Skip to content

fix: use current model's context window for usage_update size #412

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
timvisher-dd wants to merge 2 commits into
base: main
Choose a base branch
from
+534 −6
Open

fix: use current model's context window for usage_update size #412

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
2399a9e
chore: add bin/test for local CI validation
timvisher-dd Mar 13, 2026
fea5979
fix: use current model's context window for usage_update size
timvisher-dd Mar 13, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
53 changes: 53 additions & 0 deletions bin/test
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,53 @@
#!/usr/bin/env bash
# Runs the same checks as CI by parsing .github/workflows/ci.yml directly.
# If CI steps change, this script automatically picks them up.
#
# Local adaptations:
# - `npm ci` checks if node_modules is in sync with package-lock.json
# and runs a clean install if not (CI always does npm ci).
# - `npm run format:check` checks only git-tracked files because CI
# runs on a clean checkout but locally we have untracked x.* scratch
# files that fail prettier.
set -euo pipefail

cd "$(git rev-parse --show-toplevel)"

ci_yaml=".github/workflows/ci.yml"

if ! command -v yq &>/dev/null; then
echo "error: yq is required (brew install yq)" >&2
exit 1
fi

# Extract run steps
mapfile -t names < <(yq '.jobs.build.steps[] | select(.run) | .name' "$ci_yaml")
mapfile -t commands < <(yq '.jobs.build.steps[] | select(.run) | .run' "$ci_yaml")

for i in "${!commands[@]}"; do
cmd="${commands[$i]}"
name="${names[$i]}"

echo "=== ${name} ==="

if [[ "$cmd" == "npm ci" ]]; then
# Check if node_modules matches package-lock.json. If not, run
# npm ci to match what CI does. This catches stale-dependency bugs
# like sdk-tools.d.ts resolving locally but not in CI.
if npm ls --all >/dev/null 2>&1; then
echo "(node_modules in sync — skipping npm ci)"
else
echo "(node_modules out of sync — running npm ci)"
npm ci
fi
elif [[ "$cmd" == "npm run format:check" ]]; then
# Local override: format:check on git-tracked files only
git ls-files -z '*.ts' '*.tsx' '*.js' '*.jsx' '*.json' '*.md' '*.yml' '*.yaml' '*.css' '*.html' \
| xargs -0 npx prettier --check
else
eval "$cmd"
fi

echo ""
done

echo "=== All CI checks passed ==="
59 changes: 53 additions & 6 deletions src/acp-agent.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -482,6 +482,8 @@ export class ClaudeAcpAgent implements Agent {
};

let lastAssistantTotalUsage: number | null = null;
let lastAssistantModel: string | null = null;
let lastContextWindowSize: number = 200000;

const userMessage = promptToClaude(params);

Expand Down Expand Up @@ -538,9 +540,26 @@ export class ClaudeAcpAgent implements Agent {
break;
}
case "compact_boundary": {
// We don't know the exact size, but since we compacted,
// we set it to zero. The client gets the exact size on the next message.
// Send used:0 immediately so the client doesn't keep showing
// the stale pre-compaction context size until the next turn.
//
// This is a deliberate approximation: we don't know the exact
// post-compaction token count (only the SDK's next API call
// reveals that). But used:0 is directionally correct — context
// just dropped dramatically — and the real value replaces it
// within seconds when the next result message arrives.
// The alternative (no update) leaves the client showing e.g.
// "944k/1m" right after the user sees "Compacting completed",
// which is confusing and wrong.
lastAssistantTotalUsage = 0;
await this.client.sessionUpdate({
sessionId: message.session_id,
update: {
sessionUpdate: "usage_update",
used: 0,
size: lastContextWindowSize,
},
});
await this.client.sessionUpdate({
sessionId: message.session_id,
update: {
Expand Down Expand Up @@ -584,10 +603,23 @@ export class ClaudeAcpAgent implements Agent {
session.accumulatedUsage.cachedReadTokens += message.usage.cache_read_input_tokens;
session.accumulatedUsage.cachedWriteTokens += message.usage.cache_creation_input_tokens;

