Add harness.agent-dev-local@1.1.1 on claude-sonnet-5; block 1.1.0
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Assistant: claude-code Assistant-Model: opus Assistant-Process: 226514@bnt-lap001 Assistant-Session: 26ba103d-05fe-45a1-9cd7-9475bf239df6
This commit is contained in:
parent
e17e2eb990
commit
30be818081
5 changed files with 53 additions and 7 deletions
|
|
@ -5,7 +5,7 @@ Live residual: `GLAS-IN-0002`. Owner dependency: `SAND-WP-0014-T05`.
|
|||
|
||||
Start from `harness.agent-dev-local@1.0.0`: rein-aharness / Claude Code /
|
||||
`claude-sonnet-4-6` / `green-commit-only`. The future proof candidate is
|
||||
`harness.agent-dev-local@1.1.0`; it is not yet in the catalog because its runtime,
|
||||
`harness.agent-dev-local@1.1.1` (1.1.0 superseded before use); it is not yet in the catalog because its runtime,
|
||||
sandbox profile, and credential route are not resolved.
|
||||
|
||||
## Reviewed baseline, 2026-09-05
|
||||
|
|
@ -130,7 +130,7 @@ After the owner requirements and candidate review pass, run:
|
|||
|
||||
```bash
|
||||
.venv/bin/python scripts/prove-local-profile.py \
|
||||
--harness-profile harness.agent-dev-local@1.1.0
|
||||
--harness-profile harness.agent-dev-local@1.1.1
|
||||
```
|
||||
|
||||
Use `--profile-dir <reviewed-directory>` if the versioned candidate is staged
|
||||
|
|
|
|||
|
|
@ -3,8 +3,8 @@ version: "1.1.0"
|
|||
contract_version: "1.0"
|
||||
status: enabled
|
||||
operational_readiness:
|
||||
status: unverified
|
||||
reason: bounded GLAS-WP-0012-T04 proof candidate; no positive real-run evidence yet
|
||||
status: blocked
|
||||
reason: superseded by 1.1.1 (claude-sonnet-5) before any real run; runtime 2cfc7b5f bundles the earlier unverified copy and is not used
|
||||
owner: glas-harness
|
||||
evidence_ref: GLAS-WP-0012-T03
|
||||
rein:
|
||||
|
|
|
|||
37
profiles/harness.agent-dev-local-1.1.1.yaml
Normal file
37
profiles/harness.agent-dev-local-1.1.1.yaml
Normal file
|
|
@ -0,0 +1,37 @@
|
|||
id: harness.agent-dev-local
|
||||
version: "1.1.1"
|
||||
contract_version: "1.0"
|
||||
status: enabled
|
||||
operational_readiness:
|
||||
status: unverified
|
||||
reason: bounded GLAS-WP-0012-T04 proof candidate; no positive real-run evidence yet
|
||||
owner: glas-harness
|
||||
evidence_ref: GLAS-WP-0012-T03
|
||||
rein:
|
||||
id: rein-aharness
|
||||
required_capabilities:
|
||||
session_style: unattended
|
||||
model_class: frontier
|
||||
sandbox_profile: profile.bwrap-local
|
||||
tool_profile: green-commit-only
|
||||
model:
|
||||
provider: anthropic
|
||||
model: claude-sonnet-5
|
||||
model_class: frontier
|
||||
route: claude-code-cli
|
||||
limits:
|
||||
budget_tokens: 60000
|
||||
timeout_seconds: 900
|
||||
max_turns: 30
|
||||
max_budget_usd: 5.0
|
||||
credential_route_refs: []
|
||||
metadata:
|
||||
latency_class: fast
|
||||
stream_tool_events: true
|
||||
# Candidate for the metered one-cycle owner (rein-aharness metered-once).
|
||||
# The provider key stays with MessagesOwner outside bwrap; the workload gets
|
||||
# only an owner-socket route, so the sandbox keeps empty egress and no
|
||||
# credential route refs. max_budget_usd matches the accepted spend envelope
|
||||
# hfact-glas-anthropic-2026-09 (infd-20260921-b01). Model claude-sonnet-5 is
|
||||
# the operator's 2026-09-23 choice; it supersedes 1.1.0 (claude-sonnet-4-6).
|
||||
# Rollback: pin 1.0.0; 1.1.0 is blocked.
|
||||
|
|
@ -67,7 +67,7 @@ def test_committed_catalog_resolves_both_constellations() -> None:
|
|||
|
||||
contexts = catalog.validate_all()
|
||||
|
||||
assert len(contexts) == 4
|
||||
assert len(contexts) == 5
|
||||
assert {context.rein_id for context in contexts} == {
|
||||
"rein-aharness",
|
||||
"rein-openweights",
|
||||
|
|
@ -77,7 +77,8 @@ def test_committed_catalog_resolves_both_constellations() -> None:
|
|||
for context in contexts
|
||||
} == {
|
||||
("harness.agent-dev-local", "1.0.0", "blocked"),
|
||||
("harness.agent-dev-local", "1.1.0", "unverified"),
|
||||
("harness.agent-dev-local", "1.1.0", "blocked"),
|
||||
("harness.agent-dev-local", "1.1.1", "unverified"),
|
||||
("harness.agent-dev-openweights-local", "1.0.0", "blocked"),
|
||||
("harness.agent-dev", "1.0.0", "unverified"),
|
||||
}
|
||||
|
|
@ -101,8 +102,9 @@ def test_committed_catalog_resolves_both_constellations() -> None:
|
|||
|
||||
def test_metered_candidate_carries_native_limits_and_no_credential_route() -> None:
|
||||
catalog = ProfileCatalog()
|
||||
profile, _ = catalog.resolve("harness.agent-dev-local@1.1.0")
|
||||
profile, _ = catalog.resolve("harness.agent-dev-local@1.1.1")
|
||||
catalog.require_operational(profile)
|
||||
assert profile.model.model == "claude-sonnet-5"
|
||||
assert profile.operational_readiness.status == "unverified"
|
||||
assert profile.sandbox_profile == "profile.bwrap-local"
|
||||
assert profile.credential_route_refs == []
|
||||
|
|
|
|||
|
|
@ -381,3 +381,10 @@ Source return: 3e49a98a0e2c4a64539a5ccd674be8cd67ac9845; receipt in
|
|||
T02 remains wait for trusted owner execution configuration, native credentials
|
||||
and real-provider evidence. T03-T06 retain candidate review, real-model run,
|
||||
readiness and recovery obligations. No profile readiness or schedule changes.
|
||||
|
||||
2026-09-23 candidate superseded before use: operator chose `claude-sonnet-5`.
|
||||
`harness.agent-dev-local@1.1.1` is identical to 1.1.0 except the model; 1.1.0 is
|
||||
now `blocked`. Runtime `2cfc7b5f` bundles the earlier 1.1.0 copy (profile sha
|
||||
`e84b9d0f…`) and must be rebuilt with this revision and placed again
|
||||
(SECRETS-WP-0009-T03). Claude CLI 2.1.266 in that runtime lists
|
||||
`claude-sonnet-5` in its model registry (static check). 134 tests pass.
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue