drawio

Name: drawio
Author: GoogleCloudPlatform

78stars64forks

How to install drawio?

googlecloudplatform/cxas-scrapi/drawio

$npx -y skills add GoogleCloudPlatform/cxas-scrapi --skill drawio

Installs into the current project.

Summary

Extracts conversational transcripts from .drawio XML files.

Files · 1

View on GitHub

SKILL.md

1	# Draw.io Ingestor Skill
2
3	Use this skill when you need to extract dialogue turns and conversation flows
4	from `.drawio` diagram files.
5
6	## Protocol for Flowchart & Diagram Traversal
7
8	When processing `.drawio` flowchart designs, you must parse them as **complete
9	directed graphs** to construct dialogue paths, rather than extracting flat text
10	turns.
11
12	### 1. Parse Graph Structural Topology
13
14	* Scan `<mxCell>` tags where `vertex="1"`. Extract their `id` and `value`
15	(text label).
16	* Scan `<mxCell>` tags where `edge="1"`. Match `source` and `target` vertex
17	IDs to identify directed pathways.
18	* Clean cell text by stripping internal HTML styling, tags, and formatting
19	metadata (e.g., `<div>`, `<font>`, CSS styles). Unescape HTML entities
20	(e.g., `"` or `<`) to obtain clean text.
21
22	### 2. Dialogue State & Path Traversal
23
24	* Locate the start node (typically a top-level cell labeled "Start",
25	"Welcome", or "Entry").
26	* Perform a step-by-step traversal (DFS/BFS) along the directed edges. Each
27	unique path from start to a terminal node constitutes a distinct
28	conversation flow.
29	* State Mapping:
30	* Prompt Node: Maps to an Agent Turn (using the node's cleaned
31	label).
32	* Directed Edge Label: Maps to a User Input (representing speech
33	utterances or DTMF keys that trigger that transition).
34	* Computational Node: Maps to a System Action / Webhook (e.g.,
35	database lookups or validation calls).
36	* Decision Diamond: Represents logical branching.
37
38	### 3. Brand and Domain Consistency Mapping
39
40	Ensure that you map all domain concepts consistently to the target brand or
41	theme requested by the user or specified in the requirements. Do not mix
42	multiple industrial domains in a single report.
43
44	* If the target brand has specific concepts, map requirements to fit that
45	brand's services.
46	* If no brand is specified, use clean, generic customer service phrasing
47	suitable for the context.
48	* Avoid using technical or backend-specific terminology in spoken Agent turns.
49
50	### 4. Format the Resulting CXAS Transcript
51
52	Compile the traversed paths into a structured CXAS YAML format complying with
53	the root-level turn structures (`turns` containing `speaker: Agent\|User`,
54	`text`, and optional actions). Ensure every scenario starts with the
55	standardized Agent welcome greeting, is voice-realistic, and ends with the
56	standardized sign-off calling the `end_session` tool.
57
58	## Example Target Schema
59
60	```yaml
61	subintent_id: check_loyalty_status
62	subintent_name: "Check Loyalty VIP Status"
63	parent_cuj: "Loyalty Rewards"
64	turns:
65	- speaker: Agent
66	text: "Hello! Thanks for calling [Brand]. How can I help you today?"
67	- speaker: User
68	text: "I need to check my loyalty rewards point balance."
69	- speaker: Agent
70	text: "I'd be happy to help you check your loyalty rewards point balance. May I have your membership phone number?"
71	- speaker: User
72	text: "Yes, it is 555-0199."
73	- speaker: Agent
74	text: "Thank you. Let me verify your rewards balance."
75	webhook_call:
76	name: check_rewards_balance
77	payload:
78	phone_number: "5550199"
79	response:
80	vip_status: "ELITE"
81	points_balance: 1500
82	- speaker: Agent
83	text: "I've found your account. You are an Elite member with a balance of 1,500 points. Is there anything else I can help you with today?"
84	- speaker: User
85	text: "No, that's all. Thank you."
86	- speaker: Agent
87	text: "Thank you for calling [Brand]! Goodbye."
88	tool_call:
89	name: end_session
90	payload:
91	session_escalated: false
92	reason: "Conversation completed successfully"
93	```
94
95	## Linguistic & Voice Naturalness Standards
96
97	All generated spoken dialogue turns (Agent voice turns) MUST strictly adhere to
98	high-fidelity spoken voice standards. Subagents must ensure:
99
100	1. Numeric Voice Normalization: Spoken Agent turns MUST NOT contain raw
101	digits, formatted currencies, or punctuation symbols representing numbers
102	(e.g., do NOT write `"450"`, `"$909"`, `"555-0199"`). Instead, numbers must
103	be explicitly spelled out phonetically:
104	* Correct: `"four hundred fifty points"`, `"nine hundred nine dollars"`.
105	* IDs and Phone Numbers: Must be written digit-by-digit separated by
106	spaces or commas: `"five five five, zero, one, nine, nine"`.
107	2. Spoken Breath Span Limit: Agent turns must remain concise, natural, and
108	conversational. Individual spoken text blocks MUST NOT exceed **300
109	characters** inside a single turn.
110	3. Vocabulary Smoothness: Avoid robotic repetitions of the same long words
111	(do not repeat the same word of length 5+ more than 4 times in a single
112	turn).
113	4. Conversational Politeness: Agent turns must maintain standard polite
114	voice markers (`please`, `thank you`, `thanks`, `certainly`, `happy to

