[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6XENXZ42WAYK7EJ4QHTCTHF":3,"skill-same:01M6XENXZ42WAYK7EJ4QHTCTHF":52},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":26,"linkOnly":27,"markdownContent":28,"markdownBody":29,"frontMatter":30,"refName":31,"commitSha":32,"skillDirectory":33,"forksCount":13,"repositoryHtmlUrl":34,"repositoryHomepageUrl":6,"ownerHtmlUrl":35,"ownerBlogUrl":6,"ownerDescription":36,"updatedAt":37,"install":38,"resources":51},"01M6XENXZ42WAYK7EJ4QHTCTHF","discover-important-function",null,"When given a project codebase, this skill observes the important functions in the codebase for future action.","Waaaalnut\u002Fskillsbench","Waaaalnut","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F103995656?v=4","01M6XCS9K1V51EY3JMB7DGMX24","01M6XCS9Q4WPW88W62C55G1H44",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FWaaaalnut\u002Fskillsbench\u002Fblob\u002Fmain\u002Ftasks\u002Fsetup-fuzzing-py\u002Fenvironment\u002Fskills\u002Fdiscover-important-function\u002FSKILL.md",[],[18,19,20,21,22,23,24,25],"discover","discover important function","functions","project","action","for","codebase","when","2026-11-10 10:12:30",false,"---\nname: discover-important-function\ndescription: \"When given a project codebase, this skill observes the important functions in the codebase for future action.\"\n---\n\n# Fuzz Target Localizer\n\n## Purpose\n\nThis skill helps an agent quickly narrow a Python repository down to a small set of high-value fuzz targets. It produces:\n\n- A ranked list of important files\n- A ranked list of important functions and methods\n- A summary of existing unit tests and inferred oracles\n- A final shortlist of functions-under-test (FUTs) for fuzzing, each with a structured \"note to self\" for follow-up actions\n\n## When to use\n\nUse this skill when the user asks to:\n\n- Find the best functions to fuzz in a Python package\u002Flibrary\n- Identify parsers, decoders, validators, or boundary code that is fuzz-worthy\n- Decide what to fuzz based on existing test coverage and API surface\n- Produce a structured shortlist of fuzz targets with harness guidance\n- Automating testing pipeline setup for a new or existing Python project.\n\nDo not use this skill when the user primarily wants to fix a specific bug,\nrefactor code, or implement a harness immediately\n(unless they explicitly ask for target selection first).\n\n## Inputs expected from the environment\n\nThe agent should assume access to:\n\n- Repository filesystem\n- Ability to read files\n- Ability to run local commands (optional but recommended)\n\nPreferred repository listing command:\n\n- Use `tree` to get a fast, high-signal view of repository structure (limit depth if needed).\n\nIf `tree` is not available, fall back to a recursive listing via other standard shell tooling.\n\n## Outputs\n\nProduce result:\n\nA report in any format with the following information: - Localize important files - Localize important functions - Summarize existing unit tests - Decide function under test for fuzzing\nThis file should be placed in the root of the repository as `APIs.txt`,\nwhich servers as the guidelines for future fuzzing harness implementation.\n\n## Guardrails\n\n- Prefer analysis and reporting over code changes.\n- Do not modify source code unless the user explicitly requests changes.\n- If running commands could be disruptive, default to read-only analysis.\n- Avoid assumptions about runtime behavior; base conclusions on code and tests.\n- Explicitly consider existing tests in the repo when selecting targets and deciding harness shape.\n- Keep the final FUT shortlist small (typically 1–5).\n\n---\n\n# Localize important files\n\n## Goal\n\nProduce a ranked list of files that are most likely to contain fuzz-worthy logic: parsing, decoding, deserialization, validation, protocol handling, file\u002Fnetwork boundaries, or native bindings.\n\n## Procedure\n\n1. Build a repository map\n    - Start with a repository overview using `tree` to identify:\n        - Root packages and layouts (src\u002F layout vs flat layout)\n        - Test directories and configs\n        - Bindings\u002Fnative directories\n        - Examples, docs, and tooling directories\n    - Identify packaging and metadata files such as:\n        - pyproject.toml\n        - setup.cfg\n        - setup.py\n    - Identify test configuration and entry points:\n        - tests\u002F\n        - conftest.py\n        - pytest.ini\n        - tox.ini\n        - noxfile.py\n\n2. Exclude low-value areas\n    - Skip: virtual environments, build outputs, vendored code, documentation-only directories, examples-only directories, generated files.