Back to Skills

monitor-binary-version-baselines

pjt222
Updated 6 days ago
14 views
17
2
17
View on GitHub
Developmentaiapi

About

This skill establishes versioned baselines of CLI binary contents using categorized markers to track feature lifecycles across releases. It enables detection of dark-launched or removed capabilities through weighted scoring and threshold-based presence detection. Developers should use it when verifying scanning tool coverage or monitoring capability changes between binary versions.

Quick Install

Claude Code

Recommended
Primary
npx skills add pjt222/agent-almanac -a claude-code
Plugin CommandAlternative
/plugin add https://github.com/pjt222/agent-almanac
Git CloneAlternative
git clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/monitor-binary-version-baselines

Copy and paste this command in Claude Code to install this skill

Documentation

監二進位版基線

築持版鍵之 CLI 工具二進位中特系標之比錄,以機械跨版識增、刪、暗發之能。

  • 追特功於閉源 CLI 工具諸版之命周
  • 探暗發(出而閉)或默刪
  • 驗標掃器仍識舊二進位中之知善標(自身退化試)
  • 築一期基供二三四期(旗探、暗測、線捕)用
  • grep 答「今 X 在乎」而實需「X、Y、Z 之系跨版如何動」之諸境

  • :同 CLI 工具之一或多裝二進位(或解組)
  • :標定之工目錄檔(首行造,跨版擴)
  • :先行之基線檔(原處擴,勿重書)
  • :知未發之版列(略發、撤建)
  • :已追之特系列以擴非再尋

一:按類擇標

擇耐重建之串。取穩、語義之識——非組者下版重命之縮名。

六建類:

  • API——端徑、工具網之露法名
  • ——內品名、代碼名、版哨
  • ——用面配檔之識鍵
  • ——析管之發事名
  • ——閘謂消之特閘鍵
  • ——特處內之知串常(誤訊、記標)

避:似縮之短識(如 _a1bX、二字後數)、隨文修必變之內字、合組者內名約者。

得:各候標含類標與短理(「現於用面文」、「跨 N 先版穩」等)。典首掃得每系 20-50 標。

敗:標於連次小版消→目錄捕重建易變串非穩識。棄之;廣至更長更語義之子串。

二:按特系群標

合標為一系表每獨進之能。「系」乃同行之標群,因共特命周(如假 acme_widget_v3 能之諸標)。

何故群:每系評防交污。一系標缺不抑他系識,且無關系之聚計無告。

工目錄形(偽碼):

catalog:
  acme_widget_v3:
    markers:
      - { id: "acme_widget_v3_init",         category: function, weight: 10 }
      - { id: "acme.widget.v3.dialog.open",  category: telemetry, weight: 5 }
      - { id: "ACME_WIDGET_V3_DISABLE",      category: flag,     weight: 10 }
  acme_other_system:
    markers:
      - ...

得:各系有己標列;無標於二系。加新系即加頂入——勿溯動標於系間。

敗:標難分一系(重、含糊)→系定過粗。拆系,或受某標為「共底」而排於每系評。

三:按信強加權

各標賦權反其單在以證系:

  • 10 = 獨診——獨足,僅見此即足證系在(如長、系專之串無他碼徑發)
  • 3-5 = 唯佐——獨不足證,於聚評有貢(如工具於諸特復用之短遙後綴)

教約勿教具數。「診」「佐」之距勝具整數——關鍵為五之閾可分「一強信」與「諸弱信」。

得:各標有權。目錄之權偏佐標(3-5),各系有少獨診標(10)。

敗:諸標皆 10→評失分——部現之果不能。降跨多系或現於無關處之標。

四:記每版基線

各掃版,記標皆按版鍵。皆為證:版 N 之缺標如版 N+1 重引時亦告。

基線形:

baselines:
  "1.4.0":
    acme_widget_v3:
      present: ["acme_widget_v3_init", "ACME_WIDGET_V3_DISABLE"]
      absent:  ["acme.widget.v3.dialog.open"]
      score:   20
  "1.5.0":
    acme_widget_v3:
      present: ["acme_widget_v3_init", "ACME_WIDGET_V3_DISABLE", "acme.widget.v3.dialog.open"]
      absent:  []
      score:   25
  "1.4.1":
    _annotation: "never-published; skipped from upstream release timeline"

未發版宜明標非默略。默略似後讀者之數失。

得:各版生每追系一錄含 presentabsentscore,或未發版之明 _annotation

敗:基線掃得先在系之零標,未確二進位徑正、strings 出有、標 ID 確同前勿假刪。偽零污長期錄。

五:設全部現之閾

各系於聚評定二閘:

