Ben Barclay
5a1246f830
Fleet telemetry showed "unknown" as the single largest execution_surface bucket. Two construction paths were mis-attributed, both silently: 1. ACP editor sessions (VS Code / Zed / JetBrains) declare platform="acp", but "acp" was absent from EXECUTION_SURFACES, so the contract's closed-schema fallback folded every editor session into "other" -- the bucket meant for genuinely unclassifiable traffic. 2. batch_runner built agents from _AGENT_PASSTHROUGH, which omitted "platform" entirely, so every batch task run reported "unknown" despite "batch" already being a first-class surface. Neither is a reporting bug in the exporter: both are declaration gaps at the construction site. "unknown" must mean "this run genuinely could not be attributed", not "a construction site forgot to say who it was". Changes: - add "acp" to EXECUTION_SURFACES and map it to the "interactive" entrypoint alongside cli/desktop/tui - add "acp" to the v2 wire schema enum (kept in sync by an existing test) - pass platform through batch_runner: added to _AGENT_PASSTHROUGH, set self.platform = "batch" on the runner, and defaulted at the worker call site so callers that build a config without it stay attributable Wire compatibility: the ingest service validates the envelope only and stores metric bodies verbatim, so packages carrying the new value are accepted by the already-deployed server. No coordinated deploy needed. Tests: 12 new behavioural tests. Verified red before the fix (4 failed), green after. Three fix-mutants confirmed killed: M1 revert acp from EXECUTION_SURFACES -> 3 failed M2 revert acp entrypoint mapping only -> 1 failed M3 revert batch passthrough -> 1 failed No source-text assertions; every test is a contract between the surfaces the schema accepts and the surface each path declares. A guard test pins that a genuinely undeclared run still reports "unknown", so attribution cannot be "fixed" by inventing a default that hides real gaps.
78 lines
3.0 KiB
Python
78 lines
3.0 KiB
Python
"""Every AIAgent construction path must attribute its execution surface.
|
|
|
|
The shared-metrics contract normalises a missing or unrecognised ``platform``
|
|
to ``unknown`` / ``other``. That is the correct behaviour for a *bounded*
|
|
schema, but it means a construction site that forgets to declare its surface
|
|
is silently mis-attributed rather than loudly broken: fleet telemetry then
|
|
reports "unknown" for real, attributable traffic.
|
|
|
|
These are contract tests between two pieces of data -- the surfaces the
|
|
contract accepts, and the surface each construction path actually declares --
|
|
not snapshots of any current value.
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import pytest
|
|
|
|
from hermes_cli.observability import shared_metrics_contract as contract
|
|
|
|
|
|
def test_acp_editor_sessions_get_their_own_surface():
|
|
"""ACP (VS Code / Zed / JetBrains) is a real interactive surface, not 'other'.
|
|
|
|
The ACP adapter declares ``platform="acp"``. If that value is not an
|
|
accepted surface, the contract's closed-schema fallback buckets every
|
|
editor session into ``other`` alongside genuinely unclassifiable traffic.
|
|
"""
|
|
assert contract.execution_surface({"platform": "acp"}) == "acp"
|
|
|
|
|
|
def test_acp_sessions_are_interactive():
|
|
"""An editor session is a human at a keyboard, like cli/tui/desktop."""
|
|
fields = contract.task_start_fields({"platform": "acp"})
|
|
assert fields["entrypoint"] == "interactive"
|
|
assert fields["execution_surface"] == "acp"
|
|
|
|
|
|
def test_batch_runs_declare_their_surface():
|
|
"""batch_runner builds agents from a fixed passthrough tuple.
|
|
|
|
``batch`` is already an accepted surface, so the only defect is that the
|
|
runner never declares it -- every batch task run reports 'unknown'.
|
|
"""
|
|
import batch_runner
|
|
|
|
assert "platform" in batch_runner._AGENT_PASSTHROUGH, (
|
|
"batch_runner._AGENT_PASSTHROUGH omits 'platform', so batch task runs are "
|
|
f"attributed to {contract.execution_surface({})!r} despite 'batch' being a "
|
|
"valid execution surface"
|
|
)
|
|
|
|
|
|
@pytest.mark.parametrize(
|
|
"platform",
|
|
["cli", "tui", "desktop", "batch", "acp", "api_server", "cron", "telegram"],
|
|
)
|
|
def test_declared_platforms_resolve_to_a_named_surface(platform):
|
|
"""No production construction path should resolve to unknown/other.
|
|
|
|
'unknown' must mean "this run genuinely could not be attributed", not
|
|
"a construction site forgot to say who it was".
|
|
"""
|
|
surface = contract.execution_surface({"platform": platform})
|
|
assert surface not in {"unknown", "other"}, (
|
|
f"platform={platform!r} resolves to {surface!r}; a real surface is being "
|
|
"folded into the catch-all bucket"
|
|
)
|
|
|
|
|
|
def test_unattributed_runs_still_report_unknown():
|
|
"""The catch-all must survive: a genuinely undeclared run is 'unknown'.
|
|
|
|
This is the counterpart to the tests above -- fixing attribution must not
|
|
be achieved by inventing a default that hides real gaps.
|
|
"""
|
|
assert contract.execution_surface({}) == "unknown"
|
|
assert contract.task_start_fields({})["entrypoint"] == "unknown"
|