\n\n3. Score files using explainable heuristics\n   Assign a file a higher score when it matches more of these indicators:\n    - Public API exposure: **init**.py re-exports, **all**, api modules\n    - Input boundary keywords in file path or symbols: parse, load, dump, decode, encode, deserialize, serialize, validate, normalize, schema, protocol, message\n    - Format handlers: json, yaml, xml, csv, toml, protobuf, msgpack, pickle\n    - Regex-heavy or templating-heavy code\n    - Native boundaries: ctypes, cffi, cython, extension modules, bindings\n    - Central modules: high import fan-in across the package\n    - Test adjacency: directly imported or heavily referenced by tests\n\n4. Rank and select\n    - Produce a Top-N list (default 10–30) with a short rationale per file.\n\n## Output format\n\nFor each file:\n\n- path\n- score (relative, not necessarily normalized)\n- rationale (2–5 bullets)\n- indicators_hit (list)\n\n---\n\n# Localize important functions\n\n## Goal\n\nFrom the important files, identify and rank functions or methods that are strong fuzz targets while minimizing full-body reading until needed.\n\n## Approach 1: AST-based header and docstring scan\n\nFor each localized Python file, parse it using Python’s `ast` module and extract only:\n\n- Module docstring\n- Function and async function headers (name, args, defaults, annotations, decorators)\n- Class headers and method headers\n- Docstrings for modules, classes, and functions\u002Fmethods\n\nDo not read full function bodies during the initial pass unless needed for disambiguation or final selection.\n\n### Procedure\n\n1. Build per-file declarations via AST\n    - Parse the file with `ast`\n    - Enumerate:\n        - Top-level functions\n        - Classes and their methods\n        - Nested functions only if they are likely to be directly fuzzable via an exposed wrapper\n    - For each symbol, collect:\n        - Fully-qualified name\n        - Signature details (as available from AST)\n        - Decorators\n        - Docstring (if present)\n        - Location information (file, line range if available)\n\n2. Generate an initial candidate set using headers and docstrings\n   Prioritize functions\u002Fmethods that:\n    - Accept bytes, str, file-like objects, dicts, or user-controlled payloads\n    - Convert between representations (raw ↔ structured)\n    - Perform validation, normalization, parsing, decoding, deserialization\n    - Touch filesystem\u002Fnetwork\u002Fprotocol boundaries\n    - Call into native extensions or bindings\n    - Clearly document strictness, schemas, formats, or error conditions\n\n3. Use tests to refine candidate selection early\n    - Before reading full bodies, check if tests reference these functions\u002Fmodules:\n        - Direct imports in tests\n        - Fixtures that exercise particular entry points\n        - Parameterizations over formats and inputs\n    - Down-rank candidates that are already well-covered unless they are high-risk boundaries (parsers\u002Fnative bindings).\n\n4. Confirm with targeted reading only for top candidates\n   For the top candidates (typically 10–20), read the full function bodies and capture:\n    - Preconditions and assumptions\n    - Internal helpers called\n    - Error handling style and exception types\n    - Any obvious invariants and postconditions\n    - Statefulness and global dependencies\n\n5. Rank and shortlist\n   Rank candidates using an explainable rubric:\n    - Input surface and reachability\n    - Boundary risk (parsing\u002Fdecoding\u002Fnative)\n    - Structural complexity (from targeted reading only)\n    - Existing test coverage strength and breadth\n    - Ease of harnessing\n\n### Output format\n\nFor each function\u002Fmethod:\n\n- qualname\n- file\n- line_range (if available)\n- score (relative)\n- rationale (2–6 bullets)\n- dependencies (key helpers, modules, external state)\n- harnessability (low\u002Fmedium\u002Fhigh)\n\n## Approach 2: Scanning all important files yourself\n\nFor each localized Python file, read the file contents directly to extract:\n\n- Module docstring and overall structure\n- Function and async function definitions (name, args, defaults, annotations, decorators)\n- Class definitions and their methods\n- Full function bodies and implementation details\n- Docstrings for modules, classes, and functions\u002Fmethods\n\nThis approach reads complete file contents, allowing for deeper analysis at the cost of higher token usage.