People who liked this also viewed

6 picks

Type

web-design-guidelinesReview UI code for Web Interface Guidelines compliance. Use when asked to "review my UI", "check accessibility", "audit design", "review UX", or "check my site against best practices".SkillsJul 2026483k29k
lark-slides飞书幻灯片：创建和编辑幻灯片。创建演示文稿、读取幻灯片内容、管理幻灯片页面（创建、删除、读取、局部替换）。当用户需要创建或编辑幻灯片、读取或修改单个页面时使用。当用户给出 doubao.com 的 /slides/ URL/token 时，也应直接使用本 skill，不要因为域名不是飞书而回退到 WebFetch；路由依据是 URL 路径模式和 token，而不是域名。不负责：云文档内容编辑（走 SkillsJul 2026332k16k
design-taste-frontendTaste-Skill - gives your AI good taste. stops the AI from generating boring, generic slop SkillsJul 2026280k66k
ui-ux-pro-maxUI/UX design intelligence for web and mobile. Searchable local database with 84 styles, 192 color palettes, 74 font pairings, 192 product types, 98 UX guidelines, 104 icon entries, 16 GSAP motion presSkillsJul 2026280k109k
caveman-helpQuick-reference card for all caveman modes, skills, and commands. One-shot display, not a persistent mode. Trigger: /caveman-help, "caveman help", "what caveman commands", "how do I use caveman".SkillsJul 2026249k92k
codebase-designShared vocabulary for designing deep modules. Use when the user wants to design or improve a module's interface, find deepening opportunities, decide where a seam goes, make code more testable or AI-nSkillsJul 2026225k182k

drawio

78stars64forks

View on GitHub

How to install drawio?

googlecloudplatform/cxas-scrapi/drawio

$npx -y skills add GoogleCloudPlatform/cxas-scrapi --skill drawio

Installs into the current project.

Summary

Extracts conversational transcripts from .drawio XML files.

