@tduyng/pi-cost 0.0.3 → 0.0.4

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/README.md +34 -34
  2. package/package.json +1 -1
package/README.md CHANGED
@@ -1,56 +1,46 @@
1
1
  # @tduyng/pi-cost
2
2
 
3
- What the session spent, what the provider still allows, and how long each message took. None of it enters the model's context.
3
+ Session spend, live rate-limit windows, footer segments and per-message timestamps, read from files pi already wrote. None of it enters a request.
4
4
 
5
- ## What it costs
5
+ Pi's footer prints a running total and not the split: no per-provider view over the month, no record of what a delegated child cost, no sight of the rate-limit window about to end the day.
6
6
 
7
- Zero tokens. No tool, no prompt text, no session entry. The footer segments and the timestamps are display only, and the panels read session files pi already wrote.
7
+ ## The panel
8
8
 
9
- The point is the bill you cannot see otherwise. A long session's spend is one command away instead of one surprise.
9
+ `/cost` opens a TUI panel with session spend in USD, token totals, a per-provider token table, and the live rate-limit windows.
10
10
 
11
- ## /cost
11
+ The panel fits the window it opens in: body at most 70% of the rows, each line truncated to the column width, the header naming the range when the rest scrolls. Up and Down move a row, Page Up and Page Down a screen, Enter or Esc closes.
12
12
 
13
- A TUI panel with session spend in USD, token totals, provider rate-limit bars, and a per-provider token table.
14
-
15
- The panel fits the window: body at most 70% of the rows, each line truncated to the column width, and the header shows the range when the rest scrolls. Up and Down move a row, Page Up and Page Down a screen, Enter or Esc closes it.
16
-
17
- An `Extensions` row appears when another extension spent a model call on this session's behalf, such as a `subagent` child or the `harness-loop` writer. Those calls bill to the same plan but cannot land in the parent's transcript, so each appends a line to `<agentDir>/overhead.jsonl` and the panel sums them for this session. The row names the cost, the tokens, the call count, and each extension's share. The `Cost` row stays what the conversation itself cost, so the two are read side by side and never added twice.
18
-
19
- The token table groups by provider across four trailing windows: 5h, 7d, 30d, and calendar month. Input, output, cache read, and cache write are shown separately, then the total and the cost. A zero column collapses to a dot, and a provider with no tokens in any window is omitted. The cost is the `usage.cost.total` each turn recorded in its session file, so for a subscription provider such as `claude` it is the API-equivalent price rather than a charge.
20
-
21
- ## Rate-limit windows
13
+ The token table groups by provider across four trailing windows: 5h, 7d, 30d, calendar month. Input, output, cache read and cache write are separate, then the total and the cost. A zero column collapses to a dot, and a provider with no tokens is omitted.
22
14
 
23
15
  | Provider | Windows shown |
24
16
  | --- | --- |
25
- | Command Code | session 5h, weekly, remaining credits (monthly plus purchased plus free) |
17
+ | Command Code | session 5h, weekly, remaining credits |
26
18
  | OpenCode Go | rolling, weekly, monthly |
27
19
  | Claude | 5h, weekly all models, Opus, Sonnet, as percent bars |
28
20
 
29
- A section paints only when the credential it needs is on the machine. Without one the probe returns nothing, the section drops, and its cached rows are deleted.
21
+ A section paints only when the credential it needs is on the machine. Without one the probe returns nothing and the section drops. Providers with no usage endpoint (opencode, mistral, gmi) appear in the token table only. Command Code stays out of the panel unless pi holds a credential for it in `auth.json`.
30
22
 
31
- ## Footer segments
23
+ Amounts under $0.01 print with four decimals, under $1 with three, and $1 or more with two. A model with no known pricing contributes zero, so the panel understates spend instead of guessing.
32
24
 
33
- Three extra segments in pi's native footer:
25
+ ## Where the numbers come from
34
26
 
35
- - `tps` is average output tokens per second across the session.
36
- - `agents` is the count of running subagent tasks.
37
- - `git` is changed files and the PR number from `gh pr view`.
27
+ Token totals come from scanning the session `.jsonl` files. Cost is the `usage.cost.total` each turn recorded, not a recomputation from a price table. For a subscription provider such as `claude` that is the API-equivalent price rather than a charge, and the panel says so.
38
28
 
39
- Session spend and cache hit rate are left to pi's own footer, which already prints both.
29
+ Two caches keep it quick to open. `<agentDir>/cost-cache.json` stores parsed records keyed by file mtime and size, so a re-open skips unchanged files. Quota data is cached 24 hours, and files outside the scan window are skipped without a stat call.
40
30
 
