跳到正文
cccskills
免费在 GitHub 公开

hermes-ephemeral-delegation

Coordinate temporary agents for code exploration, independent implementation tasks, verification, reviews, or complex debugging. Define self-contained assignments, run independent work in parallel, and check returned evidence before combining the results.

中文简介正在准备中。当前显示英语简介。

原文查看原文说明

Trigger: a mapping need, parallel units, context backstop, high-risk verify, fresh review, or multi-step debug. Delegate via delegate_task for a named reason.

查看安装方法

适用场景

  • Delegating codebase exploration to inform a decision
  • Running independent implementation tasks in parallel
  • Getting an independent check of a high-risk change
  • Delegating multi-step debugging

包含的文件(2)

  • SKILL.md4.5 KB
  • references/tuning-knobs.md1.3 KB

SKILL.md(原文)

安装前可以查看交给 AI 代理的指令内容。

Activation Contract

Load this skill when you are acting as the parent orchestrator and the work ahead has a named reason to delegate:

  • A map you need to decide or route that exceeds the inline evidence budget (codebase mapping, approach comparison); never to read files you will read anyway before writing inline
  • A writer reason: 2+ independent units of behavior (code, tests, and docs together; each several minutes of work) launched together, or the context backstop; a large task alone is tracked, not delegated
  • A high-risk change that needs an independent verifier, or a large task's long suites and builds
  • Fresh adversarial review (diffs, PR readiness, incident audit); never restart it automatically for a candidate whose terminal approved authority was already burned and that is unchanged since, unless a new independent review is deliberately requested
  • Multi-step debugging that would flood the parent context

Do NOT load this skill if you are already inside a delegated child task — you are the executor, not the orchestrator.

Hard Rules

  • Use delegate_task only for a named reason listed above. Without one, work inline following your logbook; file count, size, or a large task alone is never a reason.
  • Workers are EPHEMERAL: each delegate_task call creates a fresh context. Do NOT request persistent agent files or profiles.
  • Pass a self-contained mission. Workers have no memory of the parent conversation.
  • Treat worker output as self-report: verify file writes, test pass/fail, URLs, and IDs before reporting success to the user.
  • Batch parallel calls only for INDEPENDENT workstreams. Sequential dependencies must run sequentially.

Decision Gates

SituationAction
Need a map to decide or route beyond the inline evidence budgetDelegate a narrow exploration worker
A large (tracked) taskTrack it; delegate a writer only for a reason (parallel units or context) with the full mission
A small task's focused test and suiteRun inline, once each
A high-risk change or a large task's long suitesDelegate an executor
Need an adversarial review of a diffDelegate a fresh-context reviewer
Multi-step debug that grows the contextDelegate a debug worker; feed results back inline
An understood change, even across files, with no named reasonDo it inline; no delegation needed
Quick git/state checkDo it inline; no delegation needed

Execution Steps

  1. Identify which gate applies. If none applies, skip delegation.
  2. Draft a self-contained mission for the worker — include:
    • Exact goal (one sentence)
    • File paths or targets to act on
    • Relevant prior context the worker needs (decisions, conventions, prior findings)
    • Constraints (style, budget)
    • For a writer: the applicable test-first policy and runner from ## Implementation Routing. Write one RED test per requested rule; for every existing command or option the change touches, add one test proving its previous behavior still holds; add no other cases; update help text and docs for any changed command, option, or message
    • For a verifier: the Verify handoff from the Delegated Verification Gate (the whole feature document, the baseline commit, the probe command forms, first-launch probes in a mktemp -d scratch copy never inside the workspace, and its severity rules)
    • Expected evidence to return (e.g., file written, test output, URL found)
    • Allowed toolsets/MCP/skills the worker should use
    • Any SKILL.md paths to load before work
  3. Call delegate_task with that mission.
  4. Wait for the worker summary.
  5. Verify the claimed output (check file existence, test result, side effect).
  6. Synthesize the verified result into your orchestrator reply.

Output Contract

After synthesizing worker results, return:

  • What was delegated and to how many workers
  • What each worker returned (verified, not just claimed)
  • Any discrepancy between worker self-report and verified evidence
  • Final answer or next step for the user

References

  • references/tuning-knobs.md — Full table of delegate_task configuration parameters and the explicit toolset/MCP/skill checklist for worker missions.

评价

还没有评价。欢迎分享你的使用感受。

评论

无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。

还没有评论。来写第一条吧。

提交后会立即公开。请勿填写邮箱、电话等个人信息。URL 最多 2 个。

0 / 1000 字

此表单受 reCAPTCHA 保护,适用 Google 的隐私权政策和服务条款。

同一仓库的技能

branch-pr

免费

Prepare pull requests by checking approved issue links, change-type labels, and required automated checks. Draft the repository’s PR template using observed test results, while leaving commits, pushes, PR creation, and merging under human control.

中文简介正在准备中。当前显示英语简介。

  • Preparing a branch for review
  • Drafting a PR linked to an approved issue
  • Checking required CI before declaring a PR merge-ready
Gentleman-Programming/gentle-ai79202026年10月22日 更新

Split changes exceeding 400 lines into linked pull requests with clear boundaries and dependencies. Choose a chain structure, keep tests and documentation with each work unit, and document review order, verification plans, and justified size exceptions.

中文简介正在准备中。当前显示英语简介。

  • Splitting a change that exceeds 400 lines
  • Planning PRs that can merge independently
  • Coordinating a feature through a draft tracker PR
Gentleman-Programming/gentle-ai79202026年10月22日 更新

Structure guides, READMEs, design documents, and pull request descriptions so readers can find answers quickly. Put outcomes first, introduce details gradually, and use examples, tables, and checklists to support understanding and review.

中文简介正在准备中。当前显示英语简介。

  • Reorganizing a long or dense README
  • Writing an onboarding guide for new contributors
  • Making a pull request's review steps clear
Gentleman-Programming/gentle-ai79202026年10月22日 更新

Draft clear, considerate comments for code reviews, issue replies, and team updates. Put the actionable point first, explain technical reasons when needed, and give a concrete next step while matching the conversation’s language.

中文简介正在准备中。当前显示英语简介。

  • Requesting a code change with a clear reason
  • Replying to a contributor’s issue
  • Writing a brief asynchronous team update
Gentleman-Programming/gentle-ai79202026年10月22日 更新

Create and verify gentle-ai journeys, which are scripted sequences of product actions. Check their declarations separately from actual execution, review outdated expectations after behavior changes, and record execution results for pull requests.

中文简介正在准备中。当前显示英语简介。

  • Adding a product operation journey
  • Updating expectations after behavior changes
  • Reproducing a bench failure from CI
Gentleman-Programming/gentle-ai79202026年10月22日 更新

Prepare Gentle AI changes for review by linking approved issues, following branch and commit naming rules, and drafting pull requests with observed test results. Check change size and required automated checks before claiming readiness.

中文简介正在准备中。当前显示英语简介。

  • Preparing a branch for a fix or feature
  • Drafting a pull request linked to an approved issue
  • Organizing test results and reviewer notes
Gentleman-Programming/gentle-ai79202026年10月22日 更新

查看 Gentleman-Programming 的全部技能

报告此技能的问题