simjecture 0.5.2rc2__tar.gz → 0.5.3rc2__tar.gz

{simjecture-0.5.2rc2 → simjecture-0.5.3rc2}/.gitignore

@@ -20,6 +20,8 @@ artifacts/
 !demos/agent_roles_validation/record/workspace/artifacts/
 !demos/agent_roles_validation/record/workspace/artifacts/*.json
 data/
+!src/conjecture_solver/llm_bench/data/
+!src/conjecture_solver/llm_bench/data/*
 .private/
 
 integrations/dsh/node_modules/

{simjecture-0.5.2rc2 → simjecture-0.5.3rc2}/CHANGELOG.md

@@ -2,6 +2,92 @@
 
 This project follows semantic versioning. Dates use ISO 8601.
 
+## 0.5.3rc2 (preview) — 2026-09-30
+
+- Polish Research tools with compact independent cards and a separate details
+  panel. Keep paths, diagnostics and logs accessible without expanding grid rows;
+  distinguish detected local builds from unavailable managed-runtime profiles.
+- Add standalone local/SSH headless numerical workers for minimal studies, with
+  frozen runtime identities, CPU/RAM/GPU admission and asynchronous execution.
+- Preserve job identity across lost replies, SSH interruptions and coordinator
+  restarts; enforce worker deadlines and confirm descendant cancellation.
+- Add the Machines workspace page, visible study pool selection, placement and
+  transport status, bounded read-only source access and verified artifact retrieval.
+- Simplify SSH onboarding to address and password with automatic hardware/runtime
+  detection, optional advanced settings, recorded agent setup commands and independent
+  30-second availability heartbeats. Polish machine cards, jobs and mobile navigation.
+- Install the matching released package on headless workers instead of a fixed
+  older candidate.
+- Preserve pool selection when preparing continuation phases. Keep research agents
+  and independent scientific review central; Simote remains optional.
+
+
+## 0.5.3rc1 (preview) — 2026-09-30
+
+First repairs from the aluminium stagnation campaign audit. Minimal remains the
+default; this preview does not change worker session policy or claim scientific
+improvements from a new long campaign.
+
+- Recognize provider `402 Insufficient Balance` errors in native result envelopes.
+  Pause with a credit/quota explanation and keep Resume and the original deadline.
+- Explain and evaluate alternative instrument requirements as **any of**, in methods
+  review, claim review, study status and the browser; FLASH or WarpX never means both.
+- Bound long plan notes before evicting current evidence. Preserve references to the
+  newest completed experiment, method and claim review even in small recovery briefs.
+- Let methods/progress reviewers request recorded JSON excerpts through the host,
+  with hash checks, bounded size and at most two retrieval rounds. Requests and failed
+  retrievals cannot grant approval; record each review packet and response.
+- Record builtin provider requests before agent-step aggregation, including completion
+  retries, failures, native cache/reasoning usage when reported, and context sizes/hashes.
+  Preserve unknown counters and avoid adding request and legacy step totals together.
+- Show live request counts, worker/reviewer token breakdown, missing usage, provider
+  retry time and actionable provider errors in terminal/browser views. Monetary cost
+  is explicitly unavailable without billing data; no price assumptions are invented.
+- Preserve native tool-call/result messages in builtin DeepSeek worker conversations,
+  including private reasoning continuity. Nineteen controlled live repair trials
+  informed this choice; retain the completion guard, session policy and review gates.
+  Record response finish reasons, call counts and text trimming for further audits.
+- Ship experimental **Simjecture Bench 0.1.0**, independently versioned CSV and
+  recorded-RZ diagnostic coding tasks, portable CLI preparation/grading and Harbor
+  task-format export. Verify varied numerical holdouts, saved outputs, input hashes
+  and numerical plot data; keep scientific claim approval separate.
+- Add a visible Benchmarks workspace entry for preparation with the chosen agent
+  and independent grading after delivery. Preparation makes no provider request.
+  Timed framework trials use the external runner; interactive chat is exploratory.
+- Document the matched MiMo 2.6 Flash/Pro and DeepSeek V4.1 Flash pilot, including
+  cache usage, continuation/setup differences and limits of single-trial evidence.
+
+See [audit and next phases](docs/research/stagnation-deep-audit-20260929.md) and
+[preview scope](docs/testing/0.5.3rc1-acceptance.md), and
+[adapter benchmark](docs/testing/native-api-adapter.md).
+
+## 0.5.2 — 2026-09-29
+
+Stable release of the conversation-first research workspace tested in rc1/rc2,
+plus continuation, steering, review recovery and the ITER diagnostics pack.
+Minimal mode remains the default; structured/frontier modes remain selectable.
+The versioned installer preserves existing research data and older program folders.
+
+- Add an optional ITER diagnostics/data pack with pinned CHERAB/Raysect,
+  CHERAB-ITER/IMAS extensions and IMAS tools. Install/check from the CLI or GUI;
+  run a visible, monitored numerical demo with plots. Add source-based agent
+  guidance and separate setup entries for JOREK, SOLPS-ITER and DINA-PS.
+
+- Offer Prepare directly or Prepare with agent from the continuation dialog.
+  Chat preparation carries parent context across turns and lets the agent draft
+  the brief and select inherited files without losing lineage or instruments.
+- Add Continue investigation and Send guidance to study cards and the experiment
+  monitor. Review a linked phase's brief, selected inherited files, model and new
+  budget before launch; retain the original study and evidence status.
+- Snapshot continuation context with verified hashes and immutable parent lineage.
+  Add CLI continuation and advisory steering; show queued/checkpoint delivery in
+  the web interface and include guidance in independent review packets.
+- Detect empty/truncated built-in review replies, record finish reasons and usage,
+  and use bounded recovery with a larger output allocation and a DeepSeek
+  non-thinking retry. Back off repeated invalid oversight without granting approval.
+- Reconcile expired deadlines/dead supervisors in live status and write final
+  reports when a supervisor exits through an external error.
+
 ## 0.5.2rc2 (preview) — 2026-09-27
 
