dongarandClaude Opus 5.5
6ae2375194
nisse-dev now runs only the conversational loop. The gated state machine it was built alongside is gone: - Removed control.py, planner.py, executor.py, approvals.py, gates.py (command_form moves to loop/reader.py), publish.py and poller.py, the old assembly in agent.py (its shared helpers move to loop/build.py), the store's state-machine view (get_state, transition), their tests, the old end-to-end and live planning/milestone/gate tests, and specs/orchestrator.qnt with its scenarios. - Roles are driver, helper, expert and classifier; driver is required. loop:, roles.planner, roles.executor and tuning.executor are refused with what to use instead; tuning.driver takes the turn cap and failure limit. - The benchmark and the capability matrix measure the driver: the loop's own system prompt and tools, the task's plan as the agreed restatement. Task plans end with finish_milestone, so earlier runs show as (old); stored runs' executor tuning reads as the driver's. - A delegation on the loop records what its helper's alias resolved to (L9), as the old path's did. - The dump shows a work item's revision and what it waits for. - Behaviour the old tests covered and the loop shares stays covered: driver runs that fail on a model call, stop mid-turn, reset the failure count; go-aheads ignoring case and punctuation; a request too big to fit is never sent; the assembly tests, now on build_loop_agent. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01PPQ14L3LikyefCMFBqEbKE
56 lines
2.2 KiB
YAML
56 lines
2.2 KiB
YAML
# Agent config for the live integration tests, aimed at test resources: the
|
|
# testing identities, the dedicated test room, and the sandbox repo. Never
|
|
# point it at production.
|
|
#
|
|
# Secrets are ${…} references. CI passes them as environment variables;
|
|
# locally they come from .env at the repository root. A missing one fails
|
|
# the live tests that need it, rather than skipping them, so a broken
|
|
# credential cannot hide as a skip.
|
|
|
|
# Unused by the live tests, which work in pytest's temporary directories;
|
|
# present because every agent config has it.
|
|
state_db: /tmp/nisse-test/state.db
|
|
|
|
projects:
|
|
nisse-dev-testing:
|
|
# Placeholders: every live test that touches a repository makes one of
|
|
# its own (tests/integration/e2e_support.py, fresh_repo), owned by the
|
|
# test operator and deleted afterwards, with its own deploy key; and runs
|
|
# on a throwaway local sshd (the `ssh_host` fixture) or the bring-up build
|
|
# VM, in temporary directories.
|
|
repo: nisse-testing-operator/made-per-run
|
|
workspace: /tmp/nisse-test/work
|
|
build_host:
|
|
address: 127.0.0.1
|
|
key: /tmp/nisse-test/unused-key
|
|
|
|
llm:
|
|
base_url: https://litellm.dongar.net
|
|
token: ${NISSE_LLM_TOKEN}
|
|
# The test key caps each request's worst-case cost ($0.10): input plus
|
|
# this much reserved output. The test agent's replies are short, so half
|
|
# the default leaves room for ~40k input tokens on Haiku.
|
|
max_output_tokens: 8192
|
|
|
|
# The live tests exercise the gated loop and the routes over a real proxy,
|
|
# not local-model competence (measured separately by test_model_capability),
|
|
# so every role uses a model that holds the tool-call contract.
|
|
roles:
|
|
driver: claude-haiku-4-5
|
|
helper: claude-haiku-4-5
|
|
expert: claude-haiku-4-5
|
|
|
|
matrix:
|
|
homeserver: https://matrix.dongar.net
|
|
user_id: "@nisse-dev-testing:dongar.net"
|
|
token: ${NISSE_MATRIX_TOKEN}
|
|
room_id: "!UxpXhTlQsFIeNZwkgv:dongar.net" # the test room, never production
|
|
operators: ["@dongar:dongar.net"]
|
|
|
|
gitea:
|
|
url: https://gitea.dongar.net
|
|
token: ${NISSE_GITEA_TOKEN}
|
|
# nisse-testing-operator plays the operator in the end-to-end tests
|
|
# (NISSE_OPERATOR_TOKEN); an operator here and nowhere else.
|
|
operators: [dongar, nisse-testing-operator]
|