@rizom/brain 0.2.0-alpha.132 → 0.2.0-alpha.134
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/brain.js +496 -496
- package/dist/entities.d.ts +3 -0
- package/dist/entities.js.map +2 -2
- package/dist/index.js +2 -2
- package/dist/index.js.map +1 -1
- package/dist/plugins.d.ts +3 -0
- package/dist/plugins.js.map +2 -2
- package/dist/seed-content/rover-seed-content-core/playbook/rover-first-knowledge-loop.md +89 -0
- package/dist/seed-content/rover-seed-content-core/playbook/rover-onboarding.md +9 -46
- package/dist/services.js.map +2 -2
- package/dist/site.js +4 -4
- package/dist/site.js.map +8 -8
- package/package.json +1 -1
|
@@ -0,0 +1,89 @@
|
|
|
1
|
+
---
|
|
2
|
+
title: Rover First Knowledge Loop
|
|
3
|
+
status: active
|
|
4
|
+
audience: anchor
|
|
5
|
+
lifecycle: onboarding
|
|
6
|
+
starterText: Save a first idea
|
|
7
|
+
description: Learn Rover by saving a first idea and seeing how your knowledge becomes reusable.
|
|
8
|
+
starterPrompt: Start playbook rover-first-knowledge-loop.
|
|
9
|
+
completionMode: agent-confirmed
|
|
10
|
+
---
|
|
11
|
+
|
|
12
|
+
# Playbook
|
|
13
|
+
|
|
14
|
+
## Purpose
|
|
15
|
+
|
|
16
|
+
Teach the operator how Rover works by saving one useful seed, retrieving it, and transforming it into working material.
|
|
17
|
+
|
|
18
|
+
## Operating Rules
|
|
19
|
+
|
|
20
|
+
- Ask one question at a time.
|
|
21
|
+
- Teach Rover by doing real actions.
|
|
22
|
+
- Use existing tools to save useful information as durable entities.
|
|
23
|
+
- After meaningful tool actions, explain what Rover just did and why it matters.
|
|
24
|
+
- Do not publish anything unless the operator explicitly asks and confirms the publishing action.
|
|
25
|
+
- Advance steps only after their Done when conditions are satisfied.
|
|
26
|
+
- Runtime evidence from entity creation and updates is attached to the active run automatically where supported.
|
|
27
|
+
- Complete the playbook only after the run reaches a final step.
|
|
28
|
+
- Do not present non-exit playbook actions as buttons or scripted choices; show progress through the current step and continue through normal chat.
|
|
29
|
+
|
|
30
|
+
## Steps
|
|
31
|
+
|
|
32
|
+
### First note
|
|
33
|
+
|
|
34
|
+
Say: Now let’s save one useful seed. Send me a rough idea, note, link, or fragment you want Rover to remember.
|
|
35
|
+
|
|
36
|
+
To do:
|
|
37
|
+
|
|
38
|
+
- Ask for one rough idea, note, link, or fragment the operator wants Rover to remember.
|
|
39
|
+
- Save it as the appropriate durable entity, usually a note or link.
|
|
40
|
+
- For a rough idea or fragment saved as the first note, call system_create with `source: { kind: "text", content }`; do not use a generate source and do not turn the seed into an async generated draft.
|
|
41
|
+
- Use "note" as the operator-facing term for note knowledge entries.
|
|
42
|
+
- Do not offer to collect another seed during this playbook; guide to the retrieval and transformation demonstration next.
|
|
43
|
+
- After saving the first seed, say it was saved or captured, then ask the operator to find/show that saved note next; do not say you found it before the operator asks for retrieval, and do not ask for another rough idea, link, note, or fragment during this playbook.
|
|
44
|
+
- Explain that rough ideas become reusable markdown knowledge inside Rover.
|
|
45
|
+
- Explain the first loop clearly: rough thought → durable knowledge → retrieval → transformation.
|
|
46
|
+
|
|
47
|
+
Done when:
|
|
48
|
+
|
|
49
|
+
- A first knowledge seed has been saved.
|
|
50
|
+
|
|
51
|
+
### Retrieve and transform
|
|
52
|
+
|
|
53
|
+
Say: Ask me to find that note now. After we bring it back, we’ll turn it into something useful — an outline for later writing, a short draft, or a reusable brief.
|
|
54
|
+
|
|
55
|
+
To do:
|
|
56
|
+
|
|
57
|
+
- At the start of this step, ask the operator to find/show the saved note; do not offer to collect another note, seed, link, idea, or fragment.
|
|
58
|
+
- If the operator asks to see, find, or show the saved note, retrieve it with system_get or system_search before saying you found it; do not rely only on conversation memory or playbook evidence.