41
- `/statusline` toggles the segments, and `--statusline-no-pr` skips the `gh pr view` lookup. The git segment refreshes when a run settles and on a debounced watcher on `.git/HEAD` with a 250ms delay, so a branch switch paints at once and an idle session spawns nothing. A worktree that stores `.git` as a file is covered by the settle refresh.
31
+ Work that bills to the same plan but cannot land in the transcript is recorded separately. A `subagent` child or the `harness-loop` writer appends a line to `<agentDir>/overhead.jsonl`, and the `Extensions` row sums them for this session with the call count and each extension's share. The `Cost` row stays what the conversation itself cost, so the two are read side by side and never added twice.
42
32
 
43
- ## /timestamps
33
+ ## Footer and timestamps
44
34
 
45
- Toggles per-message clock markers in the transcript: time of day, a relative offset such as `3m ago`, and whether the entry is a user message, an AI response, or a tool call.
35
+ Three segments added to pi's native footer:
46
36
 
47
- Markers are custom session entries, so they never enter model context and the cached prefix stays stable. Two same-role entries within a 2-second window are read as a duplicate from a double-loaded extension and suppressed.
37
+ - `tps`: average output tokens per second across the session.
38
+ - `agents`: running subagent tasks.
39
+ - `git`: changed files and the PR number from `gh pr view`.
48
40
 
49
- ## How it reads the bill
41
+ Session spend and cache hit rate stay with pi's own footer, which already prints both. `/statusline` toggles the segments and `--statusline-no-pr` skips the `gh pr view` lookup. The git segment refreshes on settle and on a debounced watch of `.git/HEAD`, so a branch switch paints at once and an idle session spawns nothing.
50
42
 
51
- Token totals come from scanning session `.jsonl` files. A cache at `<agentDir>/cost-cache.json` stores parsed records keyed by file mtime and size, so reopening `/cost` skips unchanged files. Quota data is cached 24 hours, and files older than the scan window are skipped without a stat call.
52
-
53
- Amounts under $0.01 print with four decimals, under $1 with three, and $1 or more with two. A model with no known pricing contributes zero, so the panel understates spend rather than guessing it.
43
+ `/timestamps` toggles per-message clock markers: time of day, a relative offset such as `3m ago`, and whether the entry is a user message, an assistant reply or a tool call. Markers are custom session entries, so they never enter model context, and two same-role entries inside a 2-second window are read as a double-loaded extension and suppressed.
54
44
 
55
45
  ## Install
56
46
 
@@ -60,9 +50,13 @@ pi install npm:@tduyng/pi-cost
60
50
 
61
51
  Local checkout: `pi install /path/to/dotfiles/pi/agent/extensions/pi-cost`
62
52
 
63
- ## Configure
53
+ Optional:
54
+
55
+ - `OPENCODE_API_KEY` for the OpenCode Go windows.
56
+ - A Claude OAuth credential for the Claude windows, from the macOS keychain or `~/.claude/.credentials.json`.
57
+ - `gh` on PATH for the PR segment.
64
58
 
65
- `OPENCODE_API_KEY` is needed for the OpenCode Go windows.
59
+ ## Config
66
60
 
67
61
  `<agentDir>/cost.json` controls the table and the sections. It is read each time `/cost` opens, and an invalid file is ignored with a warning. Every key is optional:
68
62
 
@@ -74,6 +68,12 @@ Local checkout: `pi install /path/to/dotfiles/pi/agent/extensions/pi-cost`
74
68
  }
75
69
  ```
76
70
 
77
- - `providers` lists the providers shown in the token table. Omit it to show every provider except `gmi`, `mistral`, `google`, `claude-bridge`, and `commandcode`, which are hidden unless named.
71
+ - `providers` lists the providers in the token table. Omit it to show every provider except `gmi`, `mistral`, `google`, `claude-bridge` and `commandcode`, which are hidden unless named.
78
72
  - `cost` lists the providers whose cost cell is filled. Omit it to fill every shown provider.
79
- - `limits` lists the rate-limit sections to fetch by provider id. Omit it to show every section with a credential.
73
+ - `limits` lists the rate-limit sections to fetch. Omit it to show every section with a credential.
74
+
75
+ ## Limits
76
+
77
+ - Network probes time out at 10 seconds and the keychain at 5, so a slow provider costs a pause.
78
+ - Quota rows older than 24 hours are marked stale rather than deleted.
79
+ - `cost-cache.json` is capped at 8 MB.
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@tduyng/pi-cost",
3
- "version": "0.0.3",
3
+ "version": "0.0.4",
4
4
  "description": "Session spend, live rate-limit windows, footer segments, and per-message timestamps. Display only, never in the model context.",
5
5
  "keywords": [
6
6
  "pi",