Skip to content

luxiaolei/autospread

v0.3.0MIT

A lightweight bilingual growth operating system for agents: diagnose constraints, choose playbooks, act within autonomy policy, and verify outcomes.

GitHub

AutoSpread

AutoSpread is a lightweight, bilingual growth operating system for AI agents. Give a capable agent a product repository, product URL, business goal, and autonomy boundary; the Lead diagnoses the largest current growth constraint, loads only the relevant playbooks, uses the connected tools, verifies real-world outcomes, and learns from the result.

It is intentionally not another multi-agent runtime. Roles are methods, not mandatory bots. One Codex/ChatGPT/Hermes/Claude-style session can perform several roles, or delegate independent work when the host supports it.

v0.3 operating model

product + goal + autonomy policy
            |
        Growth Lead
            |
    diagnose constraint
            |
  choose the smallest playbook
            |
product / funnel / monetization / SEO-GEO / content / channels / lifecycle / paid
            |
 API/MCP -> CLI -> SDK/code -> ego-browser -> human
            |
       execute + verify
            |
 activation / revenue / retention / learning
            `----> next cycle

Included

  • a Growth Lead router and constraint-diagnosis method;
  • an explicit autonomy-policy method and editable policy template;
  • project audit, customer research, CRO, monetization, lifecycle, paid acquisition and analytics playbooks;
  • SEO split into foundation, GEO/answer-engine optimization, and AI-visibility measurement;
  • a conditional SEO specialist layer for site integrity/drift, search experience, local, international, ecommerce, and agent readiness;
  • content strategy, content/video production, channel selection and channel operations;
  • a growth-experiment contract that ties work to evidence and business outcomes;
  • 25 bilingual Agent Skills;
  • five product workspace templates: product context, autonomy policy, capability inventory, experiment ledger, and work log;
  • a machine-readable catalog of 67 reviewed capability pointers (skills, plugins, MCPs, CLIs, platforms and runtimes);
  • a standard-library-only helper for validation, catalog filtering, context initialization and project-skill installation.

Use it

In a capable session, read skills/autospread/SKILL.md, then provide the product repository/path, public URL, goal, and the amount of autonomy you want to grant.

Read /path/to/AutoSpread/skills/autospread/SKILL.md.
Take over growth for /path/to/my-product and https://example.com.
Use the product's autonomy-policy.yaml. Start with a read-first audit and constraint diagnosis,
then propose or execute the smallest useful growth experiment within that policy.

Initialize the product records:

python3 scripts/autospread.py init-context /path/to/my-product/.autospread

Explore resources by role or keyword:

python3 scripts/autospread.py catalog --role geo-optimization
python3 scripts/autospread.py catalog --query stripe

Run offline checks:

python3 scripts/autospread.py validate
python3 -m pytest -q

See README.zh-CN.md, docs/en/operating-model.md, docs/zh-CN/operating-model.md, and the attribution notes for the full model.