add-generative-answers

Name: add-generative-answers
Author: microsoft

372stars76forks

How to install add-generative-answers?

microsoft/skills-for-copilot-studio/add-generative-answers

$npx -y skills add microsoft/skills-for-copilot-studio --skill add-generative-answers

Installs into the current project.

Summary

Add generative answer nodes (SearchAndSummarizeContent or AnswerQuestionWithAI) to a Copilot Studio topic. Use this instead of /add-node when the user asks to add grounded answers, knowledge search, generative answers, or AI-powered responses — these nodes require specific patter

Files · 1

View on GitHub

SKILL.md

1	# Add Generative Answers
2
3	Add `SearchAndSummarizeContent` nodes to generate responses grounded in the agent's knowledge sources.
4
5	## Important: Do You Even Need a Topic?
6
7	When knowledge sources are added to the agent (via `/add-knowledge`), the AI can directly search them without any topic if it recognizes a QnA-style query. A dedicated topic with `SearchAndSummarizeContent` is useful when:
8	- You want to restrict the search to a subset of knowledge sources (not all of them)
9	- You want to control the flow around the answer (e.g., follow-up questions, formatting, adaptive cards)
10	- You want to process the response before showing it (e.g., extract content, combine with other data)
11	- You want to use a specific input other than the user's last message
12
13	If the user just wants the agent to answer questions from its knowledge, adding the knowledge source may be enough.
14
15	## Instructions
16
17	1. Auto-discover the agent directory:
18	```
19	Glob: **/agent.mcs.yml
20	```
21	NEVER hardcode an agent name.
22
23	2. Determine the approach based on what the user needs:
24	- Add to existing topic: Read the target topic and insert a SearchAndSummarizeContent node
25	- Create new search topic: Generate a complete topic with the search pattern
26
27	3. Look up the schema for both nodes:
28	```bash
29	node ${CLAUDE_SKILL_DIR}/../../scripts/schema-lookup.bundle.js resolve CreateSearchQuery
30	node ${CLAUDE_SKILL_DIR}/../../scripts/schema-lookup.bundle.js resolve SearchAndSummarizeContent
31	```
32
33	4. Read `settings.mcs.yml` to check if `GenerativeActionsEnabled: true`. This determines the best pattern:
34	- `GenerativeActionsEnabled: true` → prefer Pattern 2 (Orchestrator): use topic inputs/outputs and let the orchestrator handle the response. This is the best approach for generative-orchestrated agents.
35	- `GenerativeActionsEnabled: false` (or not set) → use Pattern 1 (Direct Response): `autoSend: false` + manual SendActivity, or Pattern 3 (Fallback Search) for a simple all-knowledge fallback.
36	- Verbatim/exact content needed → use Pattern 4 (Precision Search): `SearchKnowledgeSources` + `CreateSearchQuery` for raw results without AI summarization (insurance policies, HR docs, legal text).
37
38	5. Ask the user to clarify the behavior (if not already clear from their request):
39	- Should it search all knowledge sources or only specific ones?
40	- Should general model knowledge also be used, or only the configured knowledge sources?
41	- Should the response be sent automatically to chat, or processed first (e.g., custom formatting, adaptive card, combined with other data)?
42
43	6. Always precede `SearchAndSummarizeContent` with `CreateSearchQuery` to preserve conversational context. Never pass `=System.Activity.Text` directly to `SearchAndSummarizeContent` — the raw last message may lack context (e.g., "tell me more about that"). `CreateSearchQuery` rewrites the input into an optimized search query. Access the result via `Topic.<ResultVar>.SearchQuery`.
44
45	7. Generate unique IDs for all nodes (format: `<nodeType>_<6-8 random alphanumeric>`).
46
47	8. Build the YAML using the appropriate pattern. For full YAML examples, see [patterns.md](patterns.md). For the complete property reference table, see [property-reference.md](property-reference.md).
48
49	## SearchAndSummarizeContent vs AnswerQuestionWithAI
50
51	\| Node \| Use When \| Data Source \| Output \|
52	\|------\|----------\|-------------\|--------\|
53	\| `SearchAndSummarizeContent` \| You want answers grounded in the agent's knowledge sources (websites, SharePoint, Dataverse) \| Agent's configured knowledge \| AI-summarized response \|
54	\| `SearchKnowledgeSources` \| You need verbatim/exact content — insurance policies, legal text, HR docs — where AI summarization could lose details \| Agent's configured knowledge \| Raw search results (no AI summary) \|
55	\| `AnswerQuestionWithAI` \| You want a response based only on conversation history and general model knowledge \| No external data \| AI-generated response \|
56
57	Use `SearchAndSummarizeContent` for the vast majority of cases (what people call "generative answers"). Use `SearchKnowledgeSources` when you need raw, unsummarized results for precision scenarios (pair with `CreateSearchQuery` for better search accuracy). Use `AnswerQuestionWithAI` only when you explicitly want the model to respond

