linxichen/dsh-rigorquant

English | 简体中文

Unattended-within-a-session, long-running empirical/computational mathematics research for DeepSeek Harness — economics, finance, portfolio construction/optimization, simulation, computational econ/finance.

RigorQuant is an agent preset + bundled skills that turns one DSH session into a context-isolated multi-agent research lab:

• Parallel explorers propose candidate methods (explorer-<n>, blank context).
• An OffGridThinker (offgrid-<n>) works off the grid when a route must be isolated: raw model intelligence plus compute tools (sympy, numpy, mpmath, Lean checkers) — no web, no literature, no other agents' results.
• A ground-truth track re-derives the analytic closed forms, invariants, and bounds for simplified cases — twice, by different means (two independent fresh doublechecker-<n> teammates).
• An adversary eliminates routes by counterexample only.
• A four-part check battery (closed-form equality, exact invariants, analytic bounds, statistical hardening) runs BEFORE numerical implementation.
• A meta-validator (rq_check.py) refuses a PASS whose evidence is missing: empty stage outputs, an empty derivations/, a registry with no audit-referenced passed route, or deliverables that do not compile. Its evidence checks read the audit record, not study.json — a study may not vouch for itself.
• Fixed-seed + LLN conventions for stochastic work.
• A jacobian MCP escalation lane (opt-in; Lean as a manual external lane) settles proof-critical claims before implementation.
• PASS → auto-implement and proceed; BLOCKED → 3 rounds of the same gap → strongest derivation + exact gap; BUDGET → 5 rounds → checkpoint + report.

The operating pattern adapts Shanmu Jin's Crouzeix-conjecture run (prompt, Lean audit) and Terence Tao's blueprint/equational-theories projects to numerical work. Full design record: docs/architecture.md.

"Unattended", precisely: the framework runs unattended within one live session. Crossing a session boundary disarms the goal; one human turn ("continue") re-arms it. It does not continue autonomously across restarts.

The research team — and how it works

Eight roles around one hub, each a teammate the orchestrator creates with its own persona and tool budget. The orchestrator is the only role that sees every report — a teammate cannot message another teammate, list the roster, or read the whole board — so the separation is enforced rather than agreed, and the producer never checks its own work: an idea dies only on a concrete counterexample, never on style or vibes.

Orchestrator · root persona — fans out the work, synthesizes, and writes the state. Bound by four rules: producer ≠ checker, counterexample-only elimination, seeds always recorded, no handwaved load-bearing claims.

Explorer · explorer-<n> — blank-context and divergent. Proposes lemmas, equations, constructions, and candidate methods with exact statements. Status reports are rejected.

OffGridThinker · offgrid-<n> — the off-grid lane. Raw model intelligence plus the pinned compute lane (sympy, numpy, mpmath, cvxpy, hypothesis, jax; Lean checkers when provisioned) — and nothing else: no web, no skills, no delegation, no other agents' results. Its own agent, not an Explorer variant: isolation is the identity.

DoubleChecker · doublechecker-<n> — blind (no web, no skills, no delegation, no drafts). Re-derives the load-bearing claims from first principles, twice by different means.

Adversary · adversary-<n> — runs the check group and hunts counterexamples. Ends in a verdict: PASS or NEEDS-EDITS.

Literature · lit-line-<n> · lit-adversary-<n> — a walled citation-graph sweep, then an independent adversary re-retrieves each claim and certifies it's real and current.

Validator · rq_check.py + schemas — refuses a PASS with missing evidence. Reads the audit record, never the study's own claims — a study cannot vouch for itself.

Document adversary · doc-adversary-<n> — an independent agent that audits each finished deliverable for self-completeness (the thing 90% of AI-generated writing drops): every jargon term, symbol, and abbreviation the document uses must be defined in the artifact itself or the audience spec's symbol registry. Returns VERDICT: PASS / VERDICT: NEEDS-EDITS; a NEEDS-EDITS is a blocking gap the validator refuses a PASS without.

The team, live — the native team view

A study runs on the harness's own Agent Teams surface, so there is no custom panel to learn: while a RigorQuant session runs, the session header's team action opens the roster (name, role, status) and the round's task board with its blocking edges — the two things to watch. The roster's model column shows the member's model selection, not the model rq-model-router routes its requests to (Decision 16), so read a role's route off the Plugins → dsh-rigorquant card, never off the roster. Each teammate is a row there: open one and its own conversation opens, so you can read a derivation or an audit while it runs (steering it is the ordinary conversation, and it breaks that teammate's blank context — documented, not prevented).

