Repository navigation
Expand file tree
/
Copy pathadvisor.ts
More file actions
145 lines (125 loc) · 5.45 KB
/
Copy pathadvisor.ts
File metadata and controls
145 lines (125 loc) · 5.45 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
import type { BetaUsage } from '@anthropic-ai/sdk/resources/beta/messages/messages.mjs'
import { getFeatureValue_CACHED_MAY_BE_STALE } from '../services/analytics/growthbook.js'
import { shouldIncludeFirstPartyOnlyBetas } from './betas.js'
import { isEnvTruthy } from './envUtils.js'
import { getInitialSettings } from './settings/settings.js'
// The SDK does not yet have types for advisor blocks.
// TODO(hackyon): Migrate to the real anthropic SDK types when this feature ships publicly
export type AdvisorServerToolUseBlock = {
type: 'server_tool_use'
id: string
name: 'advisor'
input: { [key: string]: unknown }
}
export type AdvisorToolResultBlock = {
type: 'advisor_tool_result'
tool_use_id: string
content:
| {
type: 'advisor_result'
text: string
}
| {
type: 'advisor_redacted_result'
encrypted_content: string
}
| {
type: 'advisor_tool_result_error'
error_code: string
}
}
export type AdvisorBlock = AdvisorServerToolUseBlock | AdvisorToolResultBlock
export function isAdvisorBlock(param: {
type: string
name?: string
}): param is AdvisorBlock {
return (
param.type === 'advisor_tool_result' ||
(param.type === 'server_tool_use' && param.name === 'advisor')
)
}
type AdvisorConfig = {
enabled?: boolean
canUserConfigure?: boolean
baseModel?: string
advisorModel?: string
}
function getAdvisorConfig(): AdvisorConfig {
return getFeatureValue_CACHED_MAY_BE_STALE<AdvisorConfig>(
'tengu_sage_compass',
{},
)
}
export function isAdvisorEnabled(): boolean {
if (isEnvTruthy(process.env.CLAUDE_CODE_DISABLE_ADVISOR_TOOL)) {
return false
}
// The advisor beta header is first-party only (Bedrock/Vertex 400 on it).
if (!shouldIncludeFirstPartyOnlyBetas()) {
return false
}
return getAdvisorConfig().enabled ?? false
}
export function canUserConfigureAdvisor(): boolean {
return isAdvisorEnabled() && (getAdvisorConfig().canUserConfigure ?? false)
}
export function getExperimentAdvisorModels():
| { baseModel: string; advisorModel: string }
| undefined {
const config = getAdvisorConfig()
return isAdvisorEnabled() &&
!canUserConfigureAdvisor() &&
config.baseModel &&
config.advisorModel
? { baseModel: config.baseModel, advisorModel: config.advisorModel }
: undefined
}
// @[MODEL LAUNCH]: Add the new model if it supports the advisor tool.
// Checks whether the main loop model supports calling the advisor tool.
export function modelSupportsAdvisor(model: string): boolean {
const m = model.toLowerCase()
return (
m.includes('opus-4-6') ||
m.includes('sonnet-4-6') ||
process.env.USER_TYPE === 'ant'
)
}
// @[MODEL LAUNCH]: Add the new model if it can serve as an advisor model.
export function isValidAdvisorModel(model: string): boolean {
const m = model.toLowerCase()
return (
m.includes('opus-4-6') ||
m.includes('sonnet-4-6') ||
process.env.USER_TYPE === 'ant'
)
}
export function getInitialAdvisorSetting(): string | undefined {
if (!isAdvisorEnabled()) {
return undefined
}
return getInitialSettings().advisorModel
}
export function getAdvisorUsage(
usage: BetaUsage,
): Array<BetaUsage & { model: string }> {
const iterations = usage.iterations as
| Array<{ type: string }>
| null
| undefined
if (!iterations) {
return []
}
return iterations.filter(
it => it.type === 'advisor_message',
) as unknown as Array<BetaUsage & { model: string }>
}
export const ADVISOR_TOOL_INSTRUCTIONS = `# Advisor Tool
You have access to an \`advisor\` tool backed by a stronger reviewer model. It takes NO parameters -- when you call it, your entire conversation history is automatically forwarded. The advisor sees the task, every tool call you've made, every result you've seen.
Call advisor BEFORE substantive work -- before writing code, before committing to an interpretation, before building on an assumption. If the task requires orientation first (finding files, reading code, seeing what's there), do that, then call advisor. Orientation is not substantive work. Writing, editing, and declaring an answer are.
Also call advisor:
- When you believe the task is complete. BEFORE this call, make your deliverable durable: write the file, stage the change, save the result. The advisor call takes time; if the session ends during it, a durable result persists and an unwritten one doesn't.
- When stuck -- errors recurring, approach not converging, results that don't fit.
- When considering a change of approach.
On tasks longer than a few steps, call advisor at least once before committing to an approach and once before declaring done. On short reactive tasks where the next action is dictated by tool output you just read, you don't need to keep calling -- the advisor adds most of its value on the first call, before the approach crystallizes.
Give the advice serious weight. If you follow a step and it fails empirically, or you have primary-source evidence that contradicts a specific claim (the file says X, the code does Y), adapt. A passing self-test is not evidence the advice is wrong -- it's evidence your test doesn't check what the advice is checking.
If you've already retrieved data pointing one way and the advisor points another: don't silently switch. Surface the conflict in one more advisor call -- "I found X, you suggest Y, which constraint breaks the tie?" The advisor saw your evidence but may have underweighted it; a reconcile call is cheaper than committing to the wrong branch.`