What problem does it solve? Dependency manifests quietly accumulate footguns: packages that shadow the Python standard library, obsolete backports like dataclasses or enum34, unpinned dependencies that resolve differently on every machine, and duplicate or conflicting exact pins. This Skill inspects a single manifest and reports these surface-level problems with plain-language explanations and suggested fixes, without running a full resolver or vulnerability audit. ## Core Features & Use Cases - Offline manifest diagnosis: Parses requirements.txt, PEP 621 or Poetry pyproject.toml, and package.json using only the Python standard library, with no network calls by default. - Targeted checks: Detects standard-library shadowing pins, abandoned backports, unpinned dependencies, duplicate entries, and conflicting exact pins, each with severity, line number, and a suggested fix. - Opt-in PyPI yank check: With explicit --online approval, queries pypi.org to report exact Python pins whose releases are fully yanked. - Use Case: A developer's install fails mysteriously, so they run the doctor on requirements.txt and discover pathlib==1.0.1 shadowing the standard library plus two conflicting urllib3 pins, then apply the reviewed fixes. ## Quick Start Ask your agent to check my requirements.txt for dependency problems and explain any findings with suggested fixes.