People who liked this also viewed

6 picks

Type

add-generative-answers

372stars76forks

View on GitHub

How to install add-generative-answers?

microsoft/skills-for-copilot-studio/add-generative-answers

$npx -y skills add microsoft/skills-for-copilot-studio --skill add-generative-answers

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Add Generative Answers
2
3	Add `SearchAndSummarizeContent` nodes to generate responses grounded in the agent's knowledge sources.
4
5	## Important: Do You Even Need a Topic?
6
7	When knowledge sources are added to the agent (via `/add-knowledge`), the AI can directly search them without any topic if it recognizes a QnA-style query. A dedicated topic with `SearchAndSummarizeContent` is useful when:
8	- You want to restrict the search to a subset of knowledge sources (not all of them)
9	- You want to control the flow around the answer (e.g., follow-up questions, formatting, adaptive cards)
10	- You want to process the response before showing it (e.g., extract content, combine with other data)
11	- You want to use a specific input other than the user's last message
12
13	If the user just wants the agent to answer questions from its knowledge, adding the knowledge source may be enough.
14
15	## Instructions
16
17	1. Auto-discover the agent directory:
18	```
19	Glob: **/agent.mcs.yml
20	```
21	NEVER hardcode an agent name.
22
23	2. Determine the approach based on what the user needs:
24	- Add to existing topic: Read the target topic and insert a SearchAndSummarizeContent node
25	- Create new search topic: Generate a complete topic with the search pattern
26
27	3. Look up the schema for both nodes:
28	```bash
29	node ${CLAUDE_SKILL_DIR}/../../scripts/schema-lookup.bundle.js resolve CreateSearchQuery
30	node ${CLAUDE_SKILL_DIR}/../../scripts/schema-lookup.bundle.js resolve SearchAndSummarizeContent
31	```
32
33	4. Read `settings.mcs.yml` to check if `GenerativeActionsEnabled: true`. This determines the best pattern:
34	- `GenerativeActionsEnabled: true` → prefer Pattern 2 (Orchestrator): use topic inputs/outputs and let the orchestrator handle the response. This is the best approach for generative-orchestrated agents.
35	- `GenerativeActionsEnabled: false` (or not set) → use Pattern 1 (Direct Response): `autoSend: false` + manual SendActivity, or Pattern 3 (Fallback Search) for a simple all-knowledge fallback.
36	- Verbatim/exact content needed → use Pattern 4 (Precision Search): `SearchKnowledgeSources` + `CreateSearchQuery` for raw results without AI summarization (insurance policies, HR docs, legal text).
37
38	5. Ask the user to clarify the behavior (if not already clear from their request):
39	- Should it search all knowledge sources or only specific ones?
40	- Should general model knowledge also be used, or only the configured knowledge sources?
41	- Should the response be sent automatically to chat, or processed first (e.g., custom formatting, adaptive card, combined with other data)?
42
43	6. Always precede `SearchAndSummarizeContent` with `CreateSearchQuery` to preserve conversational context. Never pass `=System.Activity.Text` directly to `SearchAndSummarizeContent` — the raw last message may lack context (e.g., "tell me more about that"). `CreateSearchQuery` rewrites the input into an optimized search query. Access the result via `Topic.<ResultVar>.SearchQuery`.
44
45	7. Generate unique IDs for all nodes (format: `<nodeType>_<6-8 random alphanumeric>`).
46
47	8. Build the YAML using the appropriate pattern. For full YAML examples, see [patterns.md](patterns.md). For the complete property reference table, see [property-reference.md](property-reference.md).
48
49	## SearchAndSummarizeContent vs AnswerQuestionWithAI
50
51	\| Node \| Use When \| Data Source \| Output \|
52	\|------\|----------\|-------------\|--------\|
53	\| `SearchAndSummarizeContent` \| You want answers grounded in the agent's knowledge sources (websites, SharePoint, Dataverse) \| Agent's configured knowledge \| AI-summarized response \|
54	\| `SearchKnowledgeSources` \| You need verbatim/exact content — insurance policies, legal text, HR docs — where AI summarization could lose details \| Agent's configured knowledge \| Raw search results (no AI summary) \|
55	\| `AnswerQuestionWithAI` \| You want a response based only on conversation history and general model knowledge \| No external data \| AI-generated response \|
56
57	Use `SearchAndSummarizeContent` for the vast majority of cases (what people call "generative answers"). Use `SearchKnowledgeSources` when you need raw, unsummarized results for precision scenarios (pair with `CreateSearchQuery` for better search accuracy). Use `AnswerQuestionWithAI` only when you explicitly want the model to respond

People who liked this also viewed

6 picks

Type