microsoft/amplifier-bundle-skills

Letzte erfasste Quellaktivität: 2. Oktober 2026 um 18:23

SkillsMP-Katalog aktualisiert: 17. Oktober 2026 um 00:09

gesammelte Skills: 33

GitHub-Stars: 12
GitHub-Forks: 11

README des Repositorys

README-Quelldatei: README.md

# amplifier-bundle-skills Skills tool and Microsoft-curated skills collection for [Amplifier](https://github.com/microsoft/amplifier) agents, following the [Agent Skills specification](https://agentskills.io/). ## Vision This repository pursues five goals: 1. Track the [Agent Skills standard](https://agentskills.io/) as it evolves, maintaining compatibility with specification updates. 2. Support Anthropic Skills 2.0 and emerging multi-provider skill formats as the ecosystem grows. 3. Welcome broader community contributions as the skills ecosystem matures beyond Microsoft-curated content. 4. Demonstrate lightweight agents-as-skills using the enhanced fork execution format (`context: fork`). 5. Provide practical skills-vs-agents guidance to help developers choose the right tool for their use case. ## What This Does Packages the tool-skills module (at `modules/tool-skills/`) with context instructions and a curated collection of reusable skills into composable behaviors for any Amplifier bundle. The tool-skills module is now maintained in this repository; the [standalone amplifier-module-tool-skills repo](https://github.com/microsoft/amplifier-module-tool-skills) is deprecated. ## Behaviors | Behavior | What you get | Use when | |----------|-------------|----------| | `skills:behaviors/skills` | Tool + instructions + curated skills | Default -- batteries included | | `skills:behaviors/skills-tool` | Tool + instructions only | Your bundle brings its own skills | ## Curated Skills | Skill | Description | |-------|-------------| | **image-vision** | LLM-based image analysis across multiple providers (Anthropic, OpenAI, Gemini, Azure) | | **code-review** | Parallel code review — spawns 3 agents (code reuse, quality, efficiency) to review recent changes | | **mass-change** | Parallel work orchestration — decomposes large changes into 5-30 independent units | | **engineering-patterns** | Thirteen engineering pattern guides behind one catalog entry — CLI packaging, one-line installers, config/state, HTTP services, auth/TLS, file IPC, plugin discovery, instance storage, container orchestration, React microfrontends, MS Graph integration, self-managing tools, Amplifier tool leverage. Each is an L3 reference read on demand. | | **session-debug** | Session diagnostics — diagnoses misconfigured tools, failing operations, unexpected behavior | | **skills-assist** | Skills expert — authoritative consultant for authoring, spec, compatibility, and skills-vs-agents guidance | | **monitor** | Bounded polling loop — watch a thing until it finishes, fails, or needs you, so the end-of-turn notification fires honestly | | **retrospective** | Review how work went, preserve what helped, and choose up to three improvements for the next round | | **create-demo-video** | Produce a narrated demo video of something you built, from a storyboard and real recordings, using the vid, aud, unfold, showrun, and Stories smart tools | ### Retrospectives `/retro` and `/retrospective` invoke the same skill, with the same arguments. They use the recent work in the current conversation, with no extra review agents by default. Add a question to focus it: ```text /retrospective why did this take so long, and what should change next round? /retro what should change next round? /retrospective session abcdef12 — review the implementation process /second-opinion Use retrospective with Astra and Fable to review this round. ``` Reviewer names resolve through your configuration; these are examples, not guaranteed aliases. Named historical sessions use the optional Context Intelligence `session_transcript` tool directly; short IDs require a prefix-capable reader. The plain current-context retro needs neither that tool nor a graph server. When combining with `second-opinion`, the parent gathers evidence once and passes the same brief-only packet to each reviewer. Recommendations do not automatically edit files or start another round. ### Power Skills The `code-review`, `mass-change`, and `session-debug` skills are **power skills** — they use the enhanced skills format to run as isolated subagents with their own tool sets and model preferences. Power skills set `disable-model-invocation: true`, meaning they are user-invoked only (via `/command`) and will not trigger automatically when the LLM processes context. See [Enhanced Skills Format](#enhanced-skills-format) below. ### Expert Skills The `skills-assist` skill is an **expert skill** — it uses `context: fork` to run in a clean context window as an isolated knowledge consultant, but it is not a delegating orchestrator. Unlike power skills, `skills-assist` does **not** have `disable-model-invocation` set, so the LLM can invoke it autonomously to consult the skills authoring expert without requiring an explicit `/skills-assist` command from the user. The `image-vision` skill is a simpler reference skill that loads inline (without forking) and activates in the agent's existing context window. ## Quick Start ### Add the reusable behavior to an existing Amplifier host ```bash amplifier bundle add git+https://github.com/microsoft/amplifier-bundle-skills@main#subdirectory=behaviors/skills.yaml --app ``` This composes the full skills capability (tool, instructions, and curated skills) into the host without selecting this repository's complete root. Use `skills-tool.yaml` in the same `behaviors/` directory when the host brings its own skill sources. ### Compose the behavior in a bundle Include the full behavior (tool + curated skills): ```yaml includes: - bundle: git+https://github.com/microsoft/amplifier-bundle-skills@main#subdirectory=behaviors/skills.yaml ``` Or include just the tool (no curated skills): ```yaml includes: - bundle: git+https://github.com/microsoft/amplifier-bundle-skills@main#subdirectory=behaviors/skills-tool.yaml ``` ### Supporting legacy root This repository also provides a complete, runnable **legacy root** for quick experimentation or when you