Octomind 0.25.0 adds declarative auto-activation rules for skills
Octomind 0.25.0 overhauls its skill system so agent skills auto-activate based on declarative rules — checking files, content, environment variables, and more — eliminating manual context loading at the start of every session.
Score breakdown
Developers using Octomind can now drop context-bootstrapping entirely — the agent arrives at the first message already loaded with project-appropriate knowledge, making AI sessions immediately productive without manual skill management.
- 01Octomind 0.25.0 replaces manual `/skill` loading with a declarative `rules:` field in skill definitions.
- 02Rule types include `file()`, `content()`, `bin()`, `env()`, `grep()`, `match()`, `session()`, and `workdir()` checks.
- 03Rules use OR-group logic per line and AND-group logic for space-separated checks within a line.
Octomind 0.25.0, authored by Ava Nuts, introduces a declarative activation system for skills — the release's biggest change since the skill system was originally introduced. Previously (in 0.24.0), skills were opt-in via the `/skill` command, meaning users often spent the first several messages of a session manually bootstrapping context. The new `rules:` field in skill definitions lets each skill declare exactly when it should activate, using a set of in-process checks evaluated on every user message.
The rule syntax supports OR-groups (one rule per line) and AND-groups (space-separated checks within a line).
The rule syntax supports OR-groups (one rule per line) and AND-groups (space-separated checks within a line). Available check types include `file()` for file or glob existence, `content()` for word-boundary matches in the user's message, `grep()` for file content searches respecting `.gitignore`, `env()` for environment variable presence, `match()` for regex matches, `bin()` for PATH executable checks, `session()` for session name patterns, and `workdir()` for working directory path matches. A `domains:` field further scopes auto-activation to specific agent roles — for example, a skill with `domains: developer` won't activate when a writing agent is in use, preventing skill spam.
For cases where skills should be active from the very first message, the `OCTOMIND_SKILLS` environment variable preloads named skills during session startup. Octomind also reads `.env` files from the working directory automatically, enabling per-project skill and configuration setups without manual intervention. Manual skill management via `/skill` commands remains available for overrides.
Key facts
- 01Octomind 0.25.0 replaces manual `/skill` loading with a declarative `rules:` field in skill definitions.
- 02Rule types include `file()`, `content()`, `bin()`, `env()`, `grep()`, `match()`, `session()`, and `workdir()` checks.
- 03Rules use OR-group logic per line and AND-group logic for space-separated checks within a line.
- 04All rules are evaluated in-process on every user message — no subprocess spawning or network calls.
- 05The `domains:` field scopes auto-activation to specific agent roles, preventing irrelevant skills from loading.
- 06The `OCTOMIND_SKILLS` environment variable preloads skills before the first message, useful for CI environments.
- 07Octomind automatically reads `.env` files from the working directory for per-project configuration.
Topics
Summary and scoring are generated automatically from the original article. We always link back to the publisher and never republish images or paywalled content. Last processed Apr 23, 2026 · 11:04 UTC. How this works →