agent-engineering
/agent-engineering
Pomaga w inżynierii promptów LLM, kompilowaniu skilli ze źródeł, systemach ekspertowych, opisach domeny i zadań.
- Autor
- jstarfilms ↗
- Repozytorium
- vibecode-protocol-suite ↗
- Licencja
- ISC
- Ostatnia zmiana repozytorium
- 17.10.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 37
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 17.10.2026, bez odwołań do wycofanych modeli i API.
ISC, otwarta.
Co robi i kiedy
Pomaga w inżynierii promptów LLM, kompilowaniu skilli ze źródeł, systemach ekspertowych, opisach domeny i zadań.
Wyzwalacze z opisu
Oryginalny opis (EN)
Use when engineering LLM prompts, compiling skills from sources ('source-to-skill'), engineering expert systems ('expert-system-engineer' CommonKADS/rules), writing agent docs ('writing-for-agents'), modeling domain vocabulary ('domain-modeling' CONTEXT.md/ADRs), creating specs ('conversation-to-spec'), curating/importing skills ('skill-curator'), spawning task DAGs ('spawn-task'), running code tournaments ('arena'), managing subagents, or delegating browser and desktop tasks to Codex CLI.
Jak działa
- Sub-Skills
Podgląd SKILL.md
---
name: agent-engineering
description: Use when engineering LLM prompts, compiling skills from sources ('source-to-skill'), engineering expert systems ('expert-system-engineer' CommonKADS/rules), writing agent docs ('writing-for-agents'), modeling domain vocabulary ('domain-modeling' CONTEXT.md/ADRs), creating specs ('conversation-to-spec'), curating/importing skills ('skill-curator'), spawning task DAGs ('spawn-task'), running code tournaments ('arena'), managing subagents, or delegating browser and desktop tasks to Codex CLI.
author: J StaR Films / Matt Pocock / P-Stack
coauthored: J StaR Films / Takomi
version: 2.1.0
---
# Agent Engineering Suite
Comprehensive suite for prompt engineering, agent steering rules, domain dictionaries (`CONTEXT.md`), task decomposition DAGs, and subagent orchestration.
> [!IMPORTANT]
> **Foundational Guide:** When authoring or editing any skill, `AGENTS.md`, or subagent prompt, read [`writing-for-agents/SKILL.md`](writing-for-agents/SKILL.md) and [`writing-for-agents/SKILL-MECHANICS.md`](writing-for-agents/SKILL-MECHANICS.md) first to ensure tight context pointers and zero token waste.
## Sub-Skills
| Sub-Skill | Purpose | Path |
|---|---|---|
| **`writing-for-agents`** | Core rules for writing skills, AGENTS.md, context pointers, and router mechanics | [`writing-for-agents/SKILL.md`](writing-for-agents/SKILL.md) |
| **`codex-control`** | Delegate browser, desktop, and native-plugin tasks through the installed Codex CLI | [`codex-control/SKILL.md`](codex-control/SKILL.md) |
| **`skill-curator`** | Intake, standardize, classify, and wire new skills into the ecosystem | [`skill-curator/SKILL.md`](skill-curator/SKILL.md) |
| **`skill-creator`** | Scaffolding new agent skills following repository conventions | [`skill-creator/SKILL.md`](skill-creator/SKILL.md) |
| **`source-to-skill`** | Distill raw sources (docs, books, transcripts, PDFs) into operational agent skills | [`source-to-skill/SKILL.md`](source-to-skill/SKILL.md) |
| **`expert-system-engineer`** | Engineer CommonKADS knowledge models, MYCIN rule systems, and inference engines | [`expert-system-engineer/SKILL.md`](expert-system-engineer/SKILL.md) |
| **`domain-modeling`** | Generates `CONTEXT.md` ubiquitous language glossary with `_Avoid_` anti-synonyms and ADRs | [`domain-modeling/SKILL.md`](domain-modeling/SKILL.md) |
| **`conversation-to-spec`** | Synthesizes chat discussions into formal technical specifications | [`conversation-to-spec/SKILL.md`](conversation-to-spec/SKILL.md) |
| **`spawn-task`** | Decomposes specs into tracer-bullet vertical slice task packets with blocking DAGs | [`spawn-task/SKILL.md`](spawn-task/SKILL.md) |
| **`to-questionnaire`** | Generates async questionnaires for domain experts and external stakeholders | [`to-questionnaire/SKILL.md`](to-questionnaire/SKILL.md) |
| **`arena`** | Spawns parallel candidate implementations and blind-judges the strongest solution | [`arena/SKILL.md`](arena/SKILL.md) |
| **`automate-me`** | Scans terminal history and repeated workflows to produce new agent skills and scripts | [`automate-me/SKILL.md`](automate-me/SKILL.md) |
