Skip to content

test(canary): calibrate quota fixture budget for native isolation #8069

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

Merged
huangruiteng merged 1 commit into from
Nov 5, 2026
Merged

test(canary): calibrate quota fixture budget for native isolation #8069

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
21 changes: 20 additions & 1 deletion docs/reference/contracts/interface-budget-contract.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,7 @@ and size/count budgets.
| --- | --- | --- | --- | --- | --- | --- |
| `heartbeat_prompt_json` | heartbeat automation | wake and route one bounded turn | `quota should-run`, `status`, or `review-packet --handoff-only` | `json_chars <= 5400` plus `interface_budget.within_budget=true` | `nested_keys <= 40` | `top_level_keys <= 30` |
| `review_packet_handoff_only_json` | project-agent handoff | forward the smallest sufficient task packet | full `review-packet` or run-history artifact | `json_chars <= 3000` plus `handoff_interface_budget.within_budget=true` | `nested_keys <= 40` | `top_level_keys <= 18` |
| `quota_should_run_json` | quota guard | decide whether the selected goal may spend compute | `status`, `history`, or active state | `json_chars <= 20500` | `nested_keys <= 575` | `top_level_keys <= 55` |
| `quota_should_run_json` | quota guard | decide whether the selected goal may spend compute | `status`, `history`, or active state | `json_chars <= 21000` | `nested_keys <= 575` | `top_level_keys <= 55` |
| `dashboard_status_json` | operator dashboard | render first-screen operator state | `history`, run artifacts, or project-local adapter output | `json_chars <= 25000` | `nested_keys <= 350` | `top_level_keys <= 27` |

These four budgets measure compact machine payloads. For
Expand Down Expand Up @@ -66,6 +66,25 @@ execution authority or change to frozen experiment or promotion criteria.
已占满旧限、没有 cadence 余量，改为 55 后保留一个键。quota 字符、status 键数、
夹具数量、采样深度和语义断言不变；不授予执行权限，也不改变冻结的实验或 promotion 标准。

The standard premerge runner adds an isolation directory around this fixture.
On base `6a92640346051b79d08fb369354180b613c8fa46` and the budget-only candidate,
quota measures 20,341 characters directly and 20,509 through the native runner,
with 565 nested and 54 top-level keys. Eight complete fixture-root references
each gain the runner's 21-character directory component, accounting for all 168
additional characters. Preserve the original 20,500-ceiling
failure. Calibrate only this character regression allowance to 21,000, leaving
491 characters in the runner; keep the 575/55 structure limits and complete
source paths. The same quota payload with an oversized reason remains rejected.
No production payload, execution budget, host hard limit or frozen qualification
result changes; this is not a compaction or efficiency claim.

标准 premerge 的隔离目录会增长同一夹具的完整路径。上述 base 与仅改预算的候选均为
直接运行 20,341 字符、原生 runner 20,509 字符，565 个嵌套键及 54 个顶层键；
8 处完整夹具路径各增加 21 字符，解释全部 168 字符差异。
保留原 20,500 超限失败，仅将字符回归余量调至 21,000，给 runner 留 491 字符；
575/55 结构限制、完整来源路径及超长 reason 拒绝继续。生产负载、执行预算、宿主
硬边界和冻结资格结果均不变，不据此声称压缩或效率提升。

The successful thin heartbeat projection contains only `schema_version`,
`ok`, `goal_id`, optional `agent_id`, `task_body`, and the compact
`interface_budget`; exact Turn identity and `bootstrap=true` are included only
Expand Down
16 changes: 15 additions & 1 deletion examples/control_plane/hot-path-interface-budget-smoke.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -79,7 +79,10 @@
# 15,419/18,636 characters, 378/513 nested keys, and 52/53 top-level
# keys. Keep the complete typed envelope rather than dropping source
# facts that the consumer must inspect before it spends compute.
"max_json_chars": 20_500,
# Native canary isolation adds a parent directory to every fixture
# path: the same payload is 20,341 chars direct / 20,509 in the runner.
# Retain complete source paths and 491 chars of runner headroom.
"max_json_chars": 21_000,
# Explicitly importing this same fixture exposes the current canonical
# read shape on both base/head: 20,287 chars, 564 nested / 54 top-level
# keys. Preserve source/Task provenance; retain 11 nested and one
Expand Down Expand Up @@ -465,6 +468,17 @@ def main() -> int:
assert_surface("quota_should_run_json", quota_payload),
assert_surface("dashboard_status_json", status_payload),
]
oversized_quota = {
**quota_payload,
"reason": "x" * int(SURFACE_BUDGETS["quota_should_run_json"]["max_json_chars"]),
}
try:
assert_surface("quota_should_run_json", oversized_quota)
except AssertionError as error:
assert error.args[0][0] == "quota_should_run_json", error
assert error.args[0][1] > SURFACE_BUDGETS["quota_should_run_json"]["max_json_chars"], error
else:
raise AssertionError("quota character budget accepted an oversized payload")
cadence = build_interface_budget_cadence(
summaries,
checked_at="2099-01-01T00:10:00+00:00",
Expand Down
Loading