burnwatch 0.8.0 → 0.8.1

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.
Files changed (2) hide show
  1. package/package.json +1 -1
  2. package/skills/burnwatch-interview/SKILL.md +75 -88
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "burnwatch",
3
- "version": "0.8.0",
3
+ "version": "0.8.1",
4
4
  "description": "Passive cost memory for vibe coding — detects paid services, tracks spend, injects budget context into your AI coding sessions.",
5
5
  "type": "module",
6
6
  "bin": {
package/skills/burnwatch-interview/SKILL.md CHANGED
@@ -9,48 +9,86 @@ allowed-tools: Read, Bash, Glob, Grep
9
9
 
10
10
  You are the interviewer. burnwatch has auto-detected services and applied defaults — your job is to walk the user through confirming or correcting each one in natural conversation.
11
11
 
12
+ ## CRITICAL: Pacing Rules
13
+
14
+ **ONE service (or one logical batch) per message. Then STOP and WAIT for the user's response.**
15
+
16
+ - Do NOT dump all services in one message
17
+ - Do NOT ask about LLMs and infrastructure in the same breath
18
+ - Do NOT proceed to the next service until the user has responded
19
+ - The only exception: free-tier services can be batched into one "any of these need updating?" question
20
+
21
+ The rhythm is: **present → ask → wait → configure → next.**
22
+
12
23
  ## Step 1: Get the current state
13
24
 
14
25
  Run this to get the structured state of all detected services:
15
26
 
16
27
  ```bash
17
- cd $PROJECT_ROOT && node ./node_modules/.bin/burnwatch interview --json
28
+ cd $PROJECT_ROOT && burnwatch interview --json
18
29
  ```
19
30
 
20
- If burnwatch is installed globally or via npx, use that path instead. Parse the JSON output.
31
+ Parse the JSON output. Do NOT show the raw JSON to the user.
21
32
 
22
- ## Step 2: Present the overview
33
+ ## Step 2: Present a brief overview, then start the interview
23
34
 
24
- Start with a brief summary:
35
+ Open with a short summary (2-3 lines max), then immediately start with the first service:
25
36
 
26
- > I found **N services** in your project. Right now they're configured with defaults — let me walk you through each one so we get accurate tracking.
37
+ > I found **N services** in your project. Let me walk through each one to make sure we're tracking accurately.
27
38
  >
28
- > **Current state:** X with API keys (LIVE), Y estimated, Z with no tracking (BLIND)
29
- > **Total budget:** $NNN/mo (all defaults — let's fix that)
39
+ > First up — **Anthropic**. You're defaulted to "API Usage" at $100/mo.
40
+ > [If probe data exists: "I checked your API — you've spent $47.23 this month."]
41
+ > Does $100/mo feel right, or want to adjust?
42
+
43
+ Then STOP. Wait for the user.
30
44
 
31
- ## Step 3: Interview by risk group
45
+ ## Step 3: Walk through services ONE AT A TIME
32
46
 
33
- Walk through services grouped by risk (the JSON is pre-sorted). For each group:
47
+ ### Order: highest risk first
34
48
 
35
- ### LLM / AI Services (ask first — highest variable cost)
36
- - These are the most important to get right
37
- - Ask: "What's your actual monthly spend on Anthropic/OpenAI? The default is $100 — too high? too low?"
38
- - If they have an admin API key, offer to probe: "If you give me your admin key I can check your actual spend"
49
+ 1. **LLM / AI Services** — one at a time (Anthropic, then OpenAI, then Gemini, etc.)
50
+ 2. **Usage-Based Services** — one at a time (Scrapfly, Stripe, Browserbase)
51
+ 3. **Infrastructure** — one at a time (Vercel, Supabase, AWS)
52
+ 4. **Free-tier / flat-rate** — batch these: "PostHog, Inngest, and Resend are all on free tiers at $0. Any of those need updating? If not, moving on."
39
53
 