The topology is hub-and-spoke, and the guards enforce it rather than asking: a teammate's message reaches the orchestrator or nowhere, a teammate cannot list the roster or the whole board, and it may read or update only a task no other teammate owns (the one its brief names, which it claims).

Attribution. The hero banner (docs/figs/agent-team-hero.svg) is reworked from the hero graphic of dsh-agent-teams by NanmiCoder (程序员阿江 / Relakkes) — Copyright (c) 2026, MIT License. The role portraits are this repo's own docs/figs/ assets.

The loop, in five moves. Each round is fan-out → ground truth → adversary → synthesize.

1. Promise — record the original question verbatim, split it into sub-problems with crisp criteria, pick hand-checkable simplified cases, and pin seeds, tolerances and the schema/validator digests.
2. Fan out — blank-context explorers and literature lines run in parallel; most are never told the favored approach.
3. Ground-truth it — the blind DoubleChecker re-derives the load-bearing claims without seeing anyone's draft; two independent derivations for anything the study rests on.
4. Attack it — the adversary runs the four-gate battery, then hunts counterexamples; divergent tracks are lined up as an adjudication docket.
5. Certify & ship — the validator checks nothing is missing; the paper and slides are assembled from validated records, never written fresh.

The check battery, run before any numerical implementation: A closed-form equality · B exact invariants · C analytic bounds · D statistical hardening (fixed seed + LLN shrinking ≈ C/√N).

With receipts: in one hard run, 21 errors were caught by a specific mechanism and none by luck (11 of them the orchestrator's own); only 35% of 81 literature claims survived independent verification; and the honesty gate is itself tested — a forged study must fail.

Install

Requires DSH >=0.1.7-rc.2 <0.1.8 (Decision 25, docs/adr/0002-declared-preset-on-dsh-0.1.7.md). The preset is a declared @deepseek-ai/dsh-agent-preset row, the routes are the router row's own profile config, and the card edits them through the Plugins page's config forms; none of that exists on an older harness, and the installer refuses one. 0.5.0 is the last release for the 0.1.6 alpha harness: nothing is backported, so stay on 0.5.0 if you cannot move the harness yet.

That range is a prerelease and the team layer is experimental: this release runs team-only on the Agent Teams bundle the harness ships as a Beta card, Agent Teams under Plugins → Official. Turn it on there yourself, or let a full install do it for you (below).

Upgrading from 0.5.0 is one cutover, done before the first 0.1.7 boot. First finish or archive the RigorQuant studies in progress: resuming a session started on the 0.1.6 harness is not promised. Then:

1. Stop dsh (every running process, the web app included).

2. Install the harness pinned to the release this range was tested on:

   npm i -g @deepseek-ai/dsh@0.1.7-rc.2   # or @deepseek-ai/dsh@next
   

   Never an unqualified npm i -g: npm latest is still 0.1.5-rc.3.

3. Run ./install.sh (or npx dsh-rigorquant) before starting dsh. It carries your saved routes and a saved RigorQuant default over from the old settings.yaml. If dsh already booted once, re-running it recovers them from settings.yaml.imported.

4. Start the web app and pick RigorQuant in the new-session picker. If the picker is missing, switch Coding Tools back on in General Settings: it is the picker's only gate, on by default, and off only if you or Desktop onboarding switched it off.

Fan-out is bounded by the host: eight live children per root (maxActiveSubagents, Plugins → Subagent). A literature-heavy study that wants four lines running beside its explorers can raise it there.

A full install turns on that Beta bundle, the Plugins page's Agent Teams card under Official, when the target profile does not have it, pinned to the core's own version (a profile that lists it at some other version is re-pinned; a bundle you enabled without a recorded version is left alone). It always removes the separate Agent Teams Web UI bundle a 0.5.0 profile has, saying so in one line: DSH 0.1.7 folded that panel into the one bundle and publishes no web bundle for this core, and if the removal fails the install stops. It raises the team service's lifetime member cap to 64 through a marked dsh-rigorquant block it appends to the profile's user patch ($DSH_HOME/profiles/<profile>/cordis.patch.yml), printing every line it writes. Re-running changes nothing once installed; --uninstall removes the lines it wrote in the marked block (a setting the harness later saved between the markers stays) and disables the bundle only if that block records the installer having turned it on, leaving a bundle you enabled yourself untouched. With no dsh on the path this step is skipped with a warning, same as the rest of the install, and the Beta card on the Plugins page is then yours to toggle (docs/adr/0001-rigorquant-on-agent-teams.md).