  • full——逾此分系於本版視為現且活
  • partial——逾此分系視為已出未全(部標現而下 full

partial = 缺(或未在,依向)。

thresholds:
  acme_widget_v3:
    full:    25
    partial: 10

擇閾:full 設健裝期發之權和;partial 設一診標加一佐信。有諸版證後重校。

得:各掃生每系標籤之果:full | partial | absentpartial 果值察——乃暗發與刪候。

敗:諸版諸系皆 partial→閾過敏(或設逾標可和)。對知善活版重校。

六:用 strings -n 8

strings 含最小長過濾為提原。-n 8 底過諸噪(短片、襯、地表雜)而不失意識,幾皆逾 8 字。

strings -n 8 path/to/binary > /tmp/binary-strings.txt

後對 /tmp/binary-strings.txt 行目錄匹(任行匹器:grep -F -f markers.txtripgrep、或小本)。

注:

  • 低底(-n 4-n 6)洪二進位垃與縮符噪;診-佐分崩
  • 高底(-n 12+)漏短旗識與配鍵
  • 某組者壓或編串;strings 近空→二進位需先解組(此技範外)

得:行串出 1k-100k 行,依二進位大。手察前 100 行可顯識識。

敗:出空或不識→二進位或包、加密、或為 strings 不能讀之碼形。停於提層治;勿由不可讀掃記基線。

七:前擴基線勿重書過往

新系或標加目錄時,唯前版為之掃。過版錄留如初書。

何故:先版基線乃彼時所掃之經驗證,非過版含何之今模。溯改之以新識標混「今知」與「彼觀」。皆有用;唯一宜於基線檔。

若真需溯掃(如試新標於版 N-3 在乎)→記為獨補

addenda:
  "1.4.0":
    scan_date: "2026-04-15"
    catalog_revision: "v7"
    findings:
      acme_new_system:
        present: ["..."]

baselines["1.4.0"] 不動。讀者可見原錄與後溯掃及各目錄修。

得:基線檔單向前增;過錄唯加含可選 addenda 塊。目錄修版以各掃可繫所用之目錄態。

敗:若欲改過版 present 列直→停。改加補。改過錄失識掃器退化之能(後掃器驗之八步倚史錄不變)。

  • 目錄各標明類標(API / identity / config / telemetry / flag / function 之一)
  • 各標賦於精一系;無標於二系
  • 權跨真範(某 10、某 3-5);非皆同
  • 各掃版有錄含每追系之 presentabsentscore
  • 未發版明標非默略
  • 各系有 fullpartial 閾;果按之標
  • strings -n 8 為提原(或非文二進位之記等)
  • 過版錄不變於最新掃;新果於溯時居 addenda

  • 記具果為目錄:目錄宜述標類形非列版固字。果形之目錄速腐且為公曝最高漏險
  • 捕縮識:如 _p3aq9X 各重建重命。今合,明日為噪。留語義識
  • 混遙事與旗:諸工具中名約共而角異。按類標(步一)以分類析淨
  • 默略未發版:版序中無注之隙似漏掃。明注:_annotation: "never-published"
  • 無基線即設閾:首掃立經驗權和;對之校閾,非預
  • 目錄擴時改先版錄:過錄為證;補乃溯掃之支模
  • 信空掃出:零標非常意「缺」。聲刪前確二進位可讀且目錄 ID 確同
  • strings -n 4-n 8:低底加噪速於信。診標幾恆 8+ 字

  • security-audit-codebase
  • audit-dependency-versions
  • probe-feature-flag-state
  • conduct-empirical-wire-capture
  • redact-for-public-disclosure

GitHub Repository

pjt222/agent-almanac
Path: i18n/wenyan-ultra/skills/monitor-binary-version-baselines
0
agentsagentskillsai-assisted-developmentclaude-codeskillsteams

Related Skills

subagent-driven-development

Development

This skill executes implementation plans by dispatching a fresh subagent for each independent task, with code review between tasks. It enables fast iteration while maintaining quality gates through this review process. Use it when working on mostly independent tasks within the same session to ensure continuous progress with built-in quality checks.

View skill

qmd

Development

qmd is a local search and indexing CLI tool that enables developers to index and search through local files using hybrid search combining BM25, vector embeddings, and reranking. It supports both command-line usage and MCP (Model Context Protocol) mode for integration with Claude. The tool uses Ollama for embeddings and stores indexes locally, making it ideal for searching documentation or codebases directly from the terminal.

View skill

mcporter

Development

The mcporter skill enables developers to manage and call Model Context Protocol (MCP) servers directly from Claude. It provides commands to list available servers, call their tools with arguments, and handle authentication and daemon lifecycle. Use this skill for integrating and testing MCP server functionality in your development workflow.

View skill

adk-deployment-specialist

Development

This skill deploys and orchestrates Vertex AI ADK agents using A2A protocol, managing AgentCard discovery, task submission, and supporting tools like Code Execution Sandbox and Memory Bank. It enables building multi-agent systems with sequential, parallel, or loop orchestration patterns in Python, Java, or Go. Use it when asked to deploy ADK agents or orchestrate agent workflows on Google Cloud.

View skill