\n\n### Procedure\n\n1. Read important files sequentially\n    - For each file from the important files list, read the full contents.\n    - Extract by direct inspection:\n        - Top-level functions and their complete implementations\n        - Classes and their methods with full bodies\n        - Nested functions if they are exposed or called by public APIs\n    - For each symbol, collect:\n        - Fully-qualified name\n        - Complete signature (from source text)\n        - Decorators\n        - Docstring (if present)\n        - Full function body\n        - Location information (file, approximate line range)\n\n2. Generate an initial candidate set using full source analysis\n   Prioritize functions\u002Fmethods that:\n    - Accept bytes, str, file-like objects, dicts, or user-controlled payloads\n    - Convert between representations (raw ↔ structured)\n    - Perform validation, normalization, parsing, decoding, deserialization\n    - Touch filesystem\u002Fnetwork\u002Fprotocol boundaries\n    - Call into native extensions or bindings\n    - Contain complex control flow, loops, or recursive calls\n    - Handle exceptions or edge cases\n    - Clearly document strictness, schemas, formats, or error conditions\n\n3. Analyze implementation details from full bodies\n   For each candidate function, inspect the body for:\n    - Preconditions and assumptions (explicit checks, assertions, early returns)\n    - Internal helpers called and their purposes\n    - Error handling style and exception types raised\n    - Invariants and postconditions (explicit or implicit)\n    - Statefulness and global dependencies\n    - Input transformations and data flow\n    - Native calls or external process invocations\n    - Resource allocation and cleanup patterns\n\n4. Use tests to refine candidate selection\n    - Check if tests reference these functions\u002Fmodules:\n        - Direct imports in tests\n        - Fixtures that exercise particular entry points\n        - Parameterizations over formats and inputs\n    - Down-rank candidates that are already well-covered unless they are high-risk boundaries (parsers\u002Fnative bindings).\n    - Note which aspects of each function are tested vs untested.\n\n5. Rank and shortlist\n   Rank candidates using an explainable rubric:\n    - Input surface and reachability\n    - Boundary risk (parsing\u002Fdecoding\u002Fnative)\n    - Structural complexity (from full body analysis)\n    - Existing test coverage strength and breadth\n    - Ease of harnessing\n    - Observable implementation risks (unsafe operations, unchecked inputs, complex state)\n\n### Output format\n\nFor each function\u002Fmethod:\n\n- qualname\n- file\n- line_range (if available)\n- score (relative)\n- rationale (2–6 bullets)\n- dependencies (key helpers, modules, external state)\n- harnessability (low\u002Fmedium\u002Fhigh)\n- implementation_notes (key observations from body analysis)\n\n---\n\n# Summarize existing unit tests\n\n## Goal\n\nSummarize what is already tested, infer test oracles, and identify gaps that fuzzing can complement.\n\n## Hard requirement\n\nAlways inspect and incorporate existing tests in the repository when:\n\n- Ranking functions\n- Selecting FUTs\n- Designing input models and oracles\n- Proposing seed corpus sources\n\n## Procedure\n\n1. Inventory tests\n    - Locate tests and their discovery configuration (pytest.ini, pyproject.toml, tox.ini, noxfile.py).\n    - Enumerate test modules and map them to source modules via imports.\n    - Identify shared fixtures and data factories (conftest.py, fixture files, test utilities).\n\n2. Summarize test intent\n   For each test module:\n    - What behaviors are asserted\n    - What inputs are used\n    - What exceptions are expected\n    - What invariants are implied\n\n3. Infer oracles and properties\n   Common fuzz-friendly oracles include:\n    - Round-trip properties\n    - Idempotence of normalization\n    - Parser consistency across equivalent inputs\n    - Deterministic output given deterministic input\n    - No-crash and no-hang for malformed inputs\n\n4. Identify coverage gaps\n    - FUT candidates with no direct tests\n    - Input classes not covered by tests (size extremes, malformed encodings, deep nesting, edge unicode, invalid schemas)\n    - Code paths guarded by complex conditionals or exception handlers with no tests\n\n## Output format\n\n- test_map: module_under_test → tests → asserted behaviors\n- inferred_oracles: list of reusable invariants with the functions they apply to\n- gaps: ranked list of untested or weakly-tested candidates\n\n---\n\n# Decide function under test for fuzzing\n\n## Goal\n\nSelect a final set of FUTs (typically 1–5) and produce a structured \"note to self\" for each FUT so the agent can proceed to harness implementation, corpus seeding, and fuzz execution.