.py
+src/tests/test_toolbox_verdict.py
+src/tests/test_usage_tracker.py
+src/tests/test_wiki.py
+src/tests/test_wiki_graphify_bench.py
+src/tests/test_wiki_graphify_quality.py
+src/tests/test_wiki_orchestrator_security.py
+src/tests/test_wiki_utils.py
+src/tests/test_wiki_visualize.py

claude_ctx-0.5.0rc1/src/claude_ctx.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

claude_ctx-0.5.0rc1/src/claude_ctx.egg-info/entry_points.txt

@@ -0,0 +1,5 @@
+[console_scripts]
+ctx-scan-repo = scan_repo:main
+ctx-skill-health = skill_health:main
+ctx-skill-quality = skill_quality:main
+ctx-toolbox = toolbox:main

claude_ctx-0.5.0rc1/src/claude_ctx.egg-info/requires.txt

@@ -0,0 +1,14 @@
+networkx<4,>=3
+numpy<3,>=1.24
+plotly<7,>=5
+pyyaml<7,>=6
+
+[dev]
+pytest>=8
+pytest-cov>=5
+mypy>=1.8
+ruff>=0.4
+
+[embeddings]
+sentence-transformers<4,>=2
+torch<3,>=2

claude_ctx-0.5.0rc1/src/claude_ctx.egg-info/top_level.txt

@@ -0,0 +1 @@
+tests

claude_ctx-0.5.0rc1/src/tests/_wiki_helpers.py

@@ -0,0 +1,117 @@
+"""
+_wiki_helpers.py -- Shared helpers for wiki_query, wiki_lint, and wiki_orchestrator tests.
+
+Exports:
+  - Schema / date constants (_SCHEMA_TAGS, _MINIMAL_SCHEMA, _TODAY, _FRESH_DATE, _STALE_DATE)
+  - make_wiki(tmp_path): build the minimal wiki skeleton used by every test
+  - make_entity_page(...): write a properly-formatted entity page to the wiki
+
+Kept out of conftest.py because these are builders, not pytest fixtures.
+Leading-underscore module name signals "internal to the tests package".
+"""
+
+from __future__ import annotations
+
+from pathlib import Path
+from typing import Any
+
+_SCHEMA_TAGS = (
+    "python", "fastapi", "docker", "testing", "architecture",
+    "patterns", "cli", "database", "async", "security",
+)
+
+_MINIMAL_SCHEMA = """\
+# Skill Wiki Schema
+
+## Domain
+Skills are reusable knowledge units.
+
+## Conventions
+Follow the naming convention: kebab-case slugs.
+
+## Tag Taxonomy
+- python: python, fastapi, testing, async
+- infra: docker, kubernetes, ci-cd
+- design: architecture, patterns, cli, database, security
+
+## Page Thresholds
+MAX_PAGE_LINES: 200
+
+## Update Policy
+Pages older than 90 days are considered stale.
+"""
+
+_TODAY = "2026-04-09"
+_FRESH_DATE = "2026-03-01"   # ~39 days before TODAY -- not stale
+_STALE_DATE = "2024-01-01"   # >90 days before TODAY -- stale
+
+
+def make_wiki(tmp_path: Path) -> Path:
+    """Create the minimal wiki skeleton (SCHEMA.md, index.md, log.md, entities/skills/)."""
