Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/dgenio/chainweaver/agents-mdgit clone --depth 1 https://github.com/dgenio/ChainWeaverWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.03925 | $0.03925 |
| Opus 5 | $0.01962 | $0.01962 |
| Sonnet 5 | $0.00785 | $0.00785 |
| Haiku 4.5 | $0.00392 | $0.00392 |
Grade A, and why
ChainWeaver AGENTS.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 299 lines — stays where its author put it; the contents beside it link to each section on GitHub.
ChainWeaver — Agent Instructions
Single source of truth for all coding agents working on this repository. This file carries the stable global contract; path-scoped
AGENTS.mdfiles add durable local rules (see §11), and the detailed module inventory lives in docs/agent-context/module-map.md. For tool-specific wrappers, see the documentation map at the end of this file.
1. Project identity
ChainWeaver is a deterministic orchestration layer for MCP-based agents. It compiles multi-tool flows into executable sequences that run without any LLM involvement between steps.
- Python 3.10+;
from __future__ import annotationsin every module. - Small runtime dependency set:
pydantic,typer,tenacity,packaging, anddeepdiff. - Core philosophy: compiled, not interpreted — the executor is a graph runner, not a reasoning engine.
2. Domain vocabulary
Use these terms consistently in code, docs, comments, and PR descriptions.
| Canonical term | Never use | Meaning |
|---|---|---|
| flow | chain, pipeline | A named, ordered sequence of tool invocations (Flow) |
| tool | function, action | A named callable with Pydantic input/output schemas (Tool) |
3. Repository layout
Stable top-level shape only. The full per-module inventory — every module's responsibility, exports, and issue history — is the module map (a mechanically freshness-checked reference, not policy).
chainweaver/ The package. Public API surface is __init__.py __all__.
├── executor.py FlowExecutor — the deterministic runner (main entry point)
├── _execution/ Private no-I/O collaborators shared by both execution lanes
├── flow/ Flow/FlowStep/DAGFlow model package (stable facade)
├── cli/ typer CLI command package
├── mcp/ MCP adapter + FlowServer (trust boundary; [mcp] extra)
├── integrations/ Optional third-party adapters (each guards its extra)
├── testing/ Public flow test harness
├── contrib/, export/ Curated stdlib tools; schema export adapters
└── *.py One concern per module — see the module map
tests/ Pytest suite (helpers.py = schemas/tools; conftest.py = fixtures)
examples/ Runnable standalone examples
docs/ Hosted MkDocs site + docs/agent-context/ (agent deep-dives)
scripts/ CI/maintenance scripts (not shipped)
benchmarks/ Standalone benchmark scripts
playground/ Streamlit onboarding playground (not lint/type-gated)
pytest_chainweaver.py Top-level pytest plugin (deliberately outside the package)
pyproject.toml Tooling source of truth (ruff, mypy, pytest)
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 299 lines · 3,925 tokens per session scan A a8b9b849804a
ChainWeaver AGENTS.md is an instructions file published in the GitHub repository dgenio/ChainWeaver (6 stars, last pushed 2d ago), licensed Apache-2.0. It adds 3,925 tokens to every session, about $0.0196 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
wet-mcp AGENTS.md
Instructions for n24q02m/wet-mcp, covering agents.md - wet-mcp, cau truc, lenh thuong dung, mise shortcuts and cau hinh quan trong.
maidan CLAUDE.md
Instructions for david-engelmann/maidan, covering agent guide, 30-second orientation, read order, the cluster model in one paragraph and pr workflow (the short version).
luxembourg-mcp CLAUDE.md
Instructions for amirdaraee/luxembourg-mcp, covering claude.md, commands, hard constraint: zero dependencies, architecture and adding or changing a tool touches five places.
swag-mcp CLAUDE.md
Instructions for jmagar/swag-mcp, covering swag mcp - claude memory reference, project overview, core architecture, key components and common development commands.
datalens-dev-mcp AGENTS.md
Instructions for ADIKANT/datalens-dev-mcp, covering agents.md, scope, read order, local material policy and public change hygiene.
swiss-democracy-mcp CLAUDE.md
Instructions for malkreide/swiss-democracy-mcp, covering claude.md, vor der arbeit, tests, zahlen, die eine aufzählung wiederholen and wenn etwas rot ist.