Crossframe Notebook
sickn33/agentic-awesome-skills
A skill your agent uses when CrossFrame Suite routes explicit Chinese notes for books, theories, articles, excerpts, bidirectional reading, absorption, or conflict mapping.
规范吸纳执行 Skill — 用于检查 data 最新可吸纳项、证明通用规范价值、剔除项目独有规则、输出吸纳清单、分层决策、同步消费者和验证探针；当用户说“吸纳规范 / 最新可吸纳 / 仍需吸纳 / 检查 data 是否有规范 / 开始吸纳”时使用。
AGPL-3.0Auto-check passed
Install Spec Absorption
$ npx skills add devcodex-labs/devcodex --skill spec-absorption -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install devcodex-labs/devcodex spec-absorption --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/devcodex-labs/devcodex.git skills-src && mkdir -p .claude/skills && cp -r skills-src/content/skills/spec-absorption .claude/skills/spec-absorption && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "spec-absorption" agent skill from https://github.com/devcodex-labs/devcodex/tree/main/content/skills/spec-absorption into .claude/skills/spec-absorption/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "spec-absorption", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/devcodex-labs/devcodex/tree/main/content/skills/spec-absorptionType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add devcodex-labs/devcodex --skill spec-absorption -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install devcodex-labs/devcodex spec-absorption --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/devcodex-labs/devcodex.git skills-src && mkdir -p .agents/skills && cp -r skills-src/content/skills/spec-absorption .agents/skills/spec-absorption && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "spec-absorption" agent skill from https://github.com/devcodex-labs/devcodex/tree/main/content/skills/spec-absorption into .agents/skills/spec-absorption/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "spec-absorption", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add devcodex-labs/devcodex --skill spec-absorption -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install devcodex-labs/devcodex spec-absorption --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/devcodex-labs/devcodex.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/content/skills/spec-absorption .cursor/skills/spec-absorption && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "spec-absorption" agent skill from https://github.com/devcodex-labs/devcodex/tree/main/content/skills/spec-absorption into .cursor/skills/spec-absorption/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "spec-absorption", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/devcodex-labs/devcodex.git --path content/skills/spec-absorption--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add devcodex-labs/devcodex --skill spec-absorption -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install devcodex-labs/devcodex spec-absorption --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/devcodex-labs/devcodex.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/content/skills/spec-absorption .gemini/skills/spec-absorption && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "spec-absorption" agent skill from https://github.com/devcodex-labs/devcodex/tree/main/content/skills/spec-absorption into .gemini/skills/spec-absorption/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "spec-absorption", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install devcodex-labs/devcodex spec-absorptionInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add devcodex-labs/devcodex --skill spec-absorption -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/devcodex-labs/devcodex.git skills-src && mkdir -p .github/skills && cp -r skills-src/content/skills/spec-absorption .github/skills/spec-absorption && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "spec-absorption" agent skill from https://github.com/devcodex-labs/devcodex/tree/main/content/skills/spec-absorption into .github/skills/spec-absorption/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "spec-absorption", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add devcodex-labs/devcodex --skill spec-absorption -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install devcodex-labs/devcodex spec-absorption --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/devcodex-labs/devcodex.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/content/skills/spec-absorption .opencode/skills/spec-absorption && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "spec-absorption" agent skill from https://github.com/devcodex-labs/devcodex/tree/main/content/skills/spec-absorption into .opencode/skills/spec-absorption/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "spec-absorption", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Facts
- Skill name
spec-absorption- GitHub stars
- 439
- Token cost
- ~4.7k tokens
- SKILL.md length
- 1,216 words
- Files
- 4
- Skills in repo
- 70
- Repo updated
- First seen
- Licence
- AGPL-3.0
At a glance
规范吸纳执行 Skill — 用于检查 data 最新可吸纳项、证明通用规范价值、剔除项目独有规则、输出吸纳清单、分层决策、同步消费者和验证探针；当用户说“吸纳规范 / 最新可吸纳 / 仍需吸纳 / 检查 data 是否有规范 / 开始吸纳”时使用。
- Works in 12 steps: 输入归集：读取用户消息、相关报告、运行态台账、最新记忆和源码现状。 → 规模路由与全命名空间扫描：先调用 skill-gap-analysis 的… → SourceExistenceVerificationGate（硬前置）：对每条表… → …
- SKILL.md covers 定位, 触发语义, 执行流程 and SourceExistenceVerificationGate, plus 10 more sections
- Calls npm and node
What it does
Spec Absorption is an agent skill from devcodex-labs/devcodex. 规范吸纳执行 Skill — 用于检查 data 最新可吸纳项、证明通用规范价值、剔除项目独有规则、输出吸纳清单、分层决策、同步消费者和验证探针；当用户说“吸纳规范 / 最新可吸纳 / 仍需吸纳 / 检查 data 是否有规范 / 开始吸纳”时使用。
Its SKILL.md is about 4.7k tokens, which your agent loads only when the skill is triggered. The skill folder holds 3 other files (for example `absorption-candidate-matrix.v1.schema.json`, `intent.json` and `layered-absorption-decision.v1.schema.json`).
The repository describes itself as: Intent-driven AI coding workflow runtime for consistent context, skills, approvals, validation, and handoffs across six AI coding hosts. The licence is AGPL-3.0.
Example prompts
- “吸纳规范 / 最新可吸纳 / 仍需吸纳 / 检查 data 是否有规范 / 开始吸纳”
- “/spec-absorption”
Workflow steps
12 steps, taken from the first numbered list in SKILL.md.