\n\n## Selection criteria\n\nPrefer FUTs that maximize:\n\n- Security\u002Frobustness payoff (parsing, decoding, validation, native boundary)\n- Reachability with minimal setup\n- Low existing test assurance or narrow test input coverage\n- High fuzzability (simple input channel, clear oracle or crash-only target)\n\nExplicitly weigh:\n\n- What tests already cover (and what they do not)\n- What seeds can be extracted from tests, fixtures, and sample data\n\n## Required \"note to self\" template\n\nFor each selected FUT, produce exactly this structure:\n\nFuzzing Target Note\n\n- Target:\n- File \u002F location:\n- Why this target:\n    - Input surface:\n    - Boundary or native considerations:\n    - Complexity or path depth:\n    - Current test gaps:\n- Callable contract:\n    - Required imports or initialization:\n    - Preconditions:\n    - Determinism concerns:\n    - External dependencies:\n- Input model:\n    - Primary payload type:\n    - Decoding or parsing steps:\n    - Constraints to respect:\n    - Edge classes to emphasize:\n- Oracles:\n    - Must-hold properties:\n    - Acceptable exceptions:\n    - Suspicious exceptions:\n    - Crash-only vs correctness-checking:\n- Harness plan:\n    - Recommended approach:\n    - Minimal harness signature:\n    - Seed corpus ideas:\n    - Timeouts and resource limits:\n- Risk flags:\n    - Native extension involved:\n    - Potential DoS paths:\n    - External I\u002FO:\n- Next actions:\n    1.\n    2.\n    3.\n    4.\n\n## Output format\n\n- selected_futs: list of chosen FUTs with brief justification\n- notes_to_self: one \"Fuzzing Target Note\" per FUT\n\n---\n\n# Final JSON block\n\nAt the end of the report, include a JSON object with:\n\n- important_files: [{path, score, rationale, indicators_hit}]\n- important_functions: [{qualname, file, line_range, score, rationale, dependencies, harnessability}]\n- test_summary: {test_map, inferred_oracles, gaps}\n- selected_futs: [{qualname, file, line_range, justification}]\n- notes_to_self: [{target_qualname, note}]\n","# Fuzz Target Localizer\n\n## Purpose\n\nThis skill helps an agent quickly narrow a Python repository down to a small set of high-value fuzz targets. It produces:\n\n- A ranked list of important files\n- A ranked list of important functions and methods\n- A summary of existing unit tests and inferred oracles\n- A final shortlist of functions-under-test (FUTs) for fuzzing, each with a structured \"note to self\" for follow-up actions\n\n## When to use\n\nUse this skill when the user asks to:\n\n- Find the best functions to fuzz in a Python package\u002Flibrary\n- Identify parsers, decoders, validators, or boundary code that is fuzz-worthy\n- Decide what to fuzz based on existing test coverage and API surface\n- Produce a structured shortlist of fuzz targets with harness guidance\n- Automating testing pipeline setup for a new or existing Python project.\n\nDo not use this skill when the user primarily wants to fix a specific bug,\nrefactor code, or implement a harness immediately\n(unless they explicitly ask for target selection first).\n\n## Inputs expected from the environment\n\nThe agent should assume access to:\n\n- Repository filesystem\n- Ability to read files\n- Ability to run local commands (optional but recommended)\n\nPreferred repository listing command:\n\n- Use `tree` to get a fast, high-signal view of repository structure (limit depth if needed).\n\nIf `tree` is not available, fall back to a recursive listing via other standard shell tooling.\n\n## Outputs\n\nProduce result:\n\nA report in any format with the following information: - Localize important files - Localize important functions - Summarize existing unit tests - Decide function under test for fuzzing\nThis file should be placed in the root of the repository as `APIs.txt`,\nwhich servers as the guidelines for future fuzzing harness implementation.\n\n## Guardrails\n\n- Prefer analysis and reporting over code changes.\n- Do not modify source code unless the user explicitly requests changes.\n- If running commands could be disruptive, default to read-only analysis.\n- Avoid assumptions about runtime behavior; base conclusions on code and tests.