 - Provision scientific compilers, MPI, HDF5 and other prerequisites in managed

{simjecture-0.5.2rc2 → simjecture-0.5.3rc2}/CITATION.cff

@@ -2,8 +2,8 @@ cff-version: 1.2.0
 message: "If you use this software, please cite the exact software version and Git commit."
 title: "Simjecture"
 type: software
-version: 0.5.2rc2
-date-released: 2026-09-27
+version: 0.5.3rc2
+date-released: 2026-09-30
 abstract: >-
   An evidence-governed runtime for autonomous computational experimentation and
   falsification over hypothesis trees.

{simjecture-0.5.2rc2 → simjecture-0.5.3rc2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: simjecture
-Version: 0.5.2rc2
+Version: 0.5.3rc2
 Summary: Evidence-governed autonomous experimentation and falsification for computational science
 Project-URL: Homepage, https://github.com/tomzhu0225/simjecture
 Project-URL: Documentation, https://github.com/tomzhu0225/simjecture/tree/main/docs
@@ -52,18 +52,47 @@ experiments, searches for counterexamples, and proposes small, testable repairs
 failed hypotheses. The harness records what actually ran and requires independent
 review before a scientific claim can close.
 
-**v0.5.2rc2 is a preview release for user testing; v0.5.1 remains stable.** Minimal mode is the default: agents keep their
+**v0.5.2 is the current stable release.** Minimal mode is the default: agents keep their
 native tools and choose their research strategy, while Simjecture manages recorded
 evidence, review and deadlines. Structured and frontier modes remain available.
 The system does not establish physical truth from a successful simulation or promise
 that a long agent run will produce a useful result.
 