One line, everything: the compute lane and the plugin (the declared preset, the role model router and its card on the Plugins page):

npx dsh-rigorquant
# npx dsh-rigorquant --profile tui     # a profile other than web

From a clone: the same install, from your own working tree. A checkout installs itself into the profile, so re-run this after editing dsh/ to refresh the plugin:

git clone https://github.com/linxichen/dsh-rigorquant
cd dsh-rigorquant
./install.sh
# ./install.sh --skill-only     # only the skills, for any preset, no plugin
# ./install.sh --uninstall      # removes everything, plugin included

Plugin only: the same everything, via the ecosystem's bundle path. The package declares a dsh.bundle manifest that declares the rigorquant preset itself, and one of its rows is a boot-sync half (rq-lane-sync): on the profile's next start it lands the compute lane into $DSH_HOME/share/rigorquant/, so dsh plugin add alone yields a working distribution (docs/architecture.md Decisions 22 and 25). This path neither enables Agent Teams nor carries saved routes over; the installer does both:

dsh --version                 # must be >= 0.1.7-rc.2 and < 0.1.8
dsh plugin --profile web add dsh-rigorquant

The sync is idempotent (a byte-identical tree is left untouched; derived state like .venv is never copied or pruned). It also removes the directory preset that releases before 0.6.0 copied under $DSH_HOME, but only when that copy's .rq-sync.json says this package put it there. There is no uninstall hook in DSH's plugin CLI, so removal of the lane stays explicit (./install.sh --uninstall).

Start a new DSH session and pick the RigorQuant preset. Then:

rigorquant: derive and validate a method for [problem], simplified cases first, before any numerical implementation.

Deployment notes

Four harness facts a research deployment should decide on. None of them is RigorQuant's own machinery, and installing RigorQuant changes none of them:

• The DeepSeek session log is on by default. The base bundle mounts session-log-deepseek with enabled: true: each session's canonical event log goes up as request metadata to the official DeepSeek API — no model-input tokens, but the whole run leaves the machine. To turn it off, overlay the row in the profile's user patch ($DSH_HOME/profiles/<profile>/cordis.patch.yml) and restart the profile — changes land at the next start:

  - id: session-log-deepseek
    config:
      enabled: false
  

• The goal-round driver is the harness's, and it is already mounted. The goal service and goal-round-driver are host-plane rows of the shipped base bundle (@deepseek-ai/dsh-base); the preset re-mounts only the human /goal command and the model-facing goal tool, which the web bundle disables at the host plane. Nothing in this repo ships, mounts or arms the round driver: "unattended" is a native contract, and crossing a session boundary still disarms the goal until one human turn re-arms it (Decision 10).

• The workspace-changes card is the human-visible witness of an edit. The web bundle's workspace-changes row records each top-level turn's changed files from git working-tree snapshots and renders the changed-files card under that turn, so an edit that lands after a verdict is visible exactly where and when it happened — what the frozen-write rule of Decision 19 needs a human to be able to see. Certification itself reads the study record, never the session (docs/architecture.md Decision 19).

• Scheduled tasks are neither used nor guarded. DSH 0.1.7 ships scheduled tasks (and their time context) disabled. RigorQuant does not use them, and its per-call guards do not cover them: a scheduled task you enable yourself runs outside the team's hub-and-spoke rules.

Compute lane (one-time)

Each study carries its own pinned uv lane. install.sh or the plugin's boot-sync row places a template at $DSH_HOME/share/rigorquant/env (both land identical bytes; see env/README.md). At intake the orchestrator copies its pyproject.toml and uv.lock into the study's env/, commits them with the record, and builds the venv once under the study's gitignored interim/ (venv and uv cache both live there, which is also the only place the workspace-write sandbox lets them go). A clone of the study therefore rebuilds the exact environment even after a later release replaces the template, and a study that needs another package adds it to its own lane. The validator refuses a PASS without the study's env/uv.lock. Budget about 1 GB per live study for the venv and cache, deleted at close-out: the sandbox keeps studies from sharing a uv cache. The jacobian escalation lane is pinned (jacobian@0.12.0) and is no preset row: it mounts at runtime. When a claim needs it, the orchestrator calls rq_escalate without asking, into itself or into a teammate it names, and the jacobian tools appear from the next request until the session ends. The framework still asks for approval before any one-time provisioning (npx -y jacobian@0.12.0 upgrade, or the Lean toolchain via the skill's scripts/provision-lean.sh). See mcp/jacobian.md.