|
|
59
|
+
- Prefer system_get with the saved note title/slug from the confirmed create result when the note title is known; use system_search only when the identifier is genuinely unclear.
|
|
60
|
+
- Every retrieval response in this step must end by offering the transformation choices: an outline for later writing, a short draft, or a reusable brief.
|
|
61
|
+
- After retrieval, explain the flywheel: more stored knowledge makes future answers and drafts more useful.
|
|
62
|
+
- When the operator picks an option or accepts a suggested angle with wording like "do that", transform the retrieved note directly in chat.
|
|
63
|
+
- Do not call system_create for an outline, short draft, or reusable brief unless the operator explicitly asks to save, store, create, or persist it as a durable entity.
|
|
64
|
+
- Do not ask for confirmation for an inline transformation.
|
|
65
|
+
- If the operator asks "Do that as an outline", write the outline in the response instead of creating a note.
|
|
66
|
+
- After the chat transformation, say the onboarding loop is complete: Rover saved a seed, retrieved it, and transformed it into useful working material.
|
|
67
|
+
- Do not publish anything unless the operator explicitly asks and confirms the publishing action.
|
|
68
|
+
|
|
69
|
+
Done when:
|
|
70
|
+
|
|
71
|
+
- The saved note has been retrieved and transformed in chat.
|
|
72
|
+
|
|
73
|
+
### Done
|
|
74
|
+
|
|
75
|
+
Say: You’re set up. Rover now has a first memory it can retrieve and transform.
|
|
76
|
+
|
|
77
|
+
To do:
|
|
78
|
+
|
|
79
|
+
- Explain the Rover loop: save, retrieve, connect, transform, and manage publishing work.
|
|
80
|
+
- Give a short list of useful next prompts.
|
|
81
|
+
- Remind the operator they can keep using chat to save, retrieve, transform, and manage knowledge.
|
|
82
|
+
|
|
83
|
+
## Next Prompts
|
|
84
|
+
|
|
85
|
+
- Save this idea as a note...
|
|
86
|
+
- Turn my latest note into an outline.
|
|
87
|
+
- What topics am I circling lately?
|
|
88
|
+
- Draft a LinkedIn post from this essay.
|
|
89
|
+
- Show me what is ready to publish.
|
|
@@ -5,7 +5,7 @@ audience: anchor
|
|
|
5
5
|
trigger: first-anchor-web-chat
|
|
6
6
|
lifecycle: onboarding
|
|
7
7
|
starterText: Set up Rover
|
|
8
|
-
description:
|
|
8
|
+
description: Tune Rover's identity and anchor profile before using the knowledge loop.
|
|
9
9
|
starterPrompt: Start playbook rover-onboarding.
|
|
10
10
|
completionMode: agent-confirmed
|
|
11
11
|
---
|
|
@@ -14,12 +14,12 @@ completionMode: agent-confirmed
|
|
|
14
14
|
|
|
15
15
|
## Purpose
|
|
16
16
|
|
|
17
|
-
|
|
17
|
+
Set up Rover's durable identity and anchor profile so future knowledge and publishing work has the right frame.
|
|
18
18
|
|
|
19
19
|
## Operating Rules
|
|
20
20
|
|
|
21
21
|
- Ask one question at a time.
|
|
22
|
-
- Teach Rover by doing real
|
|
22
|
+
- Teach Rover by doing real setup work.
|
|
23
23
|
- Use existing tools to save useful information as durable entities.
|
|
24
24
|
- After meaningful tool actions, explain what Rover just did and why it matters.
|
|
25
25
|
- Do not publish anything unless the operator explicitly asks and confirms the publishing action.
|
|
@@ -88,64 +88,27 @@ To do:
|
|
|
88
88
|
- Do not use fields-only updates for anchor-profile.
|
|
89
89
|
- Do not use system_create for anchor-profile; anchor-profile is an existing singleton profile record.
|
|
90
90
|
- After saving, explain that Rover uses the anchor profile to shape answers, site content, and publishing workflows around the operator.
|
|
91
|
+
- After the profile is saved and the setup playbook is complete, offer the next playbook by saying the operator can continue with `Start playbook rover-first-knowledge-loop.` to save, retrieve, and transform a first idea.
|
|
91
92
|
|
|
92
93
|
Done when:
|
|
93
94
|
|
|
94
95
|
- The anchor profile has been created or updated.
|
|
95
96
|
|
|
96
|
-
### First note
|
|
97
|
-
|
|
98
|
-
Say: Now let’s save one useful seed. Send me a rough idea, note, link, or fragment you want Rover to remember.