\n- Explicitly consider existing tests in the repo when selecting targets and deciding harness shape.\n- Keep the final FUT shortlist small (typically 1–5).\n\n---\n\n# Localize important files\n\n## Goal\n\nProduce a ranked list of files that are most likely to contain fuzz-worthy logic: parsing, decoding, deserialization, validation, protocol handling, file\u002Fnetwork boundaries, or native bindings.\n\n## Procedure\n\n1. Build a repository map\n    - Start with a repository overview using `tree` to identify:\n        - Root packages and layouts (src\u002F layout vs flat layout)\n        - Test directories and configs\n        - Bindings\u002Fnative directories\n        - Examples, docs, and tooling directories\n    - Identify packaging and metadata files such as:\n        - pyproject.toml\n        - setup.cfg\n        - setup.py\n    - Identify test configuration and entry points:\n        - tests\u002F\n        - conftest.py\n        - pytest.ini\n        - tox.ini\n        - noxfile.py\n\n2. Exclude low-value areas\n    - Skip: virtual environments, build outputs, vendored code, documentation-only directories, examples-only directories, generated files.\n\n3. Score files using explainable heuristics\n   Assign a file a higher score when it matches more of these indicators:\n    - Public API exposure: **init**.py re-exports, **all**, api modules\n    - Input boundary keywords in file path or symbols: parse, load, dump, decode, encode, deserialize, serialize, validate, normalize, schema, protocol, message\n    - Format handlers: json, yaml, xml, csv, toml, protobuf, msgpack, pickle\n    - Regex-heavy or templating-heavy code\n    - Native boundaries: ctypes, cffi, cython, extension modules, bindings\n    - Central modules: high import fan-in across the package\n    - Test adjacency: directly imported or heavily referenced by tests\n\n4. Rank and select\n    - Produce a Top-N list (default 10–30) with a short rationale per file.\n\n## Output format\n\nFor each file:\n\n- path\n- score (relative, not necessarily normalized)\n- rationale (2–5 bullets)\n- indicators_hit (list)\n\n---\n\n# Localize important functions\n\n## Goal\n\nFrom the important files, identify and rank functions or methods that are strong fuzz targets while minimizing full-body reading until needed.\n\n## Approach 1: AST-based header and docstring scan\n\nFor each localized Python file, parse it using Python’s `ast` module and extract only:\n\n- Module docstring\n- Function and async function headers (name, args, defaults, annotations, decorators)\n- Class headers and method headers\n- Docstrings for modules, classes, and functions\u002Fmethods\n\nDo not read full function bodies during the initial pass unless needed for disambiguation or final selection.\n\n### Procedure\n\n1. Build per-file declarations via AST\n    - Parse the file with `ast`\n    - Enumerate:\n        - Top-level functions\n        - Classes and their methods\n        - Nested functions only if they are likely to be directly fuzzable via an exposed wrapper\n    - For each symbol, collect:\n        - Fully-qualified name\n        - Signature details (as available from AST)\n        - Decorators\n        - Docstring (if present)\n        - Location information (file, line range if available)\n\n2. Generate an initial candidate set using headers and docstrings\n   Prioritize functions\u002Fmethods that:\n    - Accept bytes, str, file-like objects, dicts, or user-controlled payloads\n    - Convert between representations (raw ↔ structured)\n    - Perform validation, normalization, parsing, decoding, deserialization\n    - Touch filesystem\u002Fnetwork\u002Fprotocol boundaries\n    - Call into native extensions or bindings\n    - Clearly document strictness, schemas, formats, or error conditions\n\n3. Use tests to refine candidate selection early\n    - Before reading full bodies, check if tests reference these functions\u002Fmodules:\n        - Direct imports in tests\n        - Fixtures that exercise particular entry points\n        - Parameterizations over formats and inputs\n    - Down-rank candidates that are already well-covered unless they are high-risk boundaries (parsers\u002Fnative bindings).