-## Install and open the workspace preview
+The workspace includes visible **Continue investigation** and **Send
+guidance** actions in the study view and experiment monitor. Continue with a new
+budget and selected prior files, or send advisory guidance at the next checkpoint.
+See [continuation and steering](docs/how-to/continuation-steering.md).
+
+The **0.5.3rc2 preview** adds local/SSH experiment workers and refreshed research
+interfaces. In **Machines**, connect with an SSH address and password; hardware and
+worker setup are detected automatically, with optional advanced settings and agent
+preparation. Availability updates every 30 seconds. Minimal studies can dispatch
+parallel experiments with GPU reservations and recover existing jobs after SSH
+interruptions or coordinator restarts. Agents and independent review stay on the
+coordinator; Simote remains optional.
+
+Research tools now uses compact cards with build lists, diagnostics and logs in a
+separate Details panel. The preview also includes rc1's campaign audit repairs,
+request-level usage records, native DeepSeek tool-message continuity and experimental
+**Simjecture Bench 0.1.0**. See [SSH setup](docs/how-to/ssh-workers.md),
+[benchmark usage](docs/how-to/llm-bench.md), and
+[rc2 scope and validation](docs/testing/0.5.3rc2-acceptance.md).
+The stable installation commands below remain on 0.5.2.
+
+To try rc2, preserving existing research data:
+
+```bash
+curl -fsSL https://github.com/tomzhu0225/simjecture/releases/download/v0.5.3rc2/install.sh | bash
+# Or install only the Python package:
+uv tool install 'simjecture[workspace,tui]==0.5.3rc2'
+```
+
+## Install and open the workspace
 
 On Linux or inside WSL, run:
 
 ```bash
-curl -fsSL https://github.com/tomzhu0225/simjecture/releases/download/v0.5.2rc2/install.sh | bash
+curl -fsSL https://github.com/tomzhu0225/simjecture/releases/download/v0.5.2/install.sh | bash
 ```
 
 The installer downloads the versioned release, verifies its checksum, installs uv,
@@ -88,7 +117,7 @@ Use Linux with Python 3.11+, [uv](https://docs.astral.sh/uv/), and an installed,
 authenticated native agent CLI. Bubblewrap is the default numerical execution backend.
 
 ```bash
-uv tool install 'simjecture[tui]==0.5.2rc2'
+uv tool install 'simjecture[tui]==0.5.2'
 simjecture install core
 simjecture doctor --execution-backend bubblewrap
 
@@ -208,6 +237,13 @@ simjecture install warpx-cpu
 simjecture doctor --profile warpx-cpu
 ```
 
+The release also includes an [ITER ecosystem pack](docs/how-to/iter-pack.md):
+CHERAB/Raysect synthetic diagnostics, CHERAB-ITER/IMAS adapters and IMAS data tools.
+Use **Research tools → ITER pack → Install**, then **Run demo** for a monitored
+example with plots, or run `simjecture install iter-pack`. JOREK, SOLPS-ITER and
+DINA-PS have separate **Install with agent** entries and source-based guidance;
+they are not silently included in the diagnostics installation.
+
 FLASH source is not redistributed. Solver source inspection can be configured separately
 from a writable problem/build area. See [runtime deployment](docs/how-to/deploy-runtimes.md)
 and the [scientific skills](skills/).

{simjecture-0.5.2rc2 → simjecture-0.5.3rc2}/README.md

@@ -9,18 +9,47 @@ experiments, searches for counterexamples, and proposes small, testable repairs
 failed hypotheses. The harness records what actually ran and requires independent
 review before a scientific claim can close.
 
-**v0.5.2rc2 is a preview release for user testing; v0.5.1 remains stable.** Minimal mode is the default: agents keep their
+**v0.5.2 is the current stable release.** Minimal mode is the default: agents keep their
 native tools and choose their research strategy, while Simjecture manages recorded
 evidence, review and deadlines. Structured and frontier modes remain available.
 The system does not establish physical truth from a successful simulation or promise
 that a long agent run will produce a useful result.
 