- 输入归集：读取用户消息、相关报告、运行态台账、最新记忆和源码现状。
- 规模路由与全命名空间扫描：先调用 skill-gap-analysis 的 ProjectArtifactScaleRoutingGate 形成 ScaleDecisionRecord，再执行 WorkspaceDataAbsorptionScopeGate 覆盖…
- SourceExistenceVerificationGate（硬前置）：对每条表面 open 候选，在 source-root（通常 devcodex）检索同名 Gate / Skill / schema / 探针 /…
- Backlog Intake 真相复核：结合存在性结果分类为 pure-open / residual-tail / already-fixed / misclassified，非 pure-open 不得原样进入吸纳范围；already-fixed / 等价覆盖写入…
- 通用性证明：对通过存在性验证且仍为缺口的项执行 CommonNormGeneralizationGate。
- 消费者证明：对通过通用性证明的项执行 AbsorptionCandidateConsumerProofGate。
- 返工价值复核：候选声称降低返工、补复审遗漏或提升首次通过率时，执行 ReworkReductionValueGate；文本出现次数不能替代可执行 owner 和效果证据。
- 可执行吸纳与探针必要性：对拟 absorb 项执行 ExecutableAbsorptionEffectivenessGate + ProbeNecessityDecisionGate；缺执行面或「该探针却无探针计划」→ 不得进入最终 absorb 清单。
- 分层归属：形成 LayeredAbsorptionDecision，判定 global-invariant / existing-skill-subgate / new-skill-required / docs-only / case-evidence-only /…
- 阶段清单：仅输出仍需吸纳项（含探针档位与影响）+ 并列 可关账/已覆盖 清单；confirm 等待用户，Auto 持久化当前决定后继续。
- 实施同步：确认后同步 commonInstruction、Skill、promptTemplate、executionConsumer、validationProbe（按探针决策）、publicDocs、deployCopy；禁止只改 Markdown 正文。
- 验证回写：执行 targeted test、node scripts/validate.js、必要的 npm test / website / release 验证；回写 PI / PF / GAP / VL / ISSUE 状态；无探针/无消费者的不得写 absorbed。
What it can do on your machine
Read from SKILL.md and the folder at commit 1dd4525. It shows what the files ask for, not the result of running them.
Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
Shell commands in SKILL.md call:
npmnode
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md. Its commands use npm, which can reach the network depending on how they are called.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Context cost
Spec Absorption loads about 4.7k tokens when it runs. Until then it costs about 35 tokens; SKILL.md has 1,216 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
Safety
Auto-check passed
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.
SKILL.md
The full file from devcodex-labs/devcodex at commit 1dd4525, republished under its AGPL-3.0 licence (© devcodex-labs). 1,216 words, ~4,706 tokens.
Download SKILL.mdSave it as
.claude/skills/spec-absorption/SKILL.md (or your agent's skills folder). This skill also uses 3 other files; get the full folder from GitHub.- name
- spec-absorption
- description
- 规范吸纳执行 Skill — 用于检查 data 最新可吸纳项、证明通用规范价值、剔除项目独有规则、输出吸纳清单、分层决策、同步消费者和验证探针；当用户说“吸纳规范 / 最新可吸纳 / 仍需吸纳 / 检查 data 是否有规范 / 开始吸纳”时使用。
Spec Absorption Skill
定位
本 Skill 承接“规范吸纳”的执行面。spec-governance 负责 RecordRouter、台账分流、GovernanceGateRegistry 和 SCV；本 Skill 负责把 PI / PF / GAP / ISSUE / 用户确认项变成可审查、可验证、可同步的吸纳方案。
执行原则：
- 吸纳 ≠ 写进 Skill/台账关闭。若吸纳后仍可「只写正文、不挂消费者、不按流程执行」，该吸纳 无效，不得标 absorbed/closed。
- 先 source-root 存在性验证，再谈可吸纳；禁止把台账
open / pending absorption / active字面状态直接写成「可吸纳清单」。 - 每条 absorb 必须绑定执行面：Owner Skill/prompt/runtime + 验证路线；并完成
ProbeNecessityDecision（必须探针 / 结构门禁 / 清单即可 / 禁止探针）。 - 先证明是通用规范价值，再决定是否吸纳；不得把业务项目独有规则包装成 DevCodex 通用规范。
- 先证明 DevCodex 当前消费者存在，再进入规范源修改；不得只因为某个项目出现过一次就改通用规范。
- 先分层归属，再写实现；不得把成组能力默认追加到通用 instructions、
CrossProjectLearnedGuards或 prompts 长清单。 - 先输出最终待确认清单（含探针决策与执行影响）；用户确认后再实施、验证、回写台账和同步部署副本。
触发语义
| 用户语义 | 必做 |
|---|---|
| “检查 data 是否有可以吸纳的” | 扫描 .devcodex/*/data/ 全命名空间，对每条候选先跑 SourceExistenceVerificationGate，再输出候选、通用性判断、跳过理由与可关账项 |
| “列出仍需吸纳清单” | 执行遗漏过滤，排除已吸纳、已关闭、source 已覆盖（ledger-stale）、重复和项目独有项，只列最终仍需吸纳 |
| “开始吸纳 / 确认执行” | 执行分层决策、消费者同步、验证探针、台账回写、报告和记忆 |
| “这个也要记录 / 以后要这样” | 先交给 spec-governance 做 RecordRouter，再由本 Skill 判断是否进入吸纳 |
