AGENT SKILL

dependency-doctor

Checks requirements.txt, pyproject.toml, and package.json dependency manifests for surface-level direct-dependency footguns: standard-library shadowing pins, abandoned backports, unpinned dependencies, and obvious intra-manifest conflicts, plus opt-in PyPI yanked releases. Use when the user asks to check a manifest for dependency problems, asks why dependencies won't install or whether anything is wrong with their dependencies, wants a dependency autopsy, or suspects dependency manifest rot. Runs offline by default as a local tool for the user's own project, not repository CI.

License 许可
Apache-2.0
最近更新
2026-11-14
Stars
★ 0
Forks
0

01 前言 Front Matter

name
dependency-doctor
license
Apache-2.0
metadata
{ "author": "Matt Van Horn", "source": "https://github.com/Shubhamsaboo/awesome-llm-apps", "version": "1.0.0" }
description
Checks requirements.txt, pyproject.toml, and package.json dependency manifests for surface-level direct-dependency footguns: standard-library shadowing pins, abandoned backports, unpinned dependencies, and obvious intra-manifest conflicts, plus opt-in PyPI yanked releases. Use when the user asks to check a manifest for dependency problems, asks why dependencies won't install or whether anything is wrong with their dependencies, wants a dependency autopsy, or suspects dependency manifest rot. Runs offline by default as a local tool for the user's own project, not repository CI.
compatibility
Python 3.11+. Offline by default. Network access to pypi.org occurs only when the user explicitly approves --online.

02 资源文件 6

03 SKILL.md 正文

Dependency Doctor

Inspect one dependency manifest on the user's machine for direct, surface-level footguns. Explain each finding in plain language, then offer a small, reviewable fix. This does not diagnose a failed pip or uv resolution.

This is a local developer tool for a project the user chooses. It is not a repository-wide lint rule, a CI gate, or a proposal to enforce dependency policy across unrelated apps.

When to use

  • The user asks to check, audit, diagnose, or autopsy a dependency manifest
  • The user wants to rule out direct-manifest issues before deeper install debugging
  • The user suspects stale pins, backports, duplicate entries, or dependency rot
  • The user asks whether anything looks wrong with their dependencies

When not to use

  • Installing the current dependencies without diagnosing them
  • Upgrading every package or adding a new package
  • A full vulnerability audit. Use pip-audit, npm audit, or the project's approved security scanner for CVE coverage
  • Creating a repo-wide CI check. This skill is user-invoked and local

Choose the manifest

Use the path the user names. If no path is given and several manifests exist, ask which one to inspect. Do not sweep the repository or edit anything merely because the skill was triggered.

Supported inputs:

  • requirements.txt
  • pyproject.toml using PEP 621 or common Poetry dependency tables
  • package.json dependency sections

Run the offline diagnosis

From this skill directory:

python3 scripts/dep_doctor.py /path/to/requirements.txt --json

The default path is fully offline. It reads only the selected manifest. The report shape is:

{
  "file": "/path/to/requirements.txt",
  "findings": [
    {
      "severity": "high",
      "kind": "stdlib-shadowing",
      "package": "pathlib",
      "line": 4,
      "why": "...",
      "fix": "..."
    }
  ],
  "summary": {
    "total": 1,
    "by_severity": {"high": 1},
    "by_kind": {"stdlib-shadowing": 1},
    "online": false
  }
}

The offline checks cover:

  • Python standard-library names published as packages
  • Known backports that should not be installed on supported Python versions
  • Dependencies without a usable version constraint
  • Repeated package entries
  • Conflicting exact pins for the same package

For package.json, Python-specific standard-library and backport checks do not apply. The doctor still checks unpinned values and repeated dependency entries.

Explain the diagnosis

Read references/dependency-pitfalls.md before presenting findings. Lead with high severity items, then medium and low. For each finding, include:

  1. Package and source line
  2. What can break
  3. The suggested fix

Do not call every range a conflict. The deterministic core reports conflicting constraints only when exact pins disagree. Compatible constraints split across multiple lines are duplicate entries that should be combined.

If there are no findings, say what was checked and note the limits. A clean report is not a CVE audit or a full dependency resolver.