intentionally want its Foundation-based session: ```bash amplifier bundle add git+https://github.com/microsoft/amplifier-bundle-skills@main amplifier bundle use skills ``` The root composes `skills:behaviors/skills`; it is not the normal reusable capability-install surface. ### Add your own skills alongside curated ones If your bundle ships its own skills, declare them in your behavior YAML following the proven pattern from [amplifier-bundle-superpowers](https://github.com/microsoft/amplifier-bundle-superpowers): ```yaml tools: - module: tool-skills source: git+https://github.com/microsoft/amplifier-bundle-skills@main#subdirectory=modules/tool-skills config: skills: - "git+https://github.com/microsoft/amplifier-bundle-skills@main#subdirectory=skills" - "git+https://github.com/your-org/your-bundle@main#subdirectory=skills" ``` > **Warning:** Do NOT use a top-level `skills:` key in your bundle frontmatter. The foundation layer does not process it -- skill sources placed there will be **silently ignored**. Always use the `tools:` config pattern shown above. ## Architecture ``` amplifier-bundle-skills/ ├── bundle.md # Supporting legacy root (Foundation + full behavior) ├── behaviors/ │ ├── skills.yaml # Full: tool + instructions + curated skills │ └── skills-tool.yaml # Minimal: just the tool + instructions ├── context/ │ └── skills-instructions.md # Agent-facing skills system instructions ├── modules/ │ └── tool-skills/ # tool-skills module (maintained here; standalone repo deprecated) │ ├── pyproject.toml │ ├── amplifier_module_tool_skills/ │ └── tests/ └── skills/ ├── image-vision/ # LLM-based image analysis │ ├── SKILL.md │ └── ... ├── code-review/ # Parallel code review (power skill) │ └── SKILL.md ├── mass-change/ # Parallel work orchestration (power skill) │ └── SKILL.md ├── session-debug/ # Session diagnostics (power skill) │ └── SKILL.md ├── skills-assist/ # Skills authoring expert (expert skill) │ ├── SKILL.md │ ├── authoring-guide.md │ ├── spec-reference.md │ ├── compatibility-matrix.md │ └── skills-vs-agents.md └── monitor/ # Bounded polling loop └── SKILL.md ``` **Design**: Two behaviors serve different consumers. The full behavior (`skills`) pre-configures the curated skills collection via `git+https://` URL. The minimal behavior (`skills-tool`) provides just the tool and instructions for bundles that manage their own skill sources. ## Enhanced Skills Format Power skills use an enhanced SKILL.md frontmatter format that goes beyond the base Agent Skills specification: | Field | Purpose | |-------|---------| | `context: fork` | Skill runs as an isolated subagent with its own conversation | | `auto-load: true` | Skill activates at session start via embedded hooks | | `disable-model-invocation: true` | User-invoked only (via `/command`), not triggered by the model | | `model_role` | Semantic model selection via routing matrix (e.g., `coding`, `reasoning`, `critique`) | | `allowed-tools` | Restricts which tools the subagent can use | | `$ARGUMENTS`, `${SKILL_DIR}` | String substitution in skill body at load time | | `` !`command` `` | Dynamic shell preprocessing — output is spliced into the skill content | These features are implemented by the tool-skills module at `modules/tool-skills/`. See the [README in that directory](modules/tool-skills/README.md) for full format documentation. Power skills can also be authored and validated using the `/skills-assist` expert skill. ## CLI Integration Skills marked `user-invocable: true` register as slash commands and are available directly from the Amplifier CLI: - `/code-review` — Run parallel code review on recent changes - `/mass-change` — Decompose and execute a large change in parallel - `/session-debug` — Diagnose session issues - `/skills-assist` — Consult the skills authoring expert for help creating skills, spec compatibility questions, and skills-vs-agents guidance - `/monitor` — Watch something until it finishes, fails, or needs you (e.g. `/monitor the CI run for PR 412, check every 2m, stop after 1h`) - `/retrospective` (alias `/retro`) — Review recent work and recommend concrete changes for the next round These commands appear in `/help` and `/skills`. They are powered by the `SkillsDiscovery` capability exposed by the tool-skills module, which the CLI queries at startup to register user-invocable skills. ## Contributing > [!NOTE] > This project is not currently accepting external contributions, but we're actively working toward opening this up. We value community input and look forward to collaborating in the future. For now, feel free to fork and experiment! Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit [Contributor License Agreements](https://cla.opensource.microsoft.com). When you submit a pull request, a CLA bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., status check, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA. This project has adopted the [Microsoft Open Source Code of Conduct](https://opensource.microsoft.com/codeofconduct/). For more information see the [Code of Conduct FAQ](https://opensource.microsoft.com/codeofconduct/faq/) or contact [opencode@microsoft.com](mailto:opencode@microsoft.com) with any additional questions or comments. ## Trademarks This project may contain trademarks or logos for projects, products, or services. Authorized use of Microsoft trademarks or logos is subject to and must follow [Microsoft's Trademark & Brand Guidelines](https://www.microsoft.com/legal/intellectualproperty/trademarks/usage/general). Use of Microsoft trademarks or logos in modified versions of this project must not cause confusion or imply Microsoft sponsorship. Any use of third-party trademarks or logos are subject to those third-party's policies. ## License MIT - See [LICENSE](LICENSE) ### Remote download ownership Hosts can set `tool-skills.config.cache_dir` to an owned filesystem directory. This applies to both startup sources and sources added during a session. Local skill paths and user skill discovery remain unchanged. Without this option, remote downloads retain the existing `AMPLIFIER_HOME/cache/skills` default; CLI users do not need to change their configuration. An explicitly configured cache directory remains authoritative.