| “之前都吸纳到通用规范了，重新分层” | 叠加 HistoricalCommonNormLayeringGate，先建逐文件矩阵 |
执行流程
- 输入归集：读取用户消息、相关报告、运行态台账、最新记忆和源码现状。
- 规模路由与全命名空间扫描：先调用
skill-gap-analysis的ProjectArtifactScaleRoutingGate形成ScaleDecisionRecord，再执行WorkspaceDataAbsorptionScopeGate覆盖.devcodex/*/data/；大语料必须分批/checkpoint，错误 glob、超时和派生产物污染结果标 invalid/discarded。 - SourceExistenceVerificationGate（硬前置）：对每条表面 open 候选，在 source-root（通常
devcodex）检索同名 Gate / Skill / schema / 探针 / 测试入口；未完成存在性验证的候选不得进入「最终可吸纳清单」。 - Backlog Intake 真相复核：结合存在性结果分类为
pure-open / residual-tail / already-fixed / misclassified，非pure-open不得原样进入吸纳范围；already-fixed/ 等价覆盖写入 可关账清单。 - 通用性证明：对通过存在性验证且仍为缺口的项执行
CommonNormGeneralizationGate。 - 消费者证明：对通过通用性证明的项执行
AbsorptionCandidateConsumerProofGate。 - 返工价值复核：候选声称降低返工、补复审遗漏或提升首次通过率时，执行
ReworkReductionValueGate；文本出现次数不能替代可执行 owner 和效果证据。 - 可执行吸纳与探针必要性：对拟 absorb 项执行
ExecutableAbsorptionEffectivenessGate+ProbeNecessityDecisionGate；缺执行面或「该探针却无探针计划」→ 不得进入最终 absorb 清单。 - 分层归属：形成
LayeredAbsorptionDecision，判定global-invariant / existing-skill-subgate / new-skill-required / docs-only / case-evidence-only / project-local / already-covered。 - 阶段清单：仅输出仍需吸纳项（含探针档位与影响）+ 并列 可关账/已覆盖 清单；confirm 等待用户，Auto 持久化当前决定后继续。
- 实施同步：确认后同步 commonInstruction、Skill、promptTemplate、executionConsumer、validationProbe（按探针决策）、publicDocs、deployCopy；禁止只改 Markdown 正文。
- 验证回写：执行 targeted test、
node scripts/validate.js、必要的npm test/ website / release 验证；回写 PI / PF / GAP / VL / ISSUE 状态；无探针/无消费者的不得写 absorbed。 - 报告记忆：报告必须引用候选矩阵、SourceExistence、ProbeNecessity、LayeredAbsorptionDecision、验证证据、台账状态和部署副本同步。
SourceExistenceVerificationGate
🔴 防复发（2026-07-26）：仅扫台账状态输出「可吸纳」属于控制失效。最终确认清单中每条
commonDecision=absorb必须有本门禁证据；existenceStatus=present|equivalent-covered时只能already-covered/already-fixed并进可关账清单，禁止再标 absorb。
每个候选在进入通用性证明 / 最终确认清单前必须填写：
| 字段 | 说明 |
|---|---|
candidateId | PI / PF / GAP / ISSUE 编号 |
claimedCapability | 候选声称要新增的能力名（Gate / Skill / 矩阵 / 探针） |
searchAnchors | 在 source-root 检索的关键词/路径（≥1；含同名与已知近邻等价名） |
sourceRoot | 规范源码根，例如 E:/Worker/devcodex 或仓库相对路径 |
existenceStatus | absent / partial / present / equivalent-covered / unverified |
hitEvidence | 命中文件路径与符号（可空数组，但 present/partial/equivalent-covered 时至少 1 条） |
nearNeighborCoverage | 近邻已有能力说明；无则 none |
ledgerDisposition | absorb-candidate / residual-candidate / close-ledger / defer |
verifiedBy | 检索方式 + 可选生产入口测试命令与 exitCode |
判定规则：
existenceStatus=unverified或缺少searchAnchors/sourceRoot→ 阻断，不得输出为可吸纳。present或equivalent-covered→backlogClass=already-fixed（或commonDecision=already-covered），进入 可关账清单，禁止 absorb。partial→ 最多residual-tail；最终 absorb 只能是「补洞切片」，必须写清已有近邻与仍缺字段。absent且检索覆盖 claimedCapability 与主要近邻名 → 才可继续CommonNormGeneralizationGate并竞逐pure-open。- 台账写
pending absorption但 source 已实现 → 记 ledger-stale，归可关账，不得重复实施。
负向样例
| 负向样例 | 处理 |
|---|---|
只根据 状态：open / pending absorption 列入可吸纳 | invalid：ledger-status-only |
| 未在 source-root 检索同名 Gate 即标 pure-open absorb | invalid：existence-unverified |
source 已有 EvidenceFreshnessReceiptV1 + 测试绿，台账仍 residual open 又整包吸纳 | invalid：应 close-ledger |
| 有近邻 TTFV 却把 ProgressReportFastPath 写成「已存在」而不写 partial/absent | invalid：须区分父门禁与特化缺口 |
CommonNormGeneralizationGate
每个候选吸纳前必须先填写：
| 字段 | 说明 |
|---|---|
candidateId | PI / PF / GAP / ISSUE / 用户确认项编号或消息锚点 |
sourceNamespace | 来源命名空间，例如 .devcodex/<project>/data |
candidateScope | 原始问题影响范围：单项目 / 多项目 / DevCodex 控制面 / 宿主通用 |
generalizationEvidence | 可泛化证据：至少包含跨工作流复用、宿主无关性、DevCodex 当前消费者或同类问题重复证据之一 |
projectSpecificResidue | 项目私有路径、服务名、业务名词、数据库/接口/路由局部前提、私有组织流程 |
negativeExamples | 明确不吸纳的反例，防止下次误收 |
