wp-agent-harness
llms.txt0 starsChanged 13 days agoMITView on GitHub
Local, disposable WordPress harness for coding agents. Clone it, run one command, and agents talk to a real site over the official WordPress MCP Adapter. The WordPress harness is development only (loopback ports, no production WordPress deploy); site/ is the public marketing page at https://wpagent.marcopontili.com. This repository does not ship an MCP server, abilities framework, or WordPress plugin of its own; it wires Docker Compose, Application Password auth, tests, docs, examples, and curated WordPress agent skills. This file is the…
- Reads credentials
In this repository16 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .cursor/AGENTS.mdAGENTS.md
- docs/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- site/llms.txtllms.txt
- wordpress-routerSkill
- wp-abilities-apiSkill
- wp-abilities-auditSkill
- wp-abilities-verifySkill
- wp-block-developmentSkill
- wp-block-themesSkill
- wp-patternsSkill
- wp-plugin-developmentSkill
- wp-project-triageSkill
- wp-rest-apiSkill
- wp-wpcli-and-opsSkill
The file
We split this llms.txt into the rules it gives an agent
3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 13
Read the rules below before driving the site.
- Very specific: line 24
Prefer the rest-get and rest-write abilities before concluding a REST route is unreachable.
- Distinctive: line 32
Default to drafts: Do not publish, delete, or bulk-change unless asked.
≈ “Common” is an estimate: at least 1 in 5 of the 3,592 llms.txt files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
# Agent Harness for WordPress > Local, disposable WordPress harness for coding agents. Clone it, run one > command, and agents talk to a real site over the official WordPress MCP > Adapter. The WordPress harness is development only (loopback ports, no > production WordPress deploy); [`site/`](site/) is the public marketing page at > https://wpagent.marcopontili.com. > This repository does not ship an MCP server, abilities framework, or > WordPress plugin of its own; it wires Docker Compose, Application Password > auth, tests, docs, examples, and curated WordPress agent skills. This file is the entry point for AI agents and engineers adopting the harness. Read the rules below before driving the site. Discover abilities at runtime; do not invent names or parameters from memory. ## How to use this harness 1. `./bin/setup`, then connect a client: `./bin/connect` (Claude Code) or `./bin/connect --print` (Codex / Cursor / any HTTP MCP client). 2. Drive the site through the `wordpress` MCP server, or diagnose with `./bin/wp ability list` / `./bin/wp ability run` (bypasses MCP). 3. MCP meta-tools only: `mcp-adapter-discover-abilities`, `mcp-adapter-get-ability-info`, `mcp-adapter-execute-ability`. 4. Prefer the `rest-get` and `rest-write` abilities before concluding a REST route is unreachable. 5. Obey the hard rules in this file and in [docs/AGENTS.md](docs/AGENTS.md). ## Hard rules (never break) - **Never invent ability names or parameters.** Discover, then `get-ability-info`, then execute. - **Default to drafts.** Do not publish, delete, or bulk-change unless asked. - **Inspect before change.** State what you found. - **Destructive ops need an explicit ask:** trash, deactivate plugins, change site settings, `./bin/reset`. - **Writes pass the approval gate.** `bin/ability-gate` makes Claude Code and Cursor ask before every write ability; do not route a write around it through `./bin/wp` or `rest-write`. `./bin/audit` lists recorded writes. - **Never expose** this harness's MCP endpoint or WordPress beyond localhost. A separate live site is outside the harness: [docs/remote-site.md](docs/remote-site.md). - **Never put credentials** in a tracked file (`.env`, `.secrets/` stay ignored). - **Do not edit infra** (`docker-compose.yml`, `docker/`, `bin/`, `.env`) for a content or site-building task without saying so first. - **Do not switch** the local site to `@wordpress/env`, Playground, or Blueprints; this repo's Docker Compose setup is the environment. - **No agent attribution** on commits or PRs. ## Files - Agent index (canonical): [docs/AGENTS.md](docs/AGENTS.md) - Root agent mirror: [AGENTS.md](AGENTS.md) - Claude Code entry: [CLAUDE.md](CLAUDE.md) - Abilities: [docs/abilities.md](docs/abilities.md) - AI skills map: [docs/ai-skills.md](docs/ai-skills.md) - Task contract: [docs/agents/agent-contract.md](docs/agents/agent-contract.md) - Risk tiers: [docs/agents/runtime-policy.md](docs/agents/runtime-policy.md) - Architecture: [docs/architecture.md](docs/architecture.md) - Development loop: [docs/development.md](docs/development.md) - Claude Code MCP: [docs/claude-code.md](docs/claude-code.md) - Codex MCP: [docs/codex.md](docs/codex.md) · [.codex/config.toml](.codex/config.toml) - Live site (outside the harness): [docs/remote-site.md](docs/remote-site.md) - Security: [docs/security.md](docs/security.md) · [SECURITY.md](SECURITY.md) - Troubleshooting: [docs/troubleshooting.md](docs/troubleshooting.md) - Contributing: [CONTRIBUTING.md](CONTRIBUTING.md) - Examples: [examples/README.md](examples/README.md) - Skills: [.claude/skills/](.claude/skills/), linked per skill from [.agents/skills/](.agents/skills/) ## Agent skills Curated WordPress coding skills (blocks, themes, plugins, REST, Abilities API, WP-CLI, router/triage) under `.claude/skills/`. They do not replace MCP/`./bin/wp` for site work on this harness.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