-## Install and open the workspace preview
+The workspace includes visible **Continue investigation** and **Send
+guidance** actions in the study view and experiment monitor. Continue with a new
+budget and selected prior files, or send advisory guidance at the next checkpoint.
+See [continuation and steering](docs/how-to/continuation-steering.md).
+
+The **0.5.3rc2 preview** adds local/SSH experiment workers and refreshed research
+interfaces. In **Machines**, connect with an SSH address and password; hardware and
+worker setup are detected automatically, with optional advanced settings and agent
+preparation. Availability updates every 30 seconds. Minimal studies can dispatch
+parallel experiments with GPU reservations and recover existing jobs after SSH
+interruptions or coordinator restarts. Agents and independent review stay on the
+coordinator; Simote remains optional.
+
+Research tools now uses compact cards with build lists, diagnostics and logs in a
+separate Details panel. The preview also includes rc1's campaign audit repairs,
+request-level usage records, native DeepSeek tool-message continuity and experimental
+**Simjecture Bench 0.1.0**. See [SSH setup](docs/how-to/ssh-workers.md),
+[benchmark usage](docs/how-to/llm-bench.md), and
+[rc2 scope and validation](docs/testing/0.5.3rc2-acceptance.md).
+The stable installation commands below remain on 0.5.2.
+
+To try rc2, preserving existing research data:
+
+```bash
+curl -fsSL https://github.com/tomzhu0225/simjecture/releases/download/v0.5.3rc2/install.sh | bash
+# Or install only the Python package:
+uv tool install 'simjecture[workspace,tui]==0.5.3rc2'
+```
+
+## Install and open the workspace
 
 On Linux or inside WSL, run:
 
 ```bash
-curl -fsSL https://github.com/tomzhu0225/simjecture/releases/download/v0.5.2rc2/install.sh | bash
+curl -fsSL https://github.com/tomzhu0225/simjecture/releases/download/v0.5.2/install.sh | bash
 ```
 
 The installer downloads the versioned release, verifies its checksum, installs uv,
@@ -45,7 +74,7 @@ Use Linux with Python 3.11+, [uv](https://docs.astral.sh/uv/), and an installed,
 authenticated native agent CLI. Bubblewrap is the default numerical execution backend.
 
 ```bash
-uv tool install 'simjecture[tui]==0.5.2rc2'
+uv tool install 'simjecture[tui]==0.5.2'
 simjecture install core
 simjecture doctor --execution-backend bubblewrap
 
@@ -165,6 +194,13 @@ simjecture install warpx-cpu
 simjecture doctor --profile warpx-cpu
 ```
 
+The release also includes an [ITER ecosystem pack](docs/how-to/iter-pack.md):
+CHERAB/Raysect synthetic diagnostics, CHERAB-ITER/IMAS adapters and IMAS data tools.
+Use **Research tools → ITER pack → Install**, then **Run demo** for a monitored
+example with plots, or run `simjecture install iter-pack`. JOREK, SOLPS-ITER and
+DINA-PS have separate **Install with agent** entries and source-based guidance;
+they are not silently included in the diagnostics installation.
+
 FLASH source is not redistributed. Solver source inspection can be configured separately
 from a writable problem/build area. See [runtime deployment](docs/how-to/deploy-runtimes.md)
 and the [scientific skills](skills/).

{simjecture-0.5.2rc2 → simjecture-0.5.3rc2}/THIRD_PARTY_NOTICES.md

@@ -29,3 +29,11 @@ Model providers, literature services, and externally supplied guided
 commissioning packages are services or inputs rather than sublicensed parts of
 this source distribution. Run artifacts must retain the provenance and license
 information supplied by their generators.
+
+The optional ITER ecosystem pack downloads CHERAB, Raysect, CHERAB-IMAS,
+CHERAB-ITER, IMAS-Python, IMAS-Validator and their dependencies under their
+respective upstream licenses. Solver guides refer to separately distributed
+JOREK, SOLPS-ITER/EIRENE and DINA-PS. Their component code, data, citation and
+redistribution terms remain upstream terms; Simjecture does not relicense them
+or imply ITER endorsement. The diagnostic demo is independently authored using
+public APIs; its CHERAB bremsstrahlung API reference is credited in the script.