// Calculate context window size from modelUsage (minimum across all models used)
const contextWindows = Object.values(message.modelUsage).map((m) => m.contextWindow);
const contextWindowSize =
contextWindows.length > 0 ? Math.min(...contextWindows) : 200000;
// Calculate context window size from the current model's usage.
// The modelUsage keys may use the requested model alias (e.g. "claude-opus-4-6")
// while message.model on assistant messages has the resolved API response model
// (e.g. "claude-opus-4-6-20250514"), so we fall back to prefix matching.
const currentModel = lastAssistantModel;
const matchingModelUsage = currentModel
? (message.modelUsage[currentModel] ??
Object.entries(message.modelUsage)
.filter(([key]) => currentModel.startsWith(key) || key.startsWith(currentModel))
.sort((a, b) => b[0].length - a[0].length)[0]?.[1])
: undefined;
Comment on lines +611 to +616
Copy link
Copy Markdown
Contributor

@SteffenDE SteffenDE Mar 24, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

In my testing, the result modelUsage contained a claude-opus-4-6[1m] and the assistant message claude-opus-4-6 (no suffix). But assuming there can be a suffix - if the assistant message contains claude-opus-4-6-20250514 or similar, this would not match, because it checks for startsWith.

Maybe do something like this and actually calculate the common prefix:

function commonPrefixLength(a: string, b: string) {
  let i = 0;
  while (i < a.length && i < b.length && a[i] === b[i]) {
    i++;
  }
  return i;
}

function getMatchingModelUsage(modelUsage: Record<string, ModelUsage>, currentModel: string) {
  let bestKey: string | null = null;
  let bestLen = 0;

  for (const key of Object.keys(modelUsage)) {
    const len = commonPrefixLength(key, currentModel);
    if (len > bestLen) {
      bestLen = len;
      bestKey = key;
    }
  }

  if (bestKey) {
    return modelUsage[bestKey];
  }
}

and then use it as

const matchingModelUsage = lastAssistantModel ? getMatchingModelUsage(message.modelUsage, lastAssistantModel) : null;
const contextWindowSize = matchingModelUsage?.contextWindow ?? 200000;
lastContextWindowSize = contextWindowSize;

// Fallback to 200k: this is hit when lastAssistantModel is null (e.g. the
// assistant message lacked a model field) or no modelUsage key matches.
// 200k is a conservative default — the Anthropic API should always populate
// BetaMessage.model, so this path is unlikely in practice.
const contextWindowSize = matchingModelUsage?.contextWindow ?? 200000;
lastContextWindowSize = contextWindowSize;

// Send usage_update notification
if (lastAssistantTotalUsage !== null) {
Expand Down Expand Up @@ -716,6 +748,11 @@ export class ClaudeAcpAgent implements Agent {
}

// Store latest assistant usage (excluding subagents)
// Sum all token types as a proxy for post-turn context occupancy:
// current turn's output will become next turn's input.
// Note: per the Anthropic API, input_tokens excludes cache tokens —
// cache_read and cache_creation are reported separately, so summing
// all four fields is not double-counting.
if ((message.message as any).usage && message.parent_tool_use_id === null) {
const messageWithUsage = message.message as unknown as SDKResultMessage;
lastAssistantTotalUsage =
Expand All @@ -724,6 +761,16 @@ export class ClaudeAcpAgent implements Agent {
messageWithUsage.usage.cache_read_input_tokens +
messageWithUsage.usage.cache_creation_input_tokens;
}
// Track the current top-level model for context window size lookup
// (exclude subagent messages to stay in sync with lastAssistantTotalUsage)
if (
message.type === "assistant" &&
message.parent_tool_use_id === null &&
message.message.model &&
message.message.model !== "<synthetic>"
) {
lastAssistantModel = message.message.model;
}

// Slash commands like /compact can generate invalid output... doesn't match
// their own docs: https://docs.anthropic.com/en/docs/claude-code/sdk/sdk-slash-commands#%2Fcompact-compact-conversation-history
Expand Down
Loading
Loading