commonTrigger | 抽象后的通用触发语义 |
targetConsumer | DevCodex 中真实会执行该规范的 Skill / prompt / validate / report / docs |
decision | absorb / case-evidence-only / project-local / docs-only / already-covered / reject |
判定规则：
- 只有
generalizationEvidence成立且targetConsumer存在，才可进入absorb。 - 存在项目私有残留时，必须先抽象成宿主/流程/消费者无关表达；抽象失败则判
project-local。 - 只是某个项目的目录、服务、接口、业务角色或部署习惯，不得吸纳为 DevCodex 通用规范。
- 已有同等或更强规范、Skill、prompt 和探针覆盖时，判
already-covered，只回写台账，不重复新增。
负向样例
| 负向样例 | 处理 |
|---|---|
ServiceSpecRead、docs/services/<name>/ 服务开发读取链 | 项目或框架私有，不吸纳为 DevCodex 通用规范；可作为“先证明通用价值”的反例 |
| 单个业务项目的 service / route / model / schema 命名 | project-local，除非证明 DevCodex 多项目 Profile/服务规范消费者需要 |
单个库的 cacheControl、adapter 配置或返回值习惯 | case-evidence-only 或并入对应库本地规范 |
| 某个项目私有数据目录、导航层级、脚本名 | 不直接进入通用层；只能抽象为“消费者同步 / 产物边界 / 验证链”类规则 |
AbsorptionCandidateConsumerProofGate
通过通用性证明后，还必须证明 DevCodex 当前消费面：
| 字段 | 说明 |
|---|---|
devcodexConsumerEvidence | 当前仓库中会读取或执行该规则的文件和段落 |
targetOwner | 目标 owner：existing Skill、new Skill、instruction、prompt、script、README、website、Profile、部署副本 |
layerChecks | commonInstruction / skill / promptTemplate / executionConsumer / validationProbe / publicDocs / deployCopy |
validationRoute | targeted test、validate 编号、SCV、npm test、website build、manual evidence |
skipReason | 任一层 N/A 的理由 |
消费者证明失败时，不得实施为 active 规范；最多进入 PF / ISSUE，或作为 case evidence 保留。
ExecutableAbsorptionEffectivenessGate
🔴 核心不变量：规范写进 Skill 但流程上仍可不执行 = 假吸纳。用户确认 absorb 前，每条候选必须证明「不按流程执行时会被发现或阻断」，而不是只证明「文档里有这句话」。
什么叫「有效吸纳」
| 条件 | 必须满足 |
|---|---|
| 触发面 | 有明确触发语义（意图/阶段/产物类型），命中时必须加载 Owner |
| 执行面 | 有 Skill 子门禁 / Hook / CLI / planner 之一真正执行检查 |
| 验证面 | 有 validationRoute 生产入口；按 ProbeNecessityDecision 挂探针或结构门禁 |
| 失败语义 | 违反时有 fail / block / WARN 降级之一，不能静默忽略 |
| 关闭语义 | 只有执行面+验证面绿才允许台账 absorbed/closed |
什么叫「无效吸纳」（禁止标 absorbed）
| 反模式 | 说明 |
|---|---|
| text-only | 只改 SKILL.md / instructions 段落，无消费者路径 |
| checklist-only-as-absorb | 仅靠人工自觉勾选，无结构字段、无样例探针、无 fail 路径 |
| absorb-without-process | 宣称已吸纳，但同类任务仍可不跑门禁完成 |
| probe-theater | 有测试文件但不断言本规则的负向逃逸 |
| ledger-close-without-enforcement | 台账关了，runtime/Skill 仍不执行 |
判定：enforcementLevel 必须为下列之一，且与探针决策一致：
| enforcementLevel | 含义 | 可否 absorb 为 active |
|---|---|---|
hard-probe | 机器探针/schema 负向可阻断或测试红 | 是 |
structural-gate | 缺字段/矩阵 blocked（planner/candidate-review/CP） | 是 |
conditional-probe | 仅命中场景挂探针；未命中 N/A+skipReason | 是（须写触发条件） |
checklist-only | 仅清单提醒 | 否（active）；最多 docs-only / gray + 明示无效于防复发 |
none | 无执行 | 否 |
ProbeNecessityDecisionGate
每条 commonDecision=absorb 必须填写 ProbeNecessityDecision（可嵌入矩阵 probeNecessity 字段）：
| 字段 | 说明 |
|---|---|
candidateId | 编号 |
probeClass | machine-sample / structural-schema / fixture-replay / extend-existing / checklist-only / probe-forbidden |
necessity | required / conditional / not-required / forbidden |
rationale | 为何必须/不必/禁止探针（假绿成本、可机检性、频率） |
probePlan | 新建 classifier / 扩展现有探针 / 结构 required 字段 / N/A |
existingProbeReuse | 可复用的探针名；无则 none |
alwaysOnImpact | none / test-only / conditional-path / always-on-path |
complexityDelta | 维护成本：低/中/高 + 一句话 |
falsePositiveRisk | 误报风险与缓解 |
skipProbeReason | `necessity=not-required |
哪些 必须 有探针（necessity=required 或 conditional）
同时满足越多，越应 required：
- 假绿代价高：会误导完成态、发布、关账、用户可见进度/阶段
- 可机检：可用样例文本、JSON 字段、路径、exitCode 判定，不靠「是否好懂」主观分
- 重复发生或跨任务：≥2 次独立上下文或明确跨项目
- 违反后应失败：应用 fail/block，而不是仅建议
| 典型必须 | 探针形态 |
|---|---|
| 进度/TTFV 类执行逃逸 | 扩展 classifyTtfv* 或专用 progress 样例 |
| 多任务阶段/sourceDelivery 误投影 | 最终回复/进度卡 sample classifier |
| closed 产物复活 | fixture 头状态 + 拟写路径 |
