diff --git a/docs/community/openclaw-eve-esi-skill/index.md b/docs/community/openclaw-eve-esi-skill/index.md
new file mode 100644
index 0000000..0b0f3f0
--- /dev/null
+++ b/docs/community/openclaw-eve-esi-skill/index.md
@@ -0,0 +1,72 @@
+---
+search:
+  exclude: true
+
+title: OpenClaw EVE ESI Skill
+type: service
+description: An OpenClaw agent skill that answers questions about your characters in natural language, with read-only ESI access by default and tokens kept on your own machine.
+maintainer:
+  name: burnshall-ui
+  github: burnshall-ui
+---
+
+# OpenClaw EVE ESI Skill
+
+A skill for [OpenClaw](https://openclaw.ai) that lets a self-hosted agent answer
+questions about your EVE characters in plain language — "how much ISK do I
+have", "what is my skill queue", "which extractors run out tonight". It talks to
+the official EVE SSO and ESI APIs only; there is no intermediate service, and
+your tokens never leave the machine you run it on.
+
+<div class="grid cards" markdown>
+
+- [:octicons-mark-github-16: __GitHub__](https://github.com/burnshall-ui/openclaw-eve-skill){ .esi-card-link }
+- [:octicons-browser-16: __ClawHub__](https://clawhub.ai/burnshall-ui/skills/eve-esi){ .esi-card-link }
+
+</div>
+
+## Features
+
+- Natural-language queries against character, corporation and universe data —
+  wallet, assets, skills, clones, location, industry, contracts and market.
+- Planetary Interaction reporting: extractor timers, storage fill levels and
+  attention flags across every colony on a character.
+- Market price lookups, including the live Jita order book for a given type.
+- Threat assessment and route planning from ESI kill, jump, faction-warfare and
+  incursion data.
+- A validated JSON Schema vocabulary for describing alerts, reports and price
+  thresholds, so an agent's own automation has a well-defined config to act on.
+  The skill validates such a config and checks it against the granted scopes; it
+  does not itself poll, schedule or deliver notifications.
+- Read-only by default: any state-changing request is refused unless the
+  operator explicitly passes `--allow-write`.
+
+## ESI usage
+
+- Requests go to `https://esi.evetech.net` with no version prefix and carry an
+  `X-Compatibility-Date` header pinned to a reviewed date, rather than relying
+  on the oldest supported behaviour.
+- Every ESI and SSO request sends a `User-Agent` naming the skill, its version
+  and this repository; operators can add a contact address via an environment
+  variable.
+- `429` responses are retried after `Retry-After` and `420` after the
+  error-limit reset, with a bounded retry budget. Cache expiry is surfaced to
+  the caller rather than worked around.
+
+## Authentication and scopes
+
+Login uses EVE SSO with OAuth2 and PKCE, so there is no client secret to leak.
+Tokens are written to a `chmod 600` file on the operator's own machine and
+refreshed in-process; the skill is built so that a token never has to appear on
+a command line or in a log.
+
+Scopes are requested by profile rather than all at once — the default profile
+asks for eight read scopes, and wallet and mail are only included if the
+operator explicitly chooses the widest profile. The flow prints the exact scope
+list before opening the browser.
+
+## Self-hosting
+
+MIT licensed, Python standard library only, no pip dependencies. Requires an
+OpenClaw installation and an EVE Developer application for the Client ID. See
+the GitHub README for setup and configuration.