simjecture-0.5.3rc2/capabilities/iter-pack-1.0.json

@@ -0,0 +1,32 @@
+{
+  "environment": {
+    "MPLBACKEND": "Agg",
+    "OMP_NUM_THREADS": "1",
+    "OPENBLAS_NUM_THREADS": "1",
+    "PYTHONDONTWRITEBYTECODE": "1"
+  },
+  "executable": "bin/python",
+  "identity_files": [
+    "share/build-record.json",
+    "share/installed-requirements.txt"
+  ],
+  "manifest": {
+    "description": "CHERAB/Raysect synthetic diagnostics and IMAS data tools; does not include JOREK, SOLPS or DINA binaries",
+    "executable_kind": "python-iter-pack",
+    "name": "iter-pack-1.0",
+    "preflight_checks": [
+      "checks.completed",
+      "checks.chord_integral",
+      "checks.density_scaling",
+      "checks.imas_roundtrip",
+      "checks.imas_bolometers",
+      "checks.iter_geometry"
+    ],
+    "preflight_resource": "examples/diagnostics_demo.py",
+    "preflight_result": "iter_pack_demo.json",
+    "schema_version": "0.1.0",
+    "skill": "iter-pack",
+    "version": "1.0"
+  },
+  "runtime_root": "../.runtime/iter-pack-1.0"
+}

simjecture-0.5.3rc2/docs/_static/architecture/ssh-workers.svg

@@ -0,0 +1,22 @@
+<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 980 260" role="img" aria-labelledby="title desc">
+<title id="title">One research coordinator and independent execution workers</title>
+<desc id="desc">The research agent uses a central Lab service, which dispatches experiments to local and SSH workers. Hashed results return to central evidence records and independent review.</desc>
+<defs><marker id="arrow" markerWidth="8" markerHeight="8" refX="7" refY="4" orient="auto"><path d="M0 0L8 4L0 8" fill="#6d60bd"/></marker></defs>
+<rect width="980" height="260" rx="12" fill="#f8f8fc"/>
+<g fill="white" stroke="#d6d3e9" stroke-width="2">
+<rect x="20" y="90" width="170" height="70" rx="10"/>
+<rect x="245" y="65" width="260" height="125" rx="10"/>
+<rect x="595" y="15" width="195" height="58" rx="10"/>
+<rect x="595" y="99" width="195" height="58" rx="10"/>
+<rect x="595" y="183" width="195" height="58" rx="10"/>
+</g>
+<g fill="none" stroke="#6d60bd" stroke-width="2" marker-end="url(#arrow)">
+<path d="M190 125H240"/><path d="M505 92L590 44"/><path d="M505 127H590"/><path d="M505 162L590 212"/>
+</g>
+<g font-family="sans-serif" font-size="18" fill="#353144" text-anchor="middle">
+<text x="105" y="130">Research agent</text><text x="375" y="98" font-weight="bold">Coordinator / Lab</text>
+<text x="375" y="133" font-size="15">Hypothesis, methods, receipts</text><text x="375" y="162" font-size="15">Evidence and independent review</text>
+<text x="692" y="50">Local worker</text><text x="692" y="134">SSH worker A</text><text x="692" y="218">SSH worker B</text>
+<text x="880" y="118" font-size="15">Hashed results</text><text x="880" y="144" font-size="15">return to Lab</text>
+</g>
+</svg>

{simjecture-0.5.2rc2 → simjecture-0.5.3rc2}/docs/conf.py

@@ -3,7 +3,7 @@ from __future__ import annotations
 project = "Simjecture"
 author = "Bowen Zhu"
 copyright = "2026, Bowen Zhu and contributors"
-release = "0.5.2rc2"
+release = "0.5.3rc2"
 
 extensions = [
     "myst_parser",

{simjecture-0.5.2rc2 → simjecture-0.5.3rc2}/docs/getting-started/installation.md

@@ -5,10 +5,20 @@
 On Linux or inside an existing WSL distribution:
 
 ```bash
-curl -fsSL https://github.com/tomzhu0225/simjecture/releases/download/v0.5.2rc2/install.sh | bash
+curl -fsSL https://github.com/tomzhu0225/simjecture/releases/download/v0.5.2/install.sh | bash
 ```
 