| 强主张/完成态假绿 | 既有 MeasuredVerification / FinalValidation 类 |
| 分类/分层字段缺失导致假完成 | structural schema required |
conditional：仅特定工作流命中（如 consumer-validation formal、GlobalHost CP2）时强制；其它任务 N/A + skipReason。
哪些 没必要 机器探针（necessity=not-required）
| 类型 | 处理 | 仍须有的执行面 |
|---|---|---|
| 纯文档口吻/IA 文案 | docs-only 或 checklist | document-sync / 受众检查即可 |
| 高度主观质量（「是否足够低心智」全文） | 结构矩阵字段 required，禁止 LLM-as-judge 硬探针 | CP1 缺 FirstSuccessfulRequest 等字段 blocked |
| 极低频且人工确认更安全 | checklist + 用户确认 | 报告写 skipProbeReason |
| 已有更强探针完全覆盖 | already-covered / extend-existing | 复用，不新建 |
哪些 禁止 乱加探针（necessity=forbidden）
| 禁止 | 原因 |
|---|---|
| always-on 全任务扫全库 | 抬高 token/墙钟，违反 TTFV / 基座中性 |
| 无负向样例的「探针」 | probe-theater，制造假安全感 |
| 把项目私有路径写进通用探针 | 泄漏 project-local |
| 与现有 classifier 重复且更弱 | 规范税 + 漂移 |
添加探针的影响（实施前必须写清）
| alwaysOnImpact | 影响 | 默认允许 |
|---|---|---|
none | 仅文档 | 不单独作为 absorb 关闭证据 |
test-only | 只进 npm test / targeted，不改默认对话路径 | 推荐 |
conditional-path | 仅命中意图/Skill 时加载 | 推荐 |
always-on-path | 每轮或每宿主默认执行 | 须 base-changing 或明确 base-compatible + UnaffectedIntentRegression；单独确认 |
复杂度默认约束：
- 新建 Gate 家族 > 扩展现有 classifier > 结构字段 required（优先后者）
complexityDelta=高且无replacementOrRetirementCredit→ 不得 ready- 探针必须至少 1 正 + 1 负样例；disabled/misconfigured 在 base-compatible 及以上必填
负向样例
| 负向 | 处理 |
|---|---|
| 清单写 absorb，probeNecessity 缺失 | blocked |
| necessity=required 但 probePlan=「以后再说」 | blocked |
| checklist-only 标 active absorbed | invalid 假吸纳 |
| always-on 新探针无 UnaffectedIntentRegression | blocked |
| 只同步 Skill 正文，validationProbe 层 N/A 且无 skipReason | layer blocked |
Show full SKILL.md (472 more words)Show less
BaseImpactAssessmentV1 / ComplexityDeltaBudgetV1
新增或晋级规范、Skill、Prompt、流程、验证器或部署消费者时，必须把基座影响作为吸纳决策的子记录，而不是另建平行“基座治理”入口。最小字段：
changeId / servedIntent / currentGap / absorptionDecision / baseClass(base-neutral|base-compatible|base-changing) / affectedContracts / unaffectedIntents / consumers / fanout / defaultPathDelta / fallbackBehavior / migration / rollback / positiveProbe / negativeProbe / disabledOrMisconfiguredProbe / complexityDelta / replacementOrRetirementCredit / owner / reviewAt / deprecationAndDeletionCondition。
判定规则：
base-neutral：不改变常驻路径、默认 Context、强制阶段、用户确认数、公共契约或未命中任务；仍需至少一个真实消费者和UnaffectedIntentRegression。base-compatible：改变局部消费者或验证路线，但默认行为、fallback 与未受影响意图保持兼容；必须有正向、负向和 disabled/misconfigured 探针。base-changing：改变事实基座、always-on 路径、公共契约或默认强制行为；必须单独确认、写迁移/回滚和全面复审，不能由普通 auto 授权夹带通过。- 缺真实消费者、缺回滚、缺退役/删除条件、缺
ComplexityDeltaBudgetV1或没有replacementOrRetirementCredit抵消维护成本时，不得进入 active。
UnaffectedIntentRegression 至少覆盖一个普通 chat / 普通 dev / 低风险 fix 或其他未命中意图样本，证明新增能力不会让无关任务增加默认读取、默认确认、默认产物或默认验证成本。
ReworkReductionValueGate
涉及“减少返工、避免复审再发现、提升一次通过率”的吸纳候选必须填写：
| 字段 | 说明 |
|---|---|
reworkCluster | 可重复问题簇及 WorkUnit 边界，不得只写单个案例 |
currentDetectionPhase / targetDetectionPhase | 当前发现阶段与希望前移到的阶段 |
frequencySeverityLateCost | 频率、严重度、晚发现成本和可预防性证据 |
executableOwnerLayer | 能真正执行该预防动作的 Skill / prompt / runtime / probe / checklist owner |
successMetric | FirstPassYield、WorkUnitReworkRate、RepeatEscapeRate、PreventionHitRate 或等价指标 |
trialWindow | 前瞻验证的可比 WorkUnit / 独立上下文、观察周期和退出条件 |
overheadAndFalsePositiveCost | 新规则的执行成本、误报、重复检查和认知负担 |
rollbackOrSunset | 无效、有害、长期未命中或被更强 owner 替代时的处置 |
判定规则：已有规则反复出现但只有文档表述、没有执行消费者、负向探针或前瞻验证，属于 text-only recurrence，不得判定已吸纳有效。候选可以先进入 gray，但 ordinary active / 宣告 closed 前必须由 ReworkEffectivenessLoop 给出前瞻证据；不足时保持 insufficient-evidence。
分层决策
LayeredAbsorptionDecision 必须包含 candidateId / classification / targetSkill / triggerTerms / ownedArtifacts / layerChecks / validationRoute / consumerSync。本 Skill 在 Skill 层还要写 SkillAbsorptionDecision 兼容字段。