README-Quelldatei lesen

README des Repositorys wird geladen…

Skills in diesem Repository

Softwareentwickler Projektmanagementspezialisten Softwarequalitätssicherungsanalysten und -tester Sonstige Computerberufe

6 Berufskategorien · 91% klassifiziert

Skill

Beruf

Beschreibung

Aktualisiert

second-opinion

nicht klassifiziert

Independent review of current work or a past session. USE WHEN a user wants a second opinion. DO NOT USE WHEN ordinary code review is wanted — use code-review.

Quellsprache: Englisch

2026-10-02

bug-report

nicht klassifiziert

Investigate a defect and write an evidence-backed maintainer report. USE WHEN asked to file a bug, report an issue, investigate and report, or write a defect up for devs. DO NOT USE WHEN only a fix or direct debugging is wanted — use the normal debugging…

Quellsprache: Englisch

2026-10-02

create-demo-video

nicht klassifiziert

Produce a narrated demo video that shows something the user made or built, cut from recordings they supply or recordings made to their direction. Use when showing a product, feature, tool, or project in action, including explainers built around a live demo.…

Quellsprache: Englisch

2026-09-23

adapt-skill

Sonstige Computerberufe

Adapt a skill written for another AI assistant (OpenAI Codex, ChatGPT, Claude Code, Cursor) into a proper Amplifier SKILL.md. Use when the user wants to adapt a skill, port a skill, convert a skill to amplifier, translate a skill, or has a SKILL.md from…