-This command explicitly selects preview **0.5.2rc2**. Stable **0.5.1** remains unchanged.
+This command selects stable **0.5.2**. Existing research data and older program versions are preserved.
+
+For the **0.5.3rc2 preview**, including SSH workers and refreshed interfaces:
+
+```bash
+curl -fsSL https://github.com/tomzhu0225/simjecture/releases/download/v0.5.3rc2/install.sh | bash
+```
+
+The preview installs beside older versions. It retains projects, provider settings
+and runtime directories. See [rc2 scope and checks](../testing/0.5.3rc2-acceptance.md)
+and [SSH worker setup](../how-to/ssh-workers.md).
 
 No Git, Python environment or CLI agent setup is required in advance. The bootstrap
 fetches the versioned workspace bundle and verifies its SHA-256 checksum, installs uv
@@ -31,7 +41,7 @@ remove older versions or research outputs.
 To install without starting, or choose a different web port:
 
 ```bash
-curl -fsSL https://github.com/tomzhu0225/simjecture/releases/download/v0.5.2rc2/install.sh | bash -s -- --no-start
+curl -fsSL https://github.com/tomzhu0225/simjecture/releases/download/v0.5.2/install.sh | bash -s -- --no-start
 ~/simjecture/start-workspace --port 8765
 ```
 
@@ -57,7 +67,8 @@ for configuration, build, registration and actual backend readiness checks.
 