40
- ### Usage-Based Services (Scrapfly, Browserbase, Stripe)
41
- - Credit-pool services: "What Scrapfly plan are you on? The probe detected Pro with 1M credits — correct?"
42
- - Ask about actual budget, not just plan cost
54
+ ### For each paid service, present:
43
55
 
44
- ### Infrastructure (Vercel, Supabase, AWS)
45
- - Tiered services: "Are you on Supabase Free or Pro? This matters because Pro is $25/mo with overages"
56
+ 1. **What was detected**: service name, current default plan, default budget
57
+ 2. **What the probe found** (if anything): actual spend, credits used, plan detected
58
+ 3. **One clear question**: "Does this look right?" or "What plan are you actually on?"
46
59
 
47
- ### Flat-Rate / Free Tier
48
- - These are usually quick: "PostHog and Inngest are set to Free/$0 — correct? Moving on."
49
- - Don't dwell on these unless the user has something to say
60
+ Example per-service message:
50
61
 
51
- ## Step 4: For each service, use `burnwatch configure`
62
+ > **Scrapfly** — defaulted to Pro ($100/mo, 1M credits).
63
+ > Probe detected: Pro plan confirmed, 250K/1M credits used this month.
64
+ > Keep the $100/mo budget? [Y/adjust]
52
65
 
53
- After the user confirms or corrects, write it back:
66
+ Then STOP. Wait for response.
67
+
68
+ ### After the user responds:
69
+
70
+ 1. Run `burnwatch configure` with their answer
71
+ 2. Confirm what was written (one line)
72
+ 3. Present the NEXT service
73
+
74
+ Example:
75
+
76
+ > Got it — Scrapfly set to Pro, $100/mo budget. ✅
77
+ >
78
+ > Next — **Stripe**. Defaulted to "Standard (2.9% + 30¢)" at $50/mo.
79
+ > I checked your balance: $1,247.83 available.
80
+ > $50/mo budget for Stripe's processing fees — sound right?
81
+
82
+ Then STOP again.
83
+
84
+ ### For free-tier services, batch them:
85
+
86
+ > **PostHog**, **Inngest**, and **Resend** are all set to free tiers at $0/mo.
87
+ > Any of those need updating, or should I lock them in and move on?
88
+
89
+ ## Step 4: Configure each service
90
+
91
+ After the user confirms or corrects, write it back immediately:
54
92
 
55
93
  ```bash
56
94
  burnwatch configure --service <id> --plan "<plan name>" --budget <N>
@@ -66,78 +104,28 @@ To exclude a service:
66
104
  burnwatch configure --service <id> --exclude
67
105
  ```
68
106
 
69
- The configure command outputs JSON confirming what was written. Check the `success` field.
107
+ Always check the JSON output for `"success": true`.
70
108
 
71
- ## Step 5: Show the result
109
+ ## Step 5: Wrap up
72
110
 
73
- After all services are configured, run:
111
+ After all services are configured, run `burnwatch status` and present the brief:
74
112
 
75
- ```bash
76
- burnwatch status
77
- ```
78
-
79
- Present the brief and celebrate what's tracked vs what's still blind.
113
+ > All done! Here's your updated spend brief:
114
+ > [burnwatch status output]
115
+ >
116
+ > N services tracked, total budget $X/mo.
80
117
 