Quellsprache: Englisch

2026-09-20

skills-assist

Sonstige Computerberufe

Authoritative consultant for all skills-related questions. Use when creating or modifying skills, understanding the Agent Skills spec, troubleshooting skill loading or invocation issues, leveraging enhanced format features (context fork, model_role,…

Quellsprache: Englisch

2026-09-20

councilify

Sonstige Computerberufe

Build a NEW complete "council" for a domain — a panel of orthogonal review lenses that fan out cold, debate to consensus, and return a synthesized verdict with recorded dissent, exactly like /council and /design-council. Identifies the distinct lenses a…

Quellsprache: Englisch

2026-09-19

retrospective

Projektmanagementspezialisten

Review how work went and improve the next round. USE WHEN asked for a retro, delays, or process lessons. DO NOT USE for code review or transcript replay.

Quellsprache: Englisch

2026-09-19

personafy

Softwareentwickler

Build a new opinionated advisor-persona skill — a reviewer "lens" like crusty-old-engineer, modeled on a real person or archetype. Mines the subject's authentic voice, defines its one load-bearing question, drafts it to the family template, and proves it…

Quellsprache: Englisch

2026-09-07

skillify

Softwareentwickler

Capture a repeatable process from the current session into a reusable Amplifier SKILL.md skill file. Use when the user wants to create a skill, save a workflow as a skill, turn a process into a reusable skill, or mentions "skillify", "create skill", "make a…

Quellsprache: Englisch

2026-09-07

bet-sizer

Projektmanagementspezialisten

Delivery-risk reviewer that sizes a plan's investment against the team's real confidence it will land. Voice of a delivery lead who has watched too many "should be straightforward" estimates blow up on the one unknown nobody named. Use at any product…

Quellsprache: Englisch

2026-09-07

cranky-old-sam

Softwareentwickler

Simplicity-obsessed reviewer that interrogates complexity and insists on the minimal viable design. Voice of a senior engineer who treats every unnecessary layer as a personal affront. Use at any checkpoint — brainstorm, design, plan, implement, debug, review…

Quellsprache: Englisch

2026-09-07

crusty-old-engineer

Softwareentwickler

Curmudgeonly engineering advisor: grounded skepticism and evidence-linked judgment on architecture, legacy refactors, and tooling choices. Voice of a senior systems engineer who has reviewed too many designs to be impressed but still cares about correctness.…

Quellsprache: Englisch

2026-09-07

engineering-patterns

Softwareentwickler

Use for tool/service work: CLI packaging, installers, config/state, HTTP, auth/TLS, file IPC, plugins, instance storage, containers, React MFE, Graph, self-update, tool leverage.

Quellsprache: Englisch

2026-09-07

intent-keeper

Projektmanagementspezialisten

Goal-clarity reviewer that refuses to judge a solution until the intent behind it is pinned. A patient interrogator of "why are we doing this?", hunting goal drift — the deliverable quietly replacing the goal. Use at any checkpoint — brainstorm, design, plan,…

Quellsprache: Englisch

2026-09-07

outcome-cartographer

Projektmanagementspezialisten

Outcome-validity reviewer that refuses to count shipped work as progress until it maps to a measurable change in behavior. Hunts the gap between "we shipped it" and "we moved the number" — the output mistaken for an outcome. Use at any product checkpoint —…

Quellsprache: Englisch

2026-09-07

outcomist

Projektmanagementspezialisten

Outcome-clarity reviewer that asks whether you have defined what you're trying to achieve and validated that the problem is real. Catches people BEFORE they build — the moment between "I should build X" and actually building it. Use when a build is starting…

Quellsprache: Englisch

2026-09-07

positioning-critic

Marktforschungsanalysten und Marketingspezialisten

Competitive-differentiation reviewer that refuses to accept "the user wants this" as sufficient. Demands the plan say why a real customer would choose it over every alternative, including doing nothing — voice of a positioning strategist who has watched…

Quellsprache: Englisch

2026-09-07

user-advocate

Projektmanagementspezialisten

User-need reviewer that speaks for the person who isn't in the room — the one who will actually live with what gets built. Hunts the gap between "we can build this" and "they actually want this," in the patient, impatient voice of the absent user. Use at any…

Quellsprache: Englisch

2026-09-07

council-here

Softwarequalitätssicherungsanalysten und -tester

Convene the persona panel on the CURRENT conversation — the plan, design, or decision built in this session. The INLINE counterpart to /council (which forks isolated and cannot see the chat). Use when you want the council to critique what we're working on…

Quellsprache: Englisch

2026-08-30

council

Softwarequalitätssicherungsanalysten und -tester

Convene the persona panel — six orthogonal review lenses — on a target. Cold independent fan-out, debate-to-consensus, synthesized verdict with recorded dissent and a roster manifest.

