By sancovp
SDNA (Sanctuary DNA) - Gnostic agent workflow DSL with skills for chain development
⭐ 0 stars • 🕑 Updated 2026-06-17
📦 Auto-published from the monorepo • CHANGELOG • sancovp/sdna
Gnostic agent workflow composition for Claude Agent SDK.
Ariadne (threading) + Poimandres (generation) = SDNA spiral
pip install sdna-agent-sdk
from sdna import ariadne, human, inject_file, sdnac, sdna_flow, HermesConfig
# 1. Build Ariadne thread (context prep)
thread = ariadne('my-thread',
inject_file('spec.md', 'spec'),
human('Approve spec?', 'approval'), # Pauses for human input
)
# 2. Create HermesConfig (the message)
config = HermesConfig(
name="generator",
system_prompt="You are a code generator...",
)
# 3. Combine into SDNAC
unit = sdnac('generate-code', thread, config)
# 4. Execute
result = await unit.execute({'initial': 'context'})
| Component | Role | What It Does |
|---|---|---|
| Ariadne | Threader | Context manipulation: inject, weave, dovetail, human input |
| Poimandres | Divine Mind | Generation moment - takes config, runs agent, returns output |
| HermesConfig | The Message | Runner configuration Ariadne sends to Poimandres |
Is this continuous improvement / optimization loop?
├── YES → SDNA^F (SDNAFlowchain)
│ Optimizer + target pairs. Meta-optimization.
│
└── NO → Are you composing multiple agent units in sequence?
├── YES → SDNAF (SDNAFlow)
│ Flow of SDNACs. Sequential execution.
│
└── NO → SDNAC
Single unit: AriadneChain → HermesConfig → Poimandres executes
from sdna import (
ariadne, # Create chain
human, # Human input stop step
inject_file, # Inject file contents
inject_func, # Inject function result
inject_literal, # Inject literal value
inject_env, # Inject env variable
weave, # Context surgery between sessions
)
MIT
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
npx claudepluginhub sancovp/sdna --plugin sdnaAutopoiesis Plugin - Self-maintaining promise loops for honest AI work
CodeNose - Configurable code smell detection for LLMs. Sniffs out duplicate logic, architecture violations, missing logging, and other blind spots.
Instantiate a TWI Jobworld — AI-powered company with agents that run themselves. Enable this plugin and a CEO agent spawns, ready to bootstrap your company.
The doc-mirror claude system: a 1:1 documentation mirror of any codebase, maintained by an invariant doc-mirror STATE MACHINE (per-module doc(m), 6 context/ index files, fork-on-change, no random documents). Bundles doc-mirror-boot (entry + core-loop prime + router), the doc-mirror-{init,seework,change,prompts} state-skills, doc-mirror-install (the host-setup WIZARD), ship-a-plugin (the build+distribute dev-flow), doc-mirror (THE LAW), and make-ai-operating-system (the architect for new systems of this class); the brainhook work-loop + session-start + transition-guard + readonly-guard hooks (plugin-native via CLAUDE_PLUGIN_ROOT); the journal/projects/vision/plan/tracker/commit/cursor + docmirror-search CLI tools; and the doc-mirror law + behavior rules.
PromptWorld — the meta-compiler *World: a WrightMaster Engineer-CEO that has every skill plus a native subagent per component type (skill / mcp / harness / operating_system / prompt / team / workflow), and 7 PromptGym specialist AIOS agents, that compile / compose / publish prompt systems through the nomicon. Bundles the install-promptworld deploy WIZARD (the agent builds the docker image, boots the container running the Engineer-CEO + PromptGym + the assistant-ui chat / Monaco file explorer / embedded terminal, copies the Claude creds in, and hands the user one URL), compile-a-world (the meta-compiler skill — the nomicon folds on itself), and ingest-into-nomicon-app.
Reliable automation, in-depth debugging, and performance analysis in Chrome using Chrome DevTools and Puppeteer
Comprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Design fluency for frontend development. 1 skill with 23 commands (/impeccable polish, /impeccable audit, /impeccable critique, etc.) and curated anti-pattern detection.
Behavioral guidelines to reduce common LLM coding mistakes, derived from Andrej Karpathy's observations on LLM coding pitfalls