当候选会新增或改变 Rule/Skill、Prompt、MCP Resource/Resource Template/Tool、Task 增强 Tool、CLI 或 Hook 能力面时，LayeredAbsorptionDecision 只负责判定吸纳层级，不得顺带决定能力面。实施前必须消费由 spec-governance 的 CapabilitySurfaceDecisionGate 生成且通过新鲜度校验的 decisionRef；只有 spec-governance 可写 canonical decision，目标 Skill 仅保存本域触发、Owner、消费者和 decisionRef 等本地元数据。缺少 decision、identity 失效或状态为 stale/blocked 时停止能力资产创建并回到中央判定。
| classification | 使用条件 | 处理 |
|---|---|---|
global-invariant | 安全底线、入口加载、全模式硬约束、优先级 | 通用 instructions 保留正文，目标 Skill 引用 |
existing-skill-subgate | 有明确既有 Skill owner | 并入 Skill 子门禁，同步消费者和探针 |
new-skill-required | 具备独立触发、产物、状态、模板或 3 条以上子规则 | 同批创建 Skill 或写 PF / ISSUE |
docs-only | 只影响说明、历史镜像或用户文档 | 同步公开文档，不做执行门禁 |
case-evidence-only | 只能作为案例证据支持已有 Gate | 写报告/台账，不进入规范正文 |
project-local | 只适合来源项目 | 回写来源项目台账或 Profile，不吸纳到 DevCodex 通用 |
already-covered | 现有规范同等或更强覆盖 | 回写状态和证据，不重复实现 |
条件通用规范的 registry 消费
候选通过 CommonNormGeneralizationGate 与 AbsorptionCandidateConsumerProofGate 后，不在本 Skill 维护版本批次或 Gate 总表。执行方必须读取 ../spec-governance/gate-registry.json，按候选所属 gateGroup 获取触发条件、Owner Skill、证据要求、消费者与兼容锚点；来源仍含项目私有残留时先抽象再实施。
| 候选域 | registry 分组 | 主要承接面 |
|---|---|---|
| 配置、Profile、宿主与路由 | profile-service、absorption-layering | Profile/runtime 真相、部署副本与分层决策 |
| 行为语义、文档与示例 | docs-semantics-examples、docs-audience-render | Owner Skill、用户文档消费者与真实性探针 |
| 派生消费者、失败隔离与副通道 | derived-consumer-runtime、frontend-runtime | 运行时隔离、故障注入与 TestRoute |
| 功能清单、批次证据与返工闭环 | feature-inventory-batch-evidence、rework-governance | EvidenceLedger、进度卡与有效性验证 |
历史 A1~A10 名称只作为 registry 的 legacyAnchors 兼容检索入口，不再复制执行正文。新增或调整 Gate 必须先更新结构化 registry 与目标 Owner，再同步 Prompt、执行消费者、验证探针、公开文档和部署副本；若任一层不适用，记录 skipReason。
输出产物
AbsorptionCandidateMatrixV1
当扫描结果准备进入确认清单或实施批次时，必须先形成 AbsorptionCandidateMatrixV1，字段以 absorption-candidate-matrix.v1.schema.json 为准。矩阵至少包含：
| 字段 | 说明 |
|---|---|
candidateId | PI / PF / GAP / ISSUE / 用户确认项编号 |
sourceNamespace | 来源 active-root 或 workspace data namespace |
backlogClass | pure-open / residual-tail / already-fixed / misclassified |
commonDecision | absorb / case-evidence-only / project-local / docs-only / already-covered / reject / defer |
sourceExistence | SourceExistenceVerificationGate 记录；commonDecision=absorb 时必填且不得为 present/equivalent-covered/unverified |
probeNecessity | ProbeNecessityDecisionGate 记录；commonDecision=absorb 时必填 |
enforcementLevel | hard-probe / structural-gate / conditional-probe / checklist-only / none；absorb 不得为后两者 |
targetOwner | 真实执行 owner，吸纳项缺失 owner 时阻断 |
layerChecks | commonInstruction / skill / promptTemplate / executionConsumer / validationProbe / publicDocs / deployCopy |
validationRoute | 生产验证入口，不得只写“人工复审” |
prevention | 涉及防复发时记录根因、控制失效、负向样本与 rollback/sunset |
维护者可用只读 planner 生成分层计划：
node scripts/plan-absorption-candidates.js --input <matrix.json>
node scripts/plan-absorption-candidates.js --self-test该 planner 只读取输入并向 stdout 输出 AbsorptionCandidatePlanV1；禁止写台账、修改 data、自动关闭 PI/PF 或隐式更新规范源。缺 targetOwner、非 pure-open 却试图吸纳、任一必需 layer blocked 或矩阵 schema 无效时，计划必须停在 blocked/invalid。
LayeredAbsorptionDecisionV1
每个进入实施的候选必须产出 LayeredAbsorptionDecisionV1，字段以 layered-absorption-decision.v1.schema.json 为准。该决策用于把自由文本 finding 固定到目标层级和消费者，而不是用报告段落替代执行证据。
| status | 进入条件 | 后续 |
|---|---|---|
ready | `backlogClass=pure-open | residual-tail、commonDecision=absorb`、存在性合格、probeNecessity 齐全且 enforcementLevel 非 checklist-only/none、owner/layer/test/docs/deploy 证据齐全 |