+    wiki = tmp_path / "skill-wiki"
+    (wiki / "entities" / "skills").mkdir(parents=True)
+    (wiki / "SCHEMA.md").write_text(_MINIMAL_SCHEMA, encoding="utf-8")
+    (wiki / "index.md").write_text("# Index\n\n## Skills\n", encoding="utf-8")
+    (wiki / "log.md").write_text("# Log\n", encoding="utf-8")
+    return wiki
+
+
+def make_entity_page(
+    wiki_dir: Path,
+    name: str,
+    tags: list[str],
+    *,
+    body: str = "",
+    updated: str = _FRESH_DATE,
+    created: str = "2025-01-01",
+    status: str = "installed",
+    has_pipeline: bool = False,
+    wikilinks: list[str] | None = None,
+    extra_fm: dict[str, Any] | None = None,
+) -> Path:
+    """Write a properly-formatted entity page to wiki_dir/entities/skills/<name>.md.
+
+    Args:
+        wiki_dir: Root of the temporary wiki.
+        name: Slug (no extension) for the page file.
+        tags: Tag list to include in frontmatter.
+        body: Markdown body text appended after the frontmatter block.
+        updated: ISO date string for the ``updated`` frontmatter key.
+        created: ISO date string for the ``created`` frontmatter key.
+        status: ``status`` frontmatter value.
+        has_pipeline: Whether to write ``has_pipeline: true``.
+        wikilinks: Additional ``[[wikilink]]`` tokens injected into the body.
+        extra_fm: Extra key/value pairs merged into frontmatter.
+
+    Returns:
+        Path to the written file.
+    """
+    tags_str = "[" + ", ".join(tags) + "]"
+    fm_lines = [
+        "---",
+        f"title: {name}",
+        f"created: {created}",
+        f"updated: {updated}",
+        "type: skill",
+        f"tags: {tags_str}",
+        f"status: {status}",
+    ]
+    if has_pipeline:
+        # wiki_query reads `has_transformed` to populate SkillPage.has_transformed;
+        # wiki_orchestrator reads `has_pipeline` for its frontmatter checks.
+        # Write both so both modules see the flag correctly.
+        fm_lines.append("has_pipeline: true")
+        fm_lines.append("has_transformed: true")
+    if extra_fm:
+        for k, v in extra_fm.items():
+            fm_lines.append(f"{k}: {v}")
+    fm_lines.append("---")
+
+    link_block = ""
+    if wikilinks:
+        link_block = "\n" + "\n".join(f"[[{lnk}]]" for lnk in wikilinks) + "\n"
+
+    content = "\n".join(fm_lines) + "\n\n" + body + link_block
+    dest = wiki_dir / "entities" / "skills" / f"{name}.md"
+    dest.write_text(content, encoding="utf-8")
+    return dest