| **`reflect`** | Spawns 3 review subagents over transcripts to extract learnings and patch skills | [`reflect/SKILL.md`](reflect/SKILL.md) |
| **`create-verification-skill`** | Generates project-local verification harnesses driving apps end-to-end | [`create-verification-skill/SKILL.md`](create-verification-skill/SKILL.md) |
| **`maintain-verification-skill`** | Audits and updates project verification harnesses as code evolves | [`maintain-verification-skill/SKILL.md`](maintain-verification-skill/SKILL.md) |
| **`subagent-driven-development`** | Parallel subagent orchestration and implementation task loops | [`subagent-driven-development/SKILL.md`](subagent-driven-development/SKILL.md) |
| **`prompt-engineering`** | Advanced prompting patterns, meta-prompts, and eval harnesses | [`prompt-engineering/SKILL.md`](prompt-engineering/SKILL.md) |
| **`optimize-agent-context`** | Optimizing context windows and rule files | [`optimize-agent-context/SKILL.md`](optimize-agent-context/SKILL.md) |
| **`crafting-effective-readmes`** | Writing structured repository documentation | [`crafting-effective-readmes/SKILL.md`](crafting-effective-readmes/SKILL.md) |
Pliki w paczce (37)
- SKILL.md 4 KB
- arena/SKILL.md 5 KB
- automate-me/SKILL.md 8 KB
- codex-control/SKILL.md 7 KB
- codex-control/references/verification.md 3 KB
- crafting-effective-readmes/SKILL.md 3 KB
- crafting-effective-readmes/section-checklist.md 1 KB
- crafting-effective-readmes/style-guide.md 1 KB
- crafting-effective-readmes/using-references.md 1 KB
- crafting-effective-readmes/references/art-of-readme.md 24 KB
- crafting-effective-readmes/references/make-a-readme.md 7 KB
- crafting-effective-readmes/references/standard-readme-example-maximal.md 2 KB
- crafting-effective-readmes/references/standard-readme-example-minimal.md 1 KB
- crafting-effective-readmes/references/standard-readme-spec.md 10 KB
- crafting-effective-readmes/templates/internal.md 2 KB
- crafting-effective-readmes/templates/oss.md 2 KB
- crafting-effective-readmes/templates/personal.md 1 KB
- crafting-effective-readmes/templates/xdg-config.md 1 KB
- expert-system-engineer/SKILL.md 13 KB
- expert-system-engineer/references/examples.md 7 KB
- expert-system-engineer/references/knowledge-modeling.md 7 KB
- expert-system-engineer/references/rule-engineering.md 7 KB
- expert-system-engineer/references/terminology.md 6 KB
- optimize-agent-context/SKILL.md 5 KB
- prompt-engineering/SKILL.md 9 KB
- reflect/SKILL.md 5 KB
- reflect/references/divergent-reviewer.md 4 KB
- reflect/references/judgment-reviewer.md 3 KB
- reflect/references/synthesizer.md 4 KB
- reflect/references/tooling-reviewer.md 5 KB
- skill-curator/SKILL.md 4 KB
- source-to-skill/SKILL.md 8 KB
- source-to-skill/references/compilation-guide.md 5 KB
- source-to-skill/references/knowledge-spec.md 6 KB
- spawn-task/SKILL.md 3 KB
- writing-for-agents/SKILL-MECHANICS.md 3 KB
- writing-for-agents/SKILL.md 11 KB
Wersje
| Repozytorium | Wynik | Licencja |
|---|---|---|
| Ta strona · jstarfilms/vibecode-protocol-suite | 85 | ISC |
| ckorhonen/claude-skills | 100 | MIT |
| jacob-balslev/skill-graph | 100 | Apache-2.0 |
Migawka bazy z 13.11.2026 · wersja paczki 811e62a64e86 (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 95prompt-engineering
Pisze i optymalizuje prompty dla poleceń, hooków, skilli i podagentów oraz szablony promptów do produkcji. - 91context-engineering
Wyjaśnia składniki, mechanikę i ograniczenia kontekstu w systemach agentowych, przy pisaniu i optymalizacji poleceń, skilli i promptów podagentów. - 90prompt-engineering
Pomaga pisać polecenia, hooki, skille i prompty dla agentów oraz podagentów, optymalizuje odpowiedzi LLM i projektuje szablony promptów. - 100prompt-engineering
Zapewnia ramy inżynierii promptów do projektowania, optymalizowania i iteracyjnego ulepszania promptów LLM oraz instrukcji dla AI. - 95skill-template
Dostarcza szablon do tworzenia nowych skilli Agent Skills dla inżynierii kontekstu przy rozbudowie kolekcji. - 100skill-creator-expert
Projektuje, tworzy i waliduje skille PCL z wydobyciem wiedzy domenowej, obejmując zasady projektowania skilli, wzorce architektury, kontrolę jakości i szablon analizy domeny.