System And Research Workflow Overview
This repository is designed for an experimental personal multi-agent research workstation, with an emphasis on combinatorics and graph theory workflows. It is not guaranteed to work as desired in every environment. Codex, Claude, and DeepSeek each keep their own local configuration directory, but the reusable research instructions live here as canonical skill bodies. The installer links those skill bodies into whichever agents are present by default, can write thin reference adapters when symlinks are not suitable, and leaves absent agents alone.
The system has three layers:
Layer |
Role |
|---|---|
Agent frontends and targets |
Codex, Claude, DeepSeek, Copilot, OpenCode, Antigravity, Grok, and Kimi receive user requests and load installed skill instructions; OpenClaw is a restricted fake-root target for normal installer flows, with reviewed v2 real-system skill-file writes only through |
Shared skill repository |
|
Runtime and software tools |
Python, TeX, optional SageMath, local library tools, document parsers, public databases, and external retrieval helpers do the actual work when a skill needs them. |
The installer links these layers without embedding private state. It does not
store credentials, session logs, local library databases, downloaded papers, or
machine-specific paths. Instead, precheck detects logical capabilities such
as python-runtime, tex-runtime, sage-runtime, library access, and
optional Python packages on the current system.
A typical research workflow looks like this:
A request enters one installed agent, for example Codex, Claude, or DeepSeek.
The agent loads a shared skill such as
research-briefing,deep-research-workflow,zotero,docling, ortikz-draw.The skill routes to the right software capability: local libraries first, document parsing when files are involved, public databases for structured records, TeX for figures, and SageMath or Python for math checks.
Research phases keep explicit guard outputs: scope boundaries, goal/backward success checks, delegated-agent result packets, source handoffs, and a bounded iteration ledger.
The final answer passes through review or verification skills when the task needs stronger evidence control. Verification reports concrete issues and remaining gaps instead of an aggregate quality score.
Reusable failures, corrections, or missing capabilities are logged with
self-improving-agent. When they affect shared skills or settings, the learning includes a canonical repo integration plan before any target-home or repo mutation.
Examples:
Current literature brief:
research-briefingscopes the question,deep-research-workflowpreserves source IDs across search and synthesis,paper-lookupordatabase-lookupfills metadata gaps, thenresearch-report-reviewerandresearch-verification-gatecheck the final report.Paper review from a local library:
zoterochecks the paper library first,calibreis used for book-like review inputs,doclingparses local documents when structure matters, andpaper-revieworannotated-reviewperforms the review workflow.Research figure or math-heavy answer:
deep-research-workflowproduces a figure brief,tikz-drawturns it into a structural diagram using TeX, andsagemathorgraph-verifierhandles graph or algebra checks when available.Windows plus WSL-backed tools: Windows agents can receive the same skill bodies as Linux agents. Tools such as SageMath may be detected as WSL-backed capabilities, so the dependency graph records the substrate instead of hardcoding a personal path.
Reusable workflow improvement:
self-improving-agentrecords local.learnings/entries, then proposes repo-first changes acrosscanonical/,manifest/, generated docs, runtime helpers, and tests with explicit target/OS coverage limits.
Related pages: Installation, Skills, Dependencies, Multi-Agent Examples.