 ### SSH servers
 
-The GUI and computations run on the SSH server. The installer prints a forwarding
+The workspace runs on the SSH server. Experiments run there by default; rc2 can
+select additional SSH workers through Machines. The installer prints a forwarding
 command; run it on your local computer, for example:
 
 ```bash

{simjecture-0.5.2rc2 → simjecture-0.5.3rc2}/docs/getting-started/research-workspace.md

@@ -152,6 +152,17 @@ belong to that project.
 
 ## Research tools
 
+The catalogue uses compact cards for installed and optional tools. **Details**
+opens a separate panel for build variants, paths, setup diagnostics and logs;
+expanding that information does not stretch neighboring cards.
+
+A **Detected** local build is present but still needs a registered capability
+to participate in recorded studies. An unavailable optional managed runtime is
+shown in its scoped setup diagnostics; it does not imply that detected local
+builds are missing. **Connect existing tool** opens the registration form.
+Install, readiness-check, demonstration and agent-assisted setup actions retain
+their existing behavior.
+
 The catalogue reuses `simjecture install` and `simjecture doctor` implementations.
 **Installed** and **Not installed** are separate sections. Installed instruments
 also show whether a readiness check has passed, failed, or has not yet been run.

simjecture-0.5.3rc2/docs/how-to/continuation-steering.md

@@ -0,0 +1,97 @@
+# Continue and steer an investigation
+
+Available in the rc3 development preview for minimal studies. The published rc2
+installer does not yet include these actions.
+
+## Continue after a deadline or change direction
+
+Open the study in **Autonomous research**, or open its **experiment monitor**.
+Choose **Continue investigation**, describe the follow-up, choose a new wall-time
+budget and select working files to inherit. Choose **Prepare directly** to review the normal study brief and agent/model
+selection, then **Start research**.
+Preparation alone starts neither an agent nor a simulation.
+
+The new study links to its parent. Selected programs/readers are copied to
+`research/inherited/`. A compact `continuation_input/brief.json` summarizes counts, recent unreviewed
+notes and pending reviews. A hash-verified snapshot outside the writable research
+folder preserves the original files and a record summary. Large raw results stay
+at the parent's receipt paths; move/copy them explicitly if using a different
+machine. References do not grant new scientific acceptance. Prior exploration stays
+exploration, and old observations cannot satisfy a new prospective commitment.
+No prior method approval is automatically imported. Reuse its scoped validation
+as context when explaining the next method, instead of repeating irrelevant tests.
+
+The original hypothesis, instructions, reviews and deadline are unchanged. Each
+phase has a separate budget and may select a different agent/model. Required
+instrument rules are inherited unless explicitly supplied for a new CLI phase.
+The parent should remain available for receipt and raw-data inspection. If it is
+still active, the handoff captures recorded context at the new phase's launch;
+later parent results are not silently imported.
+
+CLI equivalent:
+
+```bash
+simjecture study --campaign runs/phase-2 --continue-from runs/phase-1 \
+  --instructions-file next-phase.md --inherit-file calculation.py \
+  --inherit-file analysis.py --backend builtin --model deepseek-flash \
+  --provider-config /private/provider.json --wall-seconds 7200
+```
+
+Omit `--hypothesis-file` to retain the original question. Specify a different one
+for an explicitly revised question. Configure the execution backend appropriate
+to the host. For a new phase the backend is checked afresh. Selected files are
+limited to128files,1MiB each, with a16MiB handoff including metadata. Hidden
+configuration, known credential filenames, symlinks and `lab.py` are excluded.
+
+## Prepare the continuation with an agent
+
+The continuation dialog also offers **Prepare with agent**. You can leave the
+follow-up description blank if you want to discuss the direction first. Select
+initial files and a proposed budget, then choose that button to open the chat.
+
+The interactive agent receives a parent-study context file: original question,
+previous instructions, limitations and review state, selected files, your guidance
+and the proposed budget. It can ask focused questions and call `draft_study` to
+prepare or revise the continuation brief. It may refine the inherited file list;
+the host checks every selected path against the parent's research workspace.
+Parent lineage and the inherited instrument registry survive later chat turns
+and agent-written briefs. Prior observations remain context, not fresh evidence.
+
+**Prepare with agent** also appears beside the parent-study link in an existing
+continuation conversation, so you can discuss a directly prepared draft later.
+If no model is configured, the parent selection is retained: choose the agent in
+chat and use that button again. The resulting brief appears in **Autonomous
+research**. Preparing in chat uses the selected interactive agent, but starts no
+autonomous research allocation; **Start research** remains a separate action.
+
+## Give advice during a study
+
+Choose **Send guidance** beside the study controls. Add an observation, a paper URL,
+an alternative interpretation or a suggested experiment. It is queued durably for
+the next agent checkpoint. **Operator guidance** shows its timestamp and whether
+it has been included in a worker prompt. Delivery does not mean the agent has