Files · 1

View on GitHub

SKILL.md

1	# Draw.io Ingestor Skill
2
3	Use this skill when you need to extract dialogue turns and conversation flows
4	from `.drawio` diagram files.
5
6	## Protocol for Flowchart & Diagram Traversal
7
8	When processing `.drawio` flowchart designs, you must parse them as **complete
9	directed graphs** to construct dialogue paths, rather than extracting flat text
10	turns.
11
12	### 1. Parse Graph Structural Topology
13
14	* Scan `<mxCell>` tags where `vertex="1"`. Extract their `id` and `value`
15	(text label).
16	* Scan `<mxCell>` tags where `edge="1"`. Match `source` and `target` vertex
17	IDs to identify directed pathways.
18	* Clean cell text by stripping internal HTML styling, tags, and formatting
19	metadata (e.g., `<div>`, `<font>`, CSS styles). Unescape HTML entities
20	(e.g., `"` or `<`) to obtain clean text.
21
22	### 2. Dialogue State & Path Traversal
23
24	* Locate the start node (typically a top-level cell labeled "Start",
25	"Welcome", or "Entry").
26	* Perform a step-by-step traversal (DFS/BFS) along the directed edges. Each
27	unique path from start to a terminal node constitutes a distinct
28	conversation flow.
29	* State Mapping:
30	* Prompt Node: Maps to an Agent Turn (using the node's cleaned
31	label).
32	* Directed Edge Label: Maps to a User Input (representing speech
33	utterances or DTMF keys that trigger that transition).
34	* Computational Node: Maps to a System Action / Webhook (e.g.,
35	database lookups or validation calls).
36	* Decision Diamond: Represents logical branching.
37
38	### 3. Brand and Domain Consistency Mapping
39
40	Ensure that you map all domain concepts consistently to the target brand or
41	theme requested by the user or specified in the requirements. Do not mix
42	multiple industrial domains in a single report.
43
44	* If the target brand has specific concepts, map requirements to fit that
45	brand's services.
46	* If no brand is specified, use clean, generic customer service phrasing
47	suitable for the context.
48	* Avoid using technical or backend-specific terminology in spoken Agent turns.
49
50	### 4. Format the Resulting CXAS Transcript
51
52	Compile the traversed paths into a structured CXAS YAML format complying with
53	the root-level turn structures (`turns` containing `speaker: Agent\|User`,
54	`text`, and optional actions). Ensure every scenario starts with the
55	standardized Agent welcome greeting, is voice-realistic, and ends with the
56	standardized sign-off calling the `end_session` tool.
57
58	## Example Target Schema
59
60	```yaml
61	subintent_id: check_loyalty_status
62	subintent_name: "Check Loyalty VIP Status"
63	parent_cuj: "Loyalty Rewards"
64	turns:
65	- speaker: Agent
66	text: "Hello! Thanks for calling [Brand]. How can I help you today?"
67	- speaker: User
68	text: "I need to check my loyalty rewards point balance."
69	- speaker: Agent
70	text: "I'd be happy to help you check your loyalty rewards point balance. May I have your membership phone number?"
71	- speaker: User
72	text: "Yes, it is 555-0199."
73	- speaker: Agent
74	text: "Thank you. Let me verify your rewards balance."
75	webhook_call:
76	name: check_rewards_balance
77	payload:
78	phone_number: "5550199"
79	response:
80	vip_status: "ELITE"
81	points_balance: 1500
82	- speaker: Agent
83	text: "I've found your account. You are an Elite member with a balance of 1,500 points. Is there anything else I can help you with today?"
84	- speaker: User
85	text: "No, that's all. Thank you."
86	- speaker: Agent
87	text: "Thank you for calling [Brand]! Goodbye."
88	tool_call:
89	name: end_session
90	payload:
91	session_escalated: false
92	reason: "Conversation completed successfully"
93	```
94
95	## Linguistic & Voice Naturalness Standards
96
97	All generated spoken dialogue turns (Agent voice turns) MUST strictly adhere to
98	high-fidelity spoken voice standards. Subagents must ensure:
99
100	1. Numeric Voice Normalization: Spoken Agent turns MUST NOT contain raw
101	digits, formatted currencies, or punctuation symbols representing numbers
102	(e.g., do NOT write `"450"`, `"$909"`, `"555-0199"`). Instead, numbers must
103	be explicitly spelled out phonetically:
104	* Correct: `"four hundred fifty points"`, `"nine hundred nine dollars"`.
105	* IDs and Phone Numbers: Must be written digit-by-digit separated by
106	spaces or commas: `"five five five, zero, one, nine, nine"`.
107	2. Spoken Breath Span Limit: Agent turns must remain concise, natural, and
108	conversational. Individual spoken text blocks MUST NOT exceed **300
109	characters** inside a single turn.
110	3. Vocabulary Smoothness: Avoid robotic repetitions of the same long words
111	(do not repeat the same word of length 5+ more than 4 times in a single
112	turn).
113	4. Conversational Politeness: Agent turns must maintain standard polite
114	voice markers (`please`, `thank you`, `thanks`, `certainly`, `happy to

People who liked this also viewed

6 picks

Type