Optional PyPI yank check

The online check sends package names and exact pinned versions to pypi.org. Ask for permission before enabling it, even if the user previously requested an offline diagnosis.

python3 scripts/dep_doctor.py /path/to/requirements.txt --json --online

It reports an exact Python release only when every file for that release is marked yanked. Network failures become low-severity findings instead of hiding the offline diagnosis.

Offer fixes, do not apply them silently

After explaining the report, offer a focused edit. Wait for approval before changing the manifest.

  • Remove standard-library packages from supported Python projects
  • Remove obsolete backports, or add a Python-version marker when an old runtime genuinely needs one
  • For an unpinned dependency, inspect the working environment's installed version, confirm it is intended, and propose an exact reviewed pin
  • Keep one entry for duplicates and combine compatible constraints
  • For conflicting exact pins, inspect dependents before choosing a version
  • Replace a yanked pin with a tested, non-yanked release

After any approved edit, rerun the offline diagnosis and the project's existing install or test command. Do not introduce a new CI gate.

Files

  • scripts/dep_doctor.py: stdlib-only manifest parser and diagnosis engine
  • references/dependency-pitfalls.md: reasoning guide for the reported risks

04 同仓库更多 Skills

WhiteChronos
WhiteChronos/ChatGPT

typesafe-ai

Build AI-powered software with TypeSafe: small units of AI intelligence you can use like programming primitives. Its System One models, including Jev, turn natural language and application state into typed judgments and probabilities that code can combine. Use when a feature needs programmable common sense, when brainstorming what AI could make possible in an app, or when an LLM prompt-and-parse step could become a structured decision. Applications include routing, ranking, extraction, verification, and interactive experiences; these are starting points, not the limits. Read live docs and cookbooks to find useful patterns and discover new combinations.

开发工程师
★ 0
WhiteChronos
WhiteChronos/ChatGPT

thinking-out-loud

A contract for what the agent does when a long, messy, stream-of-consciousness ramble arrives (usually voice dictation): act on nothing until the echo brief is approved. The echo audits the entire transfer, mission, locked decisions and constraints, open questions, flips and parked tangents, with the model's inferences and guesses quarantined away from the user's own phrasing, so the user verifies what the model believes, not just what it doubts. Use when the user says "let me think out loud" or wants to ramble a bit before building anything, when a message opens with a speech-to-text preamble like "switching to voice, sorry for typos", when input is a long weakly punctuated stream with restarts and mid-message reversals ("actually no, wait, scrap that idea entirely"), or when the user asks to be interviewed about a fuzzy half-formed idea. Includes an optional capture mode for rambles spread across several messages and an optional targeted interview.

律师与法务
★ 0
WhiteChronos
WhiteChronos/ChatGPT

scope-creep-detector

Analyzes git diffs against a stated intent to detect scope creep, unrelated files, broad pull requests, changes that grew beyond a fix, dependency additions, public API renames, config or CI edits, oversized hunks, and formatting-only files. Use when the user asks whether a change grew beyond the fix, a PR is too broad, or what unrelated stuff they touched, and wants keep, split, or justify guidance. Operates locally and offline.

★ 0
WhiteChronos
WhiteChronos/ChatGPT

project-graveyard

Scans the developer's machine for dead side projects, autopsies each one from its git history (died at the payments wall, killed by a newer project, finished but never shipped), surfaces their personal death patterns, and picks the corpse most worth resurrecting — then helps ship it. Use when the user mentions abandoned, unfinished, or old side projects, asks "what should I finish", wants to revive or resurrect a project, says "run the graveyard", wonders why they never finish anything, or is about to start a new project that sounds like one they already built. Runs entirely locally.

★ 0
WhiteChronos
WhiteChronos/ChatGPT

commit-archaeologist

Reconstructs why code exists from local git history, including the introducing commit, later changes, current authors, repeated companion files, and likely intent. Use when the user asks "why does this code exist", "who wrote this function and why", or to "explain the history of this function" before a rewrite, refactor, or risky edit. Runs entirely locally.

★ 0
WhiteChronos
WhiteChronos/ChatGPT

advisor-orchestrator-worker

Use when a task is too large for one model pass, needs parallel research or generation across many subtasks (like researching a dozen competitors at once), or the user asks to orchestrate multiple models, split work across a model team, run an advisor-worker loop, have a stronger model review the plan while cheap workers execute, or says "too big for one model" or "fan this out". Not for single-file edits or tasks one model handles in one pass.

★ 0