+implemented the suggestion. Recent guidance is repeated after session recovery,
+and independent reviewers also receive the advice with its scope and timestamp.
+
+```bash
+simjecture steer --campaign runs/phase-2 \
+  --message "Measure boundary Poynting flux independently; do not infer it from energy closure."
+```
+
+Guidance is advisory. It does not rewrite commitments, change acceptance rules,
+extend time or interrupt active simulations. Use the existing Pause/Stop controls
+for execution control. For new requirements, a new hypothesis or more time, use
+**Continue investigation**. This keeps the original contracts auditable while
+leaving the research agent free to plan its next experiments.
+
+## Review and status recovery
+
+The built-in reviewer now records whether output was empty or truncated and its
+provider finish reason. It makes at most two transport attempts per invocation;
+a DeepSeek retry disables thinking after an empty/truncated response. Usage includes
+both attempts. Persistent failure remains an error, never approval. Invalid
+oversight retries back off up to30minutes while other research can continue.
+
+Live status reconciles expired budgets and recorded supervisor identity. Final
+reports are written on supervisor exit, including an external-error exit. A
+continuation does not erase or silently resolve a parent's pending review.

simjecture-0.5.3rc2/docs/how-to/iter-pack.md

@@ -0,0 +1,75 @@
+# ITER ecosystem pack
+
+Simjecture 0.5.2 integrates open-source fusion tools without requiring a
+particular model backend. Installation and numerical readiness are deterministic;
+agent-guided preparation uses the user's selected model, including DeepSeek Flash.
+This is a Simjecture integration, not an ITER-endorsed distribution.
+
+## Install and run
+
+In **Research tools**, select **ITER pack · diagnostics & data → Install**.
+The installer creates a separate managed environment and runs its numerical
+preflight through the configured execution backend. **Run demo** creates a
+conversation and opens a monitored simulation with downloadable JSON, netCDF,
+CSV and a three-panel figure. It does not consume LLM tokens.
+
+Equivalent CLI commands, from a deployment checkout:
+
+```bash
+simjecture install iter-pack
+simjecture doctor --profile iter-pack
+```
+
+The runtime includes CHERAB 1.5.0, compatible Raysect 0.8.1.post1, pinned
+CHERAB-ITER and CHERAB-IMAS source revisions, IMAS-Python and IMAS-Validator.
+Build records retain package versions, requirements and installation duration.
+The older NumPy/Raysect dependencies stay outside Simjecture's Python environment.
+On hosts without a supported system Python, the installer provisions a local
+Python 3.12 environment with the existing Micromamba bootstrap. Git and a C
+compiler are required for the CHERAB extensions. No GPU or scheduler is needed.
+
+The offline example checks:
+
+- Raysect integration through a uniform sphere against its analytic chord length.
+- CHERAB free-free emission scaling with electron and ion density.
+- IMAS schema validation and a netCDF profile round-trip with an explicit DD version.
+- Construction of three five-channel bolometer cameras from upstream mock IDS data.
+- The CHERAB-ITER compiled geometry helper against a finite-difference derivative.
+
+These are bounded readiness checks. They do not validate actual ITER instruments,
+line-emission atomic data, a plasma discharge, or an arbitrary reconstruction.
+
+## Use with an agent
+
+Choose your backend/model in the workspace and ask, for example:
+
+> Use the ITER pack to compare two synthetic plasma emission profiles from
+> different viewing angles. Start from the working demo, quantify which signals
+> distinguish them, and retain the assumptions and output plots.
+
+The `iter-pack` skill is available through `read_skill` to API agents and through
+files to native CLI agents. It links focused references rather than duplicating
+upstream manuals. It explains data dictionary compatibility, atomic-data needs,
+ray sampling, geometry provenance and the distinction between measured signals
+and reconstructed profiles. In minimal studies, the registered capability works
+through the ordinary `lab.run` interface. Source packages remain readable.
+
+## Separate solver integrations
+
+| Tool | Workspace entry | Current qualification |
+| --- | --- | --- |
+| CHERAB / Raysect / IMAS | Install, Check readiness, Run demo | Automatic install and offline numerical demos verified |
+| JOREK | Install with agent; check registered builds | A local GNU/MUMPS model600 build passed a three-step sandboxed example; fresh-machine setup remains guided |
+| SOLPS-ITER | Install with agent | Public components inspected; EIRENE compiled, full coupled runtime not yet qualified |
+| DINA-PS | Install with agent | Native core/controllers compiled and supplied magnetic-geometry component test passed; full IMAS scenario workflow not yet qualified |
+
+The diagnostics pack does not contain these three solver binaries. A registered
+JOREK build advertises its specific model and toroidal harmonics, not all JOREK
+physics. Guided setup preserves source, build configuration, input identity and
+logs before registering the actual working application.
+
+See the [acceptance report](../testing/iter-pack-acceptance.md) for measured results,
+remaining dependencies and the real DeepSeek Flash test. Reproduce the fresh
+browser installation test with `python scripts/verify_iter_pack.py` after
+installing the workspace dependencies and Playwright Chromium. It downloads into
+a new `artifacts/iter-pack/oneclick-*` deployment and makes no model calls.