Quellsprache: Englisch

2026-08-30

image-vision

Softwareentwickler

Analyze images using LLM vision APIs (Anthropic Claude, OpenAI GPT-4, Google Gemini, Azure OpenAI). Use when a task requires understanding image content, describing visual elements, answering questions about images, comparing images, or extracting text from…

Quellsprache: Englisch

2026-08-30

monitor

Netzwerk- und Computersystemadministratoren

Run a bounded, self-checking polling loop (sleep -> check -> decide, repeat) WITHOUT ENDING THE TURN. The turn ends only when the watched thing is done, has failed, or genuinely needs the user's attention — which is what makes Amplifier's end-of-turn…

Quellsprache: Englisch

2026-08-30

product-council-here

Projektmanagementspezialisten

Convene the Product Development Council on the CURRENT conversation — the plan, roadmap, or scope decision built in this session. The INLINE counterpart to /product-council (which forks isolated and cannot see the chat). Use when you want the council to…

Quellsprache: Englisch

2026-08-30

product-council

Projektmanagementspezialisten

Convene the Product Development Council — six orthogonal product-delivery lenses anchored by a mandatory problem-validation gate. Cold independent fan-out, debate-to-consensus, synthesized verdict with recorded dissent and a roster manifest.

Quellsprache: Englisch

2026-08-30

restless-old-brian

Softwareentwickler

Momentum-driven engineering reviewer holding one uncompromising gate: is it REAL, proven end-to-end as a user would? Warm, blunt, forward-driving — proof over claims, plumbing before polish, fail-loud over fallbacks, trust in the model over instructions,…

Quellsprache: Englisch

2026-08-30

tester-breaker

Softwarequalitätssicherungsanalysten und -tester

Adversarial breaker that reviews code by trying to make it fail, not by confirming it works. Hunts the unhappy paths — malformed input, the empty string, the reversed range, the race condition — as a gleeful adversary who assumes everything is broken until a…

Quellsprache: Englisch

2026-08-30

verification-discipline

Softwarequalitätssicherungsanalysten und -tester

Use when verifying that completed work actually works. Auto-surface during /verify mode, post-implementation review, or before claiming a task is done. Teaches testing outcomes vs implementation, the unit/integration/smoke gradient, and what "done" actually…

Quellsprache: Englisch

2026-08-30

mass-change

Softwareentwickler

Research and plan a large-scale change, then execute it in parallel across isolated agents that each open a PR.

Quellsprache: Englisch

2026-06-22

session-debug

Softwareentwickler

Diagnose issues in the current Amplifier session — misconfigured tools, failing operations, unexpected behavior. Use when something isn't working right.

Quellsprache: Englisch

2026-06-22

amplifier-philosophy

Softwareentwickler

Amplifier design philosophy using Linux kernel metaphor. Covers mechanism vs policy, module architecture, event-driven design, and kernel principles. Use when designing new modules or making architectural decisions.

Quellsprache: Englisch

2026-03-23

module-development

Softwareentwickler

Guide for creating new Amplifier modules including protocol implementation, entry points, mount functions, and testing patterns. Use when creating new modules or understanding module architecture.

Quellsprache: Englisch

2026-03-23

python-standards

Softwareentwickler

Python coding standards for Amplifier including type hints, async patterns, error handling, and formatting. Use when writing Python code for Amplifier modules.

Quellsprache: Englisch

2026-03-23

code-review

Softwarequalitätssicherungsanalysten und -tester

Review changed code for reuse, quality, and efficiency, then fix any issues found.

Quellsprache: Englisch

2026-03-20

Es werden 33 von 33 gesammelten Skills angezeigt.

Mit einem KI-Assistenten installieren

Diesen Prompt kopieren und in den verwendeten KI-Assistenten einfügen.

Follow https://skillsmp.com/skill-install/prompt.md to install Agent Skills from https://github.com/microsoft/amplifier-bundle-skills.