81
118
  ## Key Behaviors
82
119
 
83
- - **Group questions.** Don't ask 14 questions one at a time. "Your LLMs — Anthropic and OpenAI. What plans and budgets?" is one question, not two.
84
- - **Lead with what you know.** If the probe detected their Scrapfly plan, say "I detected you're on Pro with 1M credits and you've used 250K so far. Budget $100/mo? [Y/n]" — don't make them scroll through options.
85
- - **Be brief for obvious services.** Free tier PostHog? "PostHog: Free, $0. Next."
86
- - **Highlight what matters.** If Anthropic is at $87 of a $100 budget, say so. If Scrapfly credits are 85% consumed, flag it.
87
- - **Use the probe data.** The interview JSON includes `probeResult` for services where we got live data. Show it: "Your Anthropic spend this month is $47.23 so far."
88
- - **Respect the user's time.** If they say "defaults are fine for everything" — write them all and move on. Don't force 14 confirmations.
89
- - **One `configure` call per service.** Don't batch. This makes it clear what was written and lets the user correct individual services.
90
-
91
- ## Example Flow
92
-
93
- ```
94
- Agent: I found 8 services in your project. Let me walk through each one
95
- to make sure we're tracking accurately.
96
-
97
- Right now everything has default budgets — let's fix that.
98
-
99
- First, your LLMs:
100
- • Anthropic — defaulted to "API Usage" plan, $100/mo budget.
101
- I probed your API and you've spent $47.23 this month.
102
- • OpenAI — defaulted to "API Usage", $100/mo. No admin key,
103
- so I can't check actual spend.
104
-
105
- What are your actual budgets for these?
106
-
107
- User: Anthropic $150 is better, I've been using it heavily.
108
- OpenAI keep at $100, and here's my admin key: sk-admin-xxx
109
-
110
- Agent: Got it. Let me update both...
111
-
112
- [runs: burnwatch configure --service anthropic --budget 150]
113
- [runs: burnwatch configure --service openai --key sk-admin-xxx --budget 100]
114
-
115
- ✅ Anthropic: API Usage | LIVE | $150/mo
116
- ✅ OpenAI: API Usage | LIVE | $100/mo (probed: 12.3K tokens this month)
117
-
118
- Next, Scrapfly — I detected you're on Pro ($100/mo, 1M credits).
119
- You've used 250K credits so far. Keep the $100 budget?
120
-
121
- User: Yeah that's right.
122
-
123
- Agent: [runs: burnwatch configure --service scrapfly --plan pro --budget 100]
124
- ✅ Scrapfly: Pro | LIVE | $100/mo | 1M credits
125
-
126
- PostHog, Inngest, and Resend are all on free tiers at $0.
127
- Sound right?
128
-
129
- User: Yep.
130
-
131
- Agent: [runs configure for each]
132
-
133
- All done! Here's your updated spend brief:
134
- [runs: burnwatch status]
135
- ```
120
+ - **Lead with what you know.** If the probe detected their plan, state it confidently and ask for confirmation — don't make them pick from a list.
121
+ - **One question at a time.** Never ask about budget AND plan AND API key in the same message. If the plan is confirmed, ask about budget. If budget is confirmed, offer the API key option.
122
+ - **Be brief.** Each message should be 2-4 lines, not paragraphs.
123
+ - **Respect shortcuts.** If the user says "defaults are fine for everything" — configure them all and show the summary. Don't force 14 rounds.
124
+ - **If they offer a key unprompted**, use it immediately and show what the probe found.
125
+ - **Surface concerns.** If Anthropic spend is $87 of $100 budget, say so. If Scrapfly credits are 85% consumed, flag it.
136
126
 
137
127
  ## Probe Data Reference
138
128
 
139
- Services with probes and what they return:
140
-
141
129
  | Service | Confidence | What's detected |
142
130
  |---------|-----------|-----------------|
143
131
  | Scrapfly | high | Plan name, credits used/total |
@@ -149,5 +137,4 @@ Services with probes and what they return:
149
137
  | Browserbase | medium | Session count, browser hours |
150
138
  | Upstash | low | Database count (key validation) |
151
139
  | PostHog | low | Org found (key validation) |
152
- | Gemini, Resend, Inngest, Voyage AI | none | No API — ask user directly |
153
- | AWS | none | Too complex — flag as BLIND |
140
+ | Gemini, Resend, Inngest, Voyage AI, AWS | none | No API — ask user directly |