\n\n4. Confirm with targeted reading only for top candidates\n   For the top candidates (typically 10–20), read the full function bodies and capture:\n    - Preconditions and assumptions\n    - Internal helpers called\n    - Error handling style and exception types\n    - Any obvious invariants and postconditions\n    - Statefulness and global dependencies\n\n5. Rank and shortlist\n   Rank candidates using an explainable rubric:\n    - Input surface and reachability\n    - Boundary risk (parsing\u002Fdecoding\u002Fnative)\n    - Structural complexity (from targeted reading only)\n    - Existing test coverage strength and breadth\n    - Ease of harnessing\n\n### Output format\n\nFor each function\u002Fmethod:\n\n- qualname\n- file\n- line_range (if available)\n- score (relative)\n- rationale (2–6 bullets)\n- dependencies (key helpers, modules, external state)\n- harnessability (low\u002Fmedium\u002Fhigh)\n\n## Approach 2: Scanning all important files yourself\n\nFor each localized Python file, read the file contents directly to extract:\n\n- Module docstring and overall structure\n- Function and async function definitions (name, args, defaults, annotations, decorators)\n- Class definitions and their methods\n- Full function bodies and implementation details\n- Docstrings for modules, classes, and functions\u002Fmethods\n\nThis approach reads complete file contents, allowing for deeper analysis at the cost of higher token usage.\n\n### Procedure\n\n1. Read important files sequentially\n    - For each file from the important files list, read the full contents.\n    - Extract by direct inspection:\n        - Top-level functions and their complete implementations\n        - Classes and their methods with full bodies\n        - Nested functions if they are exposed or called by public APIs\n    - For each symbol, collect:\n        - Fully-qualified name\n        - Complete signature (from source text)\n        - Decorators\n        - Docstring (if present)\n        - Full function body\n        - Location information (file, approximate line range)\n\n2. Generate an initial candidate set using full source analysis\n   Prioritize functions\u002Fmethods that:\n    - Accept bytes, str, file-like objects, dicts, or user-controlled payloads\n    - Convert between representations (raw ↔ structured)\n    - Perform validation, normalization, parsing, decoding, deserialization\n    - Touch filesystem\u002Fnetwork\u002Fprotocol boundaries\n    - Call into native extensions or bindings\n    - Contain complex control flow, loops, or recursive calls\n    - Handle exceptions or edge cases\n    - Clearly document strictness, schemas, formats, or error conditions\n\n3. Analyze implementation details from full bodies\n   For each candidate function, inspect the body for:\n    - Preconditions and assumptions (explicit checks, assertions, early returns)\n    - Internal helpers called and their purposes\n    - Error handling style and exception types raised\n    - Invariants and postconditions (explicit or implicit)\n    - Statefulness and global dependencies\n    - Input transformations and data flow\n    - Native calls or external process invocations\n    - Resource allocation and cleanup patterns\n\n4. Use tests to refine candidate selection\n    - Check if tests reference these functions\u002Fmodules:\n        - Direct imports in tests\n        - Fixtures that exercise particular entry points\n        - Parameterizations over formats and inputs\n    - Down-rank candidates that are already well-covered unless they are high-risk boundaries (parsers\u002Fnative bindings).\n    - Note which aspects of each function are tested vs untested.\n\n5. Rank and shortlist\n   Rank candidates using an explainable rubric:\n    - Input surface and reachability\n    - Boundary risk (parsing\u002Fdecoding\u002Fnative)\n    - Structural complexity (from full body analysis)\n    - Existing test coverage strength and breadth\n    - Ease of harnessing\n    - Observable implementation risks (unsafe operations, unchecked inputs, complex state)\n\n### Output format\n\nFor each function\u002Fmethod:\n\n- qualname\n- file\n- line_range (if available)\n- score (relative)\n- rationale (2–6 bullets)\n- dependencies (key helpers, modules, external state)\n- harnessability (low\u002Fmedium\u002Fhigh)\n- implementation_notes (key observations from body analysis)\n\n---\n\n# Summarize existing unit tests\n\n## Goal\n\nSummarize what is already tested, infer test oracles, and identify gaps that fuzzing can complement.\n\n## Hard requirement\n\nAlways inspect and incorporate existing tests in the repository when:\n\n- Ranking functions\n- Selecting FUTs\n- Designing input models and oracles\n- Proposing seed corpus sources\n\n## Procedure\n\n1. Inventory tests\n    - Locate tests and their discovery configuration (pytest.ini, pyproject.toml, tox.ini, noxfile.py).