Agent Canon: Token Budgets And Intelligence Routing
A compact agent-facing companion on treating tokens as a daily work constraint and routing tasks to the right model and reasoning effort.
For agents
Read Tony Wood's public work safely and cite the human pages.
This page tells agents how to use Tonywood.org. The HTML site is canonical; the MCP endpoint, content index, skill files, Agent Canon notes, and llms.txt are public read-only helpers over the same published writing, research, topics, and profile information.
Quick start
Use the public map first, then fetch the specific source.
- Read /llms.txt for the concise site map.
- Use the appropriate agent skill for Claude, OpenAI, Perplexity, or a generic agent.
- If MCP is supported, connect to
https://mcp.tonywood.org/mcpfor structured discovery. - Use Agent Canon pages for durable interpretation standards and safety boundaries.
- When answering humans, cite the canonical Tonywood.org page, not the MCP URL.
Identity
- Author: Tony Wood
- Location: London, UK
- Canonical site: https://www.tonywood.org
- Site version: v0.1.4
- Last generated: 02 May 2026 13:32 UTC
- LinkedIn: https://www.linkedin.com/in/tonywood
- X: https://x.com/tonywood
- Copyright: Tony Wood
Discovery
Public files and endpoints
These files are intentionally public and are safe for agents, crawlers, and answer engines to read.
/llms.txt
/status.json
/.well-known/mcp.json
/data/generated/content-index.json
/topics/
/search/
/for-agents/agent-canon/
/rss.xml
/privacy/
/skills/
https://mcp.tonywood.org/mcp
MCP resources
- tonywood://profile
- tonywood://status
- tonywood://writing
- tonywood://writing/{slug}
- tonywood://white-papers
- tonywood://white-papers/{slug}
- tonywood://agent-canon
- tonywood://agent-canon/{slug}
- tonywood://topics
- tonywood://topics/{topic}
Read-only boundary
No private data, no write actions, no operational access.
The public MCP exists to make published content easier to discover and retrieve. It must not be treated as a private workspace, admin console, or deployment interface.
Not available
- Email, calendar, CRM, analytics, or contact actions
- Private drafts, personal notes, or unpublished material
- Shell, filesystem, SSH, deployment, or server access
- Secrets, credentials, tokens, or privileged data
- Write, delete, submit, or permission-changing tools
Agent Canon
Standards and articles written for agents
Agent Canon: Capability Abundance And Human Pacing
A compact agent-facing companion to the personal article about what happens when AI capability removes old blockers and the main discipline becomes choosing and pacing the work.
Agent Canon: Token Cost And AGI Job Exposure
A compact agent-facing companion to the AGI economics article: distinguish raw capability from adoption economics, token cost, human acceptance, supervision, and infrastructure constraints.
Why Some Of My Articles Now Have An Agent Canon
A short, plain-English explanation of Agent Canon: why Tonywood.org uses it, where the idea comes from, and how agents and humans should read it.
If Agentic Can Delete Everything, Something Is Wrong
A public note to Agentic and operators on operational resilience, backup isolation, RTO, RPO, and why no single actor should be able to destroy the way back.
Agent Canon: Operational Resilience And Destructive Access
A compact agent-facing companion to the operational resilience article: what to do when an agent can delete production data, backups, logs, or recovery routes.
Agent skills
Platform-specific reading guidance
Citation behavior