|
|
99
|
-
|
|
100
|
-
To do:
|
|
101
|
-
|
|
102
|
-
- Ask for one rough idea, note, link, or fragment the operator wants Rover to remember.
|
|
103
|
-
- Save it as the appropriate durable entity, usually a note or link.
|
|
104
|
-
- For a rough idea or fragment saved as the first note, call system_create with `source: { kind: "text", content }`; do not use a generate source and do not turn the seed into an async generated draft.
|
|
105
|
-
- Use "note" as the operator-facing term for note knowledge entries.
|
|
106
|
-
- Do not offer to collect another seed during onboarding; guide to the retrieval and transformation demonstration next.
|
|
107
|
-
- After saving the first seed, say it was saved or captured, then ask the operator to find/show that saved note next; do not say you found it before the operator asks for retrieval, and do not ask for another rough idea, link, note, or fragment during onboarding.
|
|
108
|
-
- Explain that rough ideas become reusable markdown knowledge inside Rover.
|
|
109
|
-
- Explain the first loop clearly: rough thought → durable knowledge → retrieval → transformation.
|
|
110
|
-
|
|
111
|
-
Done when:
|
|
112
|
-
|
|
113
|
-
- A first knowledge seed has been saved.
|
|
114
|
-
|
|
115
|
-
### Retrieve and transform
|
|
116
|
-
|
|
117
|
-
Say: Ask me to find that note now. After we bring it back, we’ll turn it into something useful — an outline for later writing, a short draft, or a reusable brief.
|
|
118
|
-
|
|
119
|
-
To do:
|
|
120
|
-
|
|
121
|
-
- At the start of this step, ask the operator to find/show the saved note; do not offer to collect another note, seed, link, idea, or fragment.
|
|
122
|
-
- If the operator asks to see, find, or show the saved note, retrieve it with system_get or system_search before saying you found it; do not rely only on conversation memory or playbook evidence.
|
|
123
|
-
- Prefer system_get with the saved note title/slug from the confirmed create result when the note title is known; use system_search only when the identifier is genuinely unclear.
|
|
124
|
-
- Every retrieval response in this step must end by offering the transformation choices: an outline for later writing, a short draft, or a reusable brief.
|
|
125
|
-
- After retrieval, explain the flywheel: more stored knowledge makes future answers and drafts more useful.
|
|
126
|
-
- When the operator picks an option or accepts a suggested angle with wording like "do that", transform the retrieved note directly in chat.
|
|
127
|
-
- Do not call system_create for an outline, short draft, or reusable brief unless the operator explicitly asks to save, store, create, or persist it as a durable entity.
|
|
128
|
-
- Do not ask for confirmation for an inline transformation.
|
|
129
|
-
- If the operator asks "Do that as an outline", write the outline in the response instead of creating a note.
|
|
130
|
-
- After the chat transformation, say the onboarding loop is complete: Rover saved a seed, retrieved it, and transformed it into useful working material.
|
|
131
|
-
- Do not publish anything unless the operator explicitly asks and confirms the publishing action.
|
|
132
|
-
|
|
133
|
-
Done when:
|
|
134
|
-
|
|
135
|
-
- The saved note has been retrieved and transformed in chat.
|
|
136
|
-
|
|
137
97
|
### Done
|
|
138
98
|
|
|
139
|
-
Say: You’re set up. Rover now has a clear identity
|
|
99
|
+
Say: You’re set up. Rover now has a clear identity and an anchor profile for you. Next, say `Start playbook rover-first-knowledge-loop.` when you want to save, retrieve, and transform a first idea.
|
|
140
100
|
|
|
141
101
|
To do:
|
|
142
102
|
|
|
143
|
-
- Explain
|
|
103
|
+
- Explain that this setup playbook tuned Rover's identity and the operator's anchor profile.
|
|
104
|
+
- Explain the Rover loop briefly: save, retrieve, connect, transform, and manage publishing work.
|
|
105
|
+
- Offer the first knowledge loop as the next guided playbook instead of continuing it inside this setup playbook.
|
|
144
106
|
- Give a short list of useful next prompts.
|
|
145
107
|
- Remind the operator they can keep using chat to save, retrieve, transform, and manage knowledge.
|
|
146
108
|
|
|
147
109
|
## Next Prompts
|
|
148
110
|
|
|
111
|
+
- Start playbook rover-first-knowledge-loop.
|
|
149
112
|
- Save this idea as a note...
|
|
150
113
|
- Turn my latest note into an outline.
|
|
151
114
|
- What topics am I circling lately?
|