blocked | owner 缺失、layer blocked、验证路线缺失、存在性不合格、缺探针决策/执行面无效、或 schema 无效 | 修正矩阵或回 CP |
skipped | project-local / already-covered / docs-only / case-evidence-only / reject / defer | 回写 skipReason，不改通用规范 |
防复发闭环只允许声明当前关闭证据；长期有效性必须进入 repair-prevention-assessment 的 prospective plan。禁止把本轮 planner self-test 通过写成长期 prevention 已有效。
候选矩阵
| candidateId | sourceNamespace | rawSummary | backlogClass | commonDecision | targetOwner | layerChecks | validationRoute | skipReason |
|-------------|-----------------|------------|---------------|----------------|-------------|-------------|-----------------|------------|最终确认清单
只列同时满足：commonDecision=absorb、SourceExistence 合格、ExecutableAbsorption + ProbeNecessity 合格、尚未完整吸纳：
| ID | 待吸纳规范 | existence | enforcementLevel | probeClass / necessity | alwaysOnImpact | 目标 owner | 验证路线 |
|----|------------|-----------|------------------|------------------------|----------------|------------|----------|确认清单中每条必须能回答：若不按流程执行，谁会红/阻断？ 答不出则不得列入。
可关账清单（并列输出）
| ID | 台账状态 | source 证据 | disposition |
|----|----------|-------------|------------|实施报告字段
报告至少包含：
WorkspaceDataAbsorptionScopeGate扫描范围。ProjectArtifactScaleRoutingGate的项目/root、六项规模指标、决策、排除策略、batch/checkpoint 与 invalid-run 证据。SourceExistenceVerificationGate全候选结果（含可关账）。Backlog Intake分类与范围缩减。CommonNormGeneralizationGate与AbsorptionCandidateConsumerProofGate证据。LayeredAbsorptionDecision与SkillAbsorptionDecision。ReworkReductionValueGate的问题簇、阶段前移、owner、成功指标、试运行、成本和 rollback/sunset；未触发写N/A + skipReason。ConceptSyncMap或同步消费者清单。- targeted test /
node scripts/validate.js/npm test结果。 - 台账状态回写与 active-root 归属。
验证路线
规范吸纳实施完成后至少执行（MeasuredVerificationStandard：宣称通过须记录生产入口命令与 exitCode；隔离 harness 不得冒充 V# 成败）：
node scripts/test-spec-governance.jsnode scripts/validate.js或npm run test:core（含 V84ExpertOutputQualityGate/ expert-output-quality 同步时）- 高风险控制面或 Skill/部署副本变化时执行项目
npm test - 若改 README / website / changelog / Profile，执行引用扫描和 V19 资产计数校验
- 若改用户级全局部署副本，执行
devcodex global-adapters apply（源码）或npm install -g ./ pack+tarball /npm update -g（按安装来源）；workspace 运行态才用 baredevcodex update。然后再次运行 validate - 命中示例/文档/fixture 专家质量时叠加
expert-output-qualityOwner 与 gate-registryexpert-output-quality
验证失败时先修复，再更新复审清单状态；不得只在报告中写“后续处理”。
© devcodex-labs, AGPL-3.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
Files
SKILL.md and 3 other files in content/skills/spec-absorption of devcodex-labs/devcodex.
- SKILL.md
- absorption-candidate-matrix.v1.schema.json
- intent.json
- layered-absorption-decision.v1.schema.json
Open the folder on GitHubat commit 1dd4525
Compare with similar skills
Spec Absorption next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Spec Absorption this skilldevcodex-labs/devcodex | 439 | — | ~4.7k | Automated safety check: Pass | AGPL-3.0 | |
| Crossframe Notebooksickn33/agentic-awesome-skills | 47k | 1 repos | ~1k | Automated safety check: Pass | MIT | |
| Adme Property PredictorLeoYeAI/openclaw-master-skills | 2.2k | — | ~5.8k | Automated safety check: Notes | MIT | |
| Cost Accountingcbrock84/headcount | 2k | — | ~1.1k | Automated safety check: Pass | MIT | |
| Aeon Unlock MonitorBankrBot/skills | 1.2k | — | ~633 | Automated safety check: Pass | None | |
| ADMET Prediction for Drug CandidatesGPTomics/bioSkills | 1.2k | 1 repos | ~5k | Automated safety check: Pass | MIT |
Similar skills
Adme Property Predictor
LeoYeAI/openclaw-master-skills
Predict ADME (Absorption, Distribution, Metabolism, Excretion) properties for drug candidates using cheminformatics models and molecular descriptors.