\n    - Enumerate test modules and map them to source modules via imports.\n    - Identify shared fixtures and data factories (conftest.py, fixture files, test utilities).\n\n2. Summarize test intent\n   For each test module:\n    - What behaviors are asserted\n    - What inputs are used\n    - What exceptions are expected\n    - What invariants are implied\n\n3. Infer oracles and properties\n   Common fuzz-friendly oracles include:\n    - Round-trip properties\n    - Idempotence of normalization\n    - Parser consistency across equivalent inputs\n    - Deterministic output given deterministic input\n    - No-crash and no-hang for malformed inputs\n\n4. Identify coverage gaps\n    - FUT candidates with no direct tests\n    - Input classes not covered by tests (size extremes, malformed encodings, deep nesting, edge unicode, invalid schemas)\n    - Code paths guarded by complex conditionals or exception handlers with no tests\n\n## Output format\n\n- test_map: module_under_test → tests → asserted behaviors\n- inferred_oracles: list of reusable invariants with the functions they apply to\n- gaps: ranked list of untested or weakly-tested candidates\n\n---\n\n# Decide function under test for fuzzing\n\n## Goal\n\nSelect a final set of FUTs (typically 1–5) and produce a structured \"note to self\" for each FUT so the agent can proceed to harness implementation, corpus seeding, and fuzz execution.\n\n## Selection criteria\n\nPrefer FUTs that maximize:\n\n- Security\u002Frobustness payoff (parsing, decoding, validation, native boundary)\n- Reachability with minimal setup\n- Low existing test assurance or narrow test input coverage\n- High fuzzability (simple input channel, clear oracle or crash-only target)\n\nExplicitly weigh:\n\n- What tests already cover (and what they do not)\n- What seeds can be extracted from tests, fixtures, and sample data\n\n## Required \"note to self\" template\n\nFor each selected FUT, produce exactly this structure:\n\nFuzzing Target Note\n\n- Target:\n- File \u002F location:\n- Why this target:\n    - Input surface:\n    - Boundary or native considerations:\n    - Complexity or path depth:\n    - Current test gaps:\n- Callable contract:\n    - Required imports or initialization:\n    - Preconditions:\n    - Determinism concerns:\n    - External dependencies:\n- Input model:\n    - Primary payload type:\n    - Decoding or parsing steps:\n    - Constraints to respect:\n    - Edge classes to emphasize:\n- Oracles:\n    - Must-hold properties:\n    - Acceptable exceptions:\n    - Suspicious exceptions:\n    - Crash-only vs correctness-checking:\n- Harness plan:\n    - Recommended approach:\n    - Minimal harness signature:\n    - Seed corpus ideas:\n    - Timeouts and resource limits:\n- Risk flags:\n    - Native extension involved:\n    - Potential DoS paths:\n    - External I\u002FO:\n- Next actions:\n    1.\n    2.\n    3.\n    4.\n\n## Output format\n\n- selected_futs: list of chosen FUTs with brief justification\n- notes_to_self: one \"Fuzzing Target Note\" per FUT\n\n---\n\n# Final JSON block\n\nAt the end of the report, include a JSON object with:\n\n- important_files: [{path, score, rationale, indicators_hit}]\n- important_functions: [{qualname, file, line_range, score, rationale, dependencies, harnessability}]\n- test_summary: {test_map, inferred_oracles, gaps}\n- selected_futs: [{qualname, file, line_range, justification}]\n- notes_to_self: [{target_qualname, note}]\n",{"name":5,"description":7},"main","7eeb258ada72b7f58119b0e7a2c00ea89d9beedd","tasks\u002Fsetup-fuzzing-py\u002Fenvironment\u002Fskills\u002Fdiscover-important-function","https:\u002F\u002Fgithub.com\u002FWaaaalnut\u002Fskillsbench","https:\u002F\u002Fgithub.com\u002FWaaaalnut","student of Zhejiang University","2026-04-21 23:16:23",{"source":8,"skillName":5,"commands":39,"prompt":49,"zipAvailable":50},[40,43,46],{"tool":41,"command":42},"npx","npx skills add Waaaalnut\u002Fskillsbench --skill discover-important-function",{"tool":44,"command":45},"bunx","bunx skills add Waaaalnut\u002Fskillsbench --skill discover-important-function",{"tool":47,"command":48},"pnpm","pnpm dlx skills add Waaaalnut\u002Fskillsbench --skill