nisse-core records each model call of every agent into a file beside its
state store (<state>.calls.db), for the status page (L7, X6):
- the request body exactly as sent (never headers), the raw response with
its reasoning, the model asked and the one that answered, time, tokens
and llama-server's own prompt reuse (timings.cache_n);
- its work item, role, run and purpose: Driver.run and nisse-dev's
delegation loop open runs, nested; compaction, expert consults, the
classifier and the go-ahead reader label their calls;
- provenance, worked out in BudgetedClient by comparing the events at
each stage: each event sent with its log id and content as sent, which
a compaction replaced, which fitting shortened (by how much) and
dropped.
Content is stored once by hash, compressed. It is diagnostic only: the
agent never reads it, a failed write is logged once and recording stops
while the call goes on, and the file is deleted when the agent starts,
so its layout never migrates; CallReader refuses a layout it doesn't
know. While recording, the system prompt ends with one sentence saying
so. `call_record: false` in either agent's YAML turns it off.
The benchmark records each run beside that run's own state store, and
`--keep-runs DIR` keeps both for the page.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01J5Utyfzw7gF7UhjRDEtjTk
The agent works in a clone, but benchmark workspaces were plain
directories, so a model that starts with `git log` failed two commands
in a row and was stopped as stuck within seconds (Flash-Next, 10 of 13
failures on 2026-10-24). The seed is now committed on main, locally and
on the build VM, and how a workspace is set up is part of each task's
hash, so earlier runs are reported as not comparable.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01J5Utyfzw7gF7UhjRDEtjTk
What isn't the dev agent's moves to nisse-core, unchanged in behaviour,
with its tests: canonical, render, render_anthropic, llm, catalog, probe,
context, views, state, reconcile, worker, cancel, dump, transport, matrix,
execution, testing, vault, memory, oplog, envfile and gitea (whole).
- nisse_core.state.StateStore holds the tables every agent has; an
agent's subclass adds its tables (version, migrations), roles and
views. nisse_dev.state.StateStore adds the build VM blocks (v5).
- views.project takes what each role is shown, rather than a fixed table.
- Vault takes the agent's name (agent=), for provenance.
- A test that nisse_core never imports an agent.
- The loop, config, git, delegation, expert, tools and evals stay in
nisse-dev until nisse-ops has a loop to share.
Docs: the Development Plan marks M7.1 done and notes that Phase 6 waits on
the agent host and Terraform; Architecture Part L names the package.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01J5Utyfzw7gF7UhjRDEtjTk
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
execution.SSHHost.execute is now the only way the agent runs anything
(L13). Commands travel as a script on SSH's stdin, so the secrets handed
to a call are never on a command line, and every value handed out is
redacted from the output. Timeouts kill the remote process group, and
an unreachable host is reported as such, not as a failed command.
Workspace and GitRepo run on the build VM. The clone uses the
repository's SSH URL, so no credential is stored in it (#89); fetches
and pushes get the push key over stdin, in a short-lived ssh-agent,
and PRs are still opened from the agent's host with its API token.
The bench runs on the build host too.
Tests reach "the build VM" through a throwaway local sshd (nisse_dev
.testing); CI installs openssh-server. The live push test now pushes
with the sandbox's deploy key (NISSE_PUSH_KEY).
Fixes#89.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PPQ14L3LikyefCMFBqEbKE
All seven newer local models passed the original four tasks 20 of 20,
so the benchmark no longer told them apart. The new tasks keep plans
explicit, as the planner is told to write them, and make the work
itself harder:
- trace-bug: a wrong multi-currency total whose cause is a shared
rate helper; discounts and tax go through it too, so a fix at the
call site fails the check.
- refactor-rename: rename load_config across a ten-module package,
leaving load_config_defaults and reload_config alone.
- spec-parser: implement parse_duration from an eight-rule SPEC.md
given three examples, and write tests for every rule.
- long-milestone: eight dependent steps adding due dates to a task
tracker, from the model through storage versions to the CLI.
Self-tests prove that a call-site fix, a blind text replace and a
partial rename all fail their checks.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
The report grouped runs by model, serving note and task version, so
runs made at different tuning were silently averaged together. Tuning
is now part of the key, and a tuning column shows how a row's settings
differ from today's defaults; it appears only when some row needs it.
tuning.executor.max_turns goes from 24 to 40. On the harder benchmark
tasks Haiku, the calibration ceiling, hit 24 on both refactor-rename
runs and both long-milestone runs, one of which had finished the work.
The cap guards against a model trying forever; it isn't a budget.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
The client waits RATE_LIMIT_RETRY_S and retries once on a 429, silently.
A benchmark run that waited looked slower than the model is: on
2026-09-27, fast models under the unit-test key's 200k tokens/minute
limit had 20-45 s runs that were probably waits.
Both clients now report last_rate_limit_wait_s after every call, even
one that then fails. The benchmark sums it per run into
rate_limit_wait_s, the report's time column leaves it out, and a new
429 column counts the runs that waited. Results stored before this
load unchanged.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Choosing a local executor model, and tuning its knobs, needs measured
success rates across repeated runs, comparable with every model
measured before.
- Four canned milestones under nisse_dev/bench/tasks: write a function,
fix a bug a failing test exposes, thread an option through several
files, and find one failure in output far past max_output_chars. Each
is an approved plan over a seed project, judged by a hidden check the
model never sees; a reference solution proves each is solvable and
that its check tells the difference
- Runs use the agent's own client wiring, now shared as
make_llm_clients, with delegation and ask_expert off: an escalation is
a failure of the model under test
- Each run is stored as one JSON line in bench/results/, committed, with
the model's backing name and declared facts, a note on how it was
served, the task's content hash, the agent's commit and tuning, the
outcome, turns, misbehaviour counts, context trims, tokens, time and
LiteLLM's cost; a fallback answer can never count as a success
- The report rebuilds the comparison from every stored run, with Wilson
95% intervals, marking runs against older task versions
- scripts/benchmark.py runs and reports; scripts/evaluate_model.sh takes
a new model through the context probe, the capability matrix and the
benchmark in one command
- Clients keep LiteLLM's per-call cost, and BudgetedClient counts the
requests it had to trim
- The tuning guide gains "Benchmarking models"
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
dev-agent
dongarandClaude Opus 5.5