Cost Accounting
cbrock84/headcount
Establishes what something actually costs — fixed, variable and mixed cost behavior, absorption versus variable costing, job-order, process and activity-based methods, breakeven and operating…
Aeon Unlock Monitor
BankrBot/skills
Weekly token unlock tracker ranked by Absorption Ratio (unlockusd / 7d avg daily volume), not the weak "% of circulating supply" proxy.
ADMET Prediction for Drug Candidates
GPTomics/bioSkills
Predicts absorption, distribution, metabolism, excretion and toxicity for drug candidates with ADMETlab 3.0, ADMET-AI, DeepChem and chemprop, plus druglikeness filters.
Unlock Monitor
aeonfun/aeon
Token unlock and vesting tracker - quantify supply pressure via absorption ratio, classify cliff vs linear, and deliver one-line market reads
More from devcodex-labs/devcodex
All 70 skills in this repoAccessibility I18n
devcodex-labs/devcodex
无障碍与国际化专家 Owner — 当任务涉及可访问性、键盘操作、焦点、屏幕阅读器、ARIA、语言地区、本地化、RTL、翻译资源、用户可见文案或多语言文档时使用；要求把包容性体验和本地化验证绑定到真实用户路径。
AI Agent System Architecture
devcodex-labs/devcodex
AI Agent 系统架构专家 Owner — 当任务涉及 Agent 路由、工具调用、上下文管理、记忆、状态机、权限、人机协作、可观测性、回放验证或模型辅助治理时使用；要求把 Agent 行为设计成可解释、可恢复、可审计。
API Contract Architecture
devcodex-labs/devcodex
API 契约架构专家 Owner — 当任务涉及 public API、HTTP/SDK/CLI 契约、版本兼容、错误模型、分页过滤、幂等、Schema、类型、迁移或消费者影响时使用；要求先冻结消费者契约，再设计实现与验证。
Architecture Design
devcodex-labs/devcodex
架构设计文档编排 Owner — 当用户要求架构设计、系统设计、技术架构或可指导开发、Review 与任务拆分的完整方案时使用；要求从业务流程反推节点、状态、数据、一致性、异常补偿、ADR 与实施任务。
Audit Common
devcodex-labs/devcodex
审查公共维度 G0~G5 + Profile Freshness Check — 所有 audit 子类型必先执行的基础维度层
Audit Session
devcodex-labs/devcodex
审计工作流的跨会话状态机 — 在 <audit-root/.audit-state/<session-id.json 持久化轮次/发现项/收敛状态，支持 Token 中断后精准恢复
Questions about Spec Absorption
What does Spec Absorption do?
规范吸纳执行 Skill — 用于检查 data 最新可吸纳项、证明通用规范价值、剔除项目独有规则、输出吸纳清单、分层决策、同步消费者和验证探针；当用户说“吸纳规范 / 最新可吸纳 / 仍需吸纳 / 检查 data 是否有规范 / 开始吸纳”时使用。. Spec Absorption is an agent skill from devcodex-labs/devcodex.
How do I install Spec Absorption in Claude Code?
Run `npx skills add devcodex-labs/devcodex --skill spec-absorption -a claude-code`. Or copy the skill folder (content/skills/spec-absorption in devcodex-labs/devcodex) into .claude/skills/spec-absorption in your project. Claude Code loads it when a task matches its description.
How do I install Spec Absorption in Codex?
Run `npx skills add devcodex-labs/devcodex --skill spec-absorption -a codex`. Or copy the skill folder (content/skills/spec-absorption in devcodex-labs/devcodex) into .agents/skills/spec-absorption in your project. Codex loads it when a task matches its description.
Can I use Spec Absorption in Cursor, Gemini CLI or GitHub Copilot?
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add devcodex-labs/devcodex --skill spec-absorption -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/spec-absorption, .gemini/skills/spec-absorption, .github/skills/spec-absorption and .opencode/skills/spec-absorption in your project.
What does Spec Absorption need to run?
Going by SKILL.md and its folder, Spec Absorption needs the command-line tools its instructions call (npm and node).
Does Spec Absorption access the network?
SKILL.md contains no URLs. Its commands use npm, which can reach the network depending on how they are called. This is read from the text; nothing was executed.
Is Spec Absorption safe to install?
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.
What licence does Spec Absorption use?
Spec Absorption is published under the AGPL-3.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
How many tokens does Spec Absorption use?
About 4.7k tokens (SKILL.md is roughly 19k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.
What are the alternatives to Spec Absorption?
Skills that share tags, products or a category with Spec Absorption: Crossframe Notebook (sickn33/agentic-awesome-skills, 47k stars), Adme Property Predictor (LeoYeAI/openclaw-master-skills, 2.2k stars), Cost Accounting (cbrock84/headcount, 2k stars) and Aeon Unlock Monitor (BankrBot/skills, 1.2k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains Spec Absorption?
devcodex-labs (a GitHub organization) maintains it in devcodex-labs/devcodex, which has 439 GitHub stars. The repository holds 70 skills in this directory. The repository was last updated on September 17, 2026.
Source: devcodex-labs/devcodex on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.