discover-important-function","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FWaaaalnut\u002Fskillsbench\u002Fblob\u002Fmain\u002Ftasks\u002Fsetup-fuzzing-py\u002Fenvironment\u002Fskills\u002Fdiscover-important-function\u002FSKILL.md",true,[],[53,67,82,109,131,149],{"skillNo":54,"name":55,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":56,"categories":57,"keywords":63,"publishedAt":26},"01M6XEWMJ6VY3ZC53MPDCPG3GH","data-reconciliation","https:\u002F\u002Fgithub.com\u002FWaaaalnut\u002Fskillsbench\u002Fblob\u002Fmain\u002Ftasks\u002Fxlsx-recover-data\u002Fenvironment\u002Fskills\u002Fdata-reconciliation\u002FSKILL.md",[58],{"categoryNo":59,"categoryName":60,"taxonomyType":61,"taxonomyTypeDesc":62},"01JBSEEDCAT000000000000033","数据分析师","PROFESSION","适用职业",[64,65,66],"data reconciliation","reconciliation","data",{"skillNo":68,"name":69,"nameZh":6,"description":70,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":71,"categories":72,"keywords":73,"publishedAt":26},"01M6XEW8MNPMF1KRXV6ATGFRQF","speech-to-text","Transcribe video to timestamped text using Whisper tiny model (pre-installed).","https:\u002F\u002Fgithub.com\u002FWaaaalnut\u002Fskillsbench\u002Fblob\u002Fmain\u002Ftasks\u002Fvideo-tutorial-indexer\u002Fenvironment\u002Fskills\u002Fspeech-to-text\u002FSKILL.md",[],[74,75,76,77,78,79,80,81],"speech","text","speech to text","transcribe video","whisper","model","Video","transcribe",{"skillNo":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":93,"publishedAt":26},"01M6XEW3FQZP0T1RKQQJFXVY9T","video-processor","Process videos by removing segments and concatenating remaining parts. Use when you need to remove detected pauses\u002Fopenings from videos, create highlight reels, or batch process segment removals using ffmpeg filter_complex.","https:\u002F\u002Fgithub.com\u002FWaaaalnut\u002Fskillsbench\u002Fblob\u002Fmain\u002Ftasks\u002Fvideo-silence-remover\u002Fenvironment\u002Fskills\u002Fvideo-processor\u002FSKILL.md",[88],{"categoryNo":89,"categoryName":90,"taxonomyType":91,"taxonomyTypeDesc":92},"01JBSEEDCAT000000000000016","图像与多媒体","SCENARIO","使用场景",[80,94,95,96,97,98,25,99,100,101,102,103,104,105,106,107,108],"processor","video processor","batch","create","reels","parts","segments","process","videos","you","segment","filter","ffmpeg","use","remove",{"skillNo":110,"name":111,"nameZh":6,"description":112,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":113,"categories":114,"keywords":116,"publishedAt":26},"01M6XEVXJ5KZ0NTY3XZKRY4AS1","silence-detector","Detect initial silence segments in audio\u002Fvideo using energy-based analysis. Use when you need to find low-energy periods at the start of recordings (title slides, setup time, pre-roll silence).","https:\u002F\u002Fgithub.com\u002FWaaaalnut\u002Fskillsbench\u002Fblob\u002Fmain\u002Ftasks\u002Fvideo-silence-remover\u002Fenvironment\u002Fskills\u002Fsilence-detector\u002FSKILL.md",[115],{"categoryNo":89,"categoryName":90,"taxonomyType":91,"taxonomyTypeDesc":92},[117,118,119,120,25,103,121,122,123,124,125,100,126,127,80,128,129,130,107],"detector","silence detector","title","slides","time","energy","at","recordings","detect","find","Audio","analysis","start","setup",{"skillNo":132,"name":133,"nameZh":6,"description":134,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":135,"categories":136,"keywords":137,"publishedAt":26},"01M6XEVT3NGRN1FT7JPMAPXS6M","segment-combiner","Combine multiple segment detection results into a unified list. Use when you need to merge segments from different detectors, prepare removal lists for video processing, or consolidate detection outputs.","https:\u002F\u002Fgithub.com\u002FWaaaalnut\u002Fskillsbench\u002Fblob\u002Fmain\u002Ftasks\u002Fvideo-silence-remover\u002Fenvironment\u002Fskills\u002Fsegment-combiner\u002FSKILL.md",[],[104,138,80,139,140,141,142,23,25,143,144,100,103,145,146,107,147,148],"segment combiner","detection","prepare","merge","video processing","outputs","combine","processing","consolidate","results","list",{"skillNo":150,"name":151,"nameZh":6,"description":152,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":153,"categories":154,"keywords":158,"publishedAt":26},"01M6XEVMPVASGS2KX86QK4FMT7","report-generator","Generate compression reports for video processing. Use when you need to create structured JSON reports with duration statistics, compression ratios, and segment details after video processing.","https:\u002F\u002Fgithub.com\u002FWaaaalnut\u002Fskillsbench\u002Fblob\u002Fmain\u002Ftasks\u002Fvideo-silence-remover\u002Fenvironment\u002Fskills\u002Freport-generator\u002FSKILL.md",[155],{"categoryNo":156,"categoryName":157,"taxonomyType":61,"taxonomyTypeDesc":62},"01JBSEEDCAT000000000000034","行政与办公人员",[159,160,161,104,25,23,103,162,145,163,164,165,107,142,166,80,167,168,97],"generator","report generator","report","with","statistics","structured","reports","json","compression","generate"]