Role-routed models (rq-model-router)

The bundled plugin gives each RigorQuant role a model + reasoning-effort policy, with one fallback per role. Role identity comes from the teammate's Team membership name (<role>-<n>; the Lead is the orchestrator) — the router carries the shipped tier matrix itself (DoubleChecker and adversary default to deepseek-v4-pro @ high) and overlays a saved route on top. Configure routes in Plugins → dsh-rigorquant (the bundle's own page, under its description): only Save writes, a saved route is the router row's own config in the profile's cordis.patch.yml and applies from the next request, and leaving the page drops staged edits. Shipped defaults:

Role	Primary	Fallback
DoubleChecker	deepseek-v4-pro @ high	deepseek-flash @ low
Adversary	deepseek-v4-pro @ high	deepseek-flash @ low
Root, explorers, OffGridThinker, literature/document roles	inherit (root follows the chatbox picker)	—

On a terminal primary failure (no adapter / HTTP 4xx, including the official quota response 1308 / “Usage limit reached”) the role degrades to its fallback for one forced retry, and recovers on the next success or after 10 minutes. An agent outside a RigorQuant team (another preset, or with no Team membership at all) is never touched.

Signed in with a DeepSeek account only? DSH 0.1.7 splits DeepSeek into deepseek-official (API key) and deepseek-account (account sign-in). When the official route is not routable (no API key) and the account route is, the shipped defaults above move to the same model ids on deepseek-account, and those requests are billed to your account quota, not to an API key. A route you saved yourself is never moved. Design record: docs/architecture.md Decisions 16 and 25.

Repository layout

package.json                dsh.bundle manifest (dsh plugin add support)
cordis.patch.yml            bundle patch: skill layer + rq-model-router +
                            rq-team + rq-lane-sync rows
dsh/                        host halves (role router, team composition and
                            per-call guard, boot-sync) + one persona file per
                            role, and the web client bundle (routing card)
agent-presets/
  rigorquant.patch.yml      the declared `rigorquant` preset (persona + child rows)
  rigorquant/skills/        bundled skills
    rigorquant/             SKILL.md + references/ + scripts/ + schemas/
  .../scripts/rq_check.py   the meta-validator (single canonical copy)
  .../schemas/              study.json + registry.json JSON Schemas, which the
                            validator loads — so schema and checker cannot drift
env/                        pinned uv compute lane (sympy/cvxpy/hypothesis/…)
mcp/jacobian.md             escalation lane wiring
docs/architecture.md        grilled decision record + sources
docs/figs/agent-team-hero.svg       hero banner, reworked from the
                             dsh-agent-teams hero graphic (see the credit above)
tests/                      the validator's test suite (see Testing below)
studies/                    one folder per study (Mode B; a checkout's own
                            live studies — not shipped in the npm bundle)

Testing

The validator has a test suite, and its centrepiece is a forged study — empty derivations, empty stage outputs, a one-line adversary report, and a paper whose body reads "This paper says nothing." It must FAIL. A framework whose honesty gate is not itself tested is a framework that certifies whatever it is handed.

uv sync --frozen --project env
uv run --frozen --project env python -m pytest tests/ -q

Pre-commit coverage gate (validator ≥95%)

The checked-in hook at .githooks/pre-commit runs the same full suite under coverage and refuses a commit when the shipped validator (rq_check.py) drops below 95% line coverage. ./install.sh enables it in a git checkout; in an existing checkout, enable it explicitly:

git config core.hooksPath .githooks

The validator is a subprocess, so coverage is deliberately explicit rather than plugin magic: RQ_COVERAGE=1 makes tests/conftest.py::run_check invoke coverage run --parallel; the hook combines those child data files and applies coverage report --fail-under=95. CI runs the identical gate. Bypass one commit only with Git's standard git commit --no-verify.

tests/test_repo_consistency.py covers the other half: one validator, one schema, documented commands that resolve, layout blocks that match the filesystem, and one word per concept across the tracked docs. That is the defect class this repository actually produces.

What a green validator means: nothing declared is missing, and the deliverables build. It does not mean the mathematics is right — that stays with the check battery, the independent ground-truth track, and the adversary.

Studies

A study is the assignment: one self-contained piece of work with an identical folder structure everywhere. Its durable deliverables (study.json, STUDY.md, registry.json, journal.md, derivations/, audits/, artifacts/) sit at the study root and are meant to be committed; all scratch lives in a gitignored interim/. Two modes, implied by location:

• One study per repo — study.json at the repo root.
• Multiple studies per repo — studies/<slug>/study.json; the roster is studies/*/study.json.

Intake detects an existing study and continues it silently; a new study asks one question (mode + slug) and never asks again. See docs/architecture.md §12.

Publishing

This repo is a community DSH plugin distribution (bundle + preset + skill form): it declares a dsh.bundle manifest in package.json, is tagged dsh-plugin, and is discoverable by the ecosystem's topic-based indexes — see dsh-find-plugins and the awesome-deepseek-harness list for the conventions.

MIT License.