Api Round Message Grouping
Split conversation history into API-round buckets so compaction retries cut only the chunks tied to the failed round.
Install
$ agentstack add skill-ychampion-cskill-agents-api-round-message-grouping ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Links to your public security report.[](https://agentstack.voostack.com/security/report/skill-ychampion-cskill-agents-api-round-message-grouping) Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
SKILL: API Round Message Grouping
Domain: context-management Trigger: When compacting or retrying requests, use API-round boundaries instead of human-turn heuristics to determine what past messages belong together. Source Pattern: Distilled from reviewed session memory, compaction, and context-budgeting implementations.
Core Method
Walk the message stream and emit a new bucket every time a fresh assistant message with a new message id starts. This respects the API contract that all tool results complete before the next assistant response, so each bucket represents exactly one API round-trip; malformed conversations still fall through because the grouping only fires when a real assistant boundary appears. Downstream compaction retries can then drop or replay whole rounds without breaking tool-result pairing.
Key Rules
- Track the last assistant message id and start a new group only when it changes while at least one message is already buffered.
- Push the final bucket at the end so the last API round isn’t lost.
- Do not split mid-assistant stream (IDs stay constant across streaming chunks), keeping each API response intact.
- Let dangling tool uses remain in the same group rather than inventing extra boundaries; downstream helpers (ensureToolResultPairing) repair them only when needed.
- Name each bucket explicitly when logging so retry diagnostics can report the round that triggered the fallback.
Example Application
When a compaction attempt hits prompt-too-long, first group the transcript into API rounds, identify the round that introduced the oversized assistant response, and drop or compact only that round before retrying.
Anti-Patterns (What NOT to do)
- Do not group by user turns or by time because API chunks can span multiple user messages and tool results.
- Do not start a new group on every assistant chunk; the streaming assistant may emit multiple chunks with the same
id, and splitting them would leave incomplete tool-result pairs.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: ychampion
- Source: ychampion/cskill-agents
- License: MIT
- Homepage: https://x.com/YashasGunderia/status/2039990540016693692?s=20
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.