claude_ctx-0.5.0rc1/src/tests/conftest.py

@@ -0,0 +1,183 @@
+"""
+conftest.py -- Shared pytest fixtures for the ctx test suite.
+"""
+
+import sys
+import textwrap
+from pathlib import Path
+
+import pytest
+
+# ---------------------------------------------------------------------------
+# Ensure the project root is on sys.path so imports work from any working dir
+# ---------------------------------------------------------------------------
+_PROJECT_ROOT = Path(__file__).resolve().parent.parent
+if str(_PROJECT_ROOT) not in sys.path:
+    sys.path.insert(0, str(_PROJECT_ROOT))
+
+from wiki_sync import ensure_wiki  # noqa: E402  (import after path manipulation)
+
+
+# ---------------------------------------------------------------------------
+# Markers
+# ---------------------------------------------------------------------------
+
+
+def pytest_configure(config: pytest.Config) -> None:
+    """Register custom markers so pytest doesn't warn on unknown-mark usage.
+
+    ``integration`` is used by tests that load real models (e.g. MiniLM for
+    the similarity precision/recall harness). Skip them in fast CI with
+    ``-m 'not integration'``.
+    """
+    config.addinivalue_line(
+        "markers",
+        "integration: tests that load real embedders or external services "
+        "(skip with -m 'not integration')",
+    )
+
+
+# ---------------------------------------------------------------------------
+# Fixtures
+# ---------------------------------------------------------------------------
+
+
+@pytest.fixture()
+def project_root() -> Path:
+    """Return the ctx project root directory as a Path."""
+    return _PROJECT_ROOT
+
+
+@pytest.fixture()
+def tmp_wiki(tmp_path: Path) -> Path:
+    """
+    Create a temporary wiki directory using wiki_sync.ensure_wiki.
+
+    The returned Path is the wiki root (tmp_path / "skill-wiki").
+    All required subdirectories and seed files are created by ensure_wiki.
+    """
+    wiki_dir = tmp_path / "skill-wiki"
+    ensure_wiki(str(wiki_dir))
+    return wiki_dir
+
+
+@pytest.fixture()
+def tmp_skills_dir(tmp_path: Path) -> Path:
+    """
+    Create a temporary skills directory with three fake SKILL.md files:
+
+    - short-skill/SKILL.md  (~50 lines)
+    - medium-skill/SKILL.md (~100 lines)
+    - long-skill/SKILL.md   (~250 lines)
+    """
+    skills_dir = tmp_path / "skills"
+
+    # ── Short skill (~50 lines) ───────────────────────────────────────────
+    short_dir = skills_dir / "short-skill"
+    short_dir.mkdir(parents=True)
+    short_lines = textwrap.dedent("""\
+        ---
+        title: short-skill
+        tags: [python, testing]
+        ---
+
+        # Short Skill
+
+        ## Overview
+        A short demonstration skill used for testing.
+
+        ## Usage
+        Import and call the helper function.
+
+        ## Examples
+        ```python
+        from short_skill import helper
+        helper()
+        ```
+    """)
+    # Pad to ~50 lines
+    short_lines += "\n".join(f"# line {i}" for i in range(1, 35)) + "\n"
+    (short_dir / "SKILL.md").write_text(short_lines, encoding="utf-8")
+
+    # ── Medium skill (~100 lines) ─────────────────────────────────────────
+    medium_dir = skills_dir / "medium-skill"
+    medium_dir.mkdir(parents=True)
+    medium_lines = textwrap.dedent("""\
+        ---
+        title: medium-skill
+        tags: [python, fastapi]
+        ---
+
+        # Medium Skill
+
+        ## Overview
+        A medium-length demonstration skill used for testing.
+
+        ## Background
+        Provides patterns for FastAPI service construction.
+
+        ## Usage
+        Configure the app factory and mount routers.
+
+        ## Configuration
+        Set environment variables before startup.
+
+        ## Examples
+        ```python
+        from medium_skill import create_app
+        app = create_app()
+        ```
+
+        ## Notes
+        Remember to add lifespan handlers.
+    """)
+    # Pad to ~100 lines
+    medium_lines += "\n".join(f"# line {i}" for i in range(1, 73)) + "\n"
+    (medium_dir / "SKILL.md").write_text(medium_lines, encoding="utf-8")
+
+    # ── Long skill (~250 lines) ───────────────────────────────────────────
+    long_dir = skills_dir / "long-skill"
+    long_dir.mkdir(parents=True)
+    long_lines = textwrap.dedent("""\
+        ---
+        title: long-skill
+        tags: [python, architecture, patterns]
+        ---
+
+        # Long Skill
+
+        ## Overview
+        A long demonstration skill used for testing the line_threshold logic.
+
+        ## Stage 1: Discovery
+        Locate relevant files in the repository.
+
+        ## Stage 2: Analysis
+        Parse and classify each file by type.
+
+        ## Stage 3: Extraction
+        Pull out the key patterns and idioms.
+
+        ## Stage 4: Synthesis
+        Combine findings into a coherent summary.
+
+        ## Stage 5: Output
+        Write the final report to disk.
+
+        ## Configuration Reference
+        All options are documented below.
+
+        ## Advanced Usage
+        Chain multiple stages for batch processing.
+
+        ## Troubleshooting
+        Common issues and their resolutions.
+
+        ## See Also
+        Related skills and external references.
+    """)
+    # Pad to ~250 lines
+    long_lines += "\n".join(f"# line {i}" for i in range(1, 210)) + "\n"
+    (long_dir / "SKILL.md").write_text(long_lines, encoding="utf-8")
+
+    return skills_dir