Skills Overview
The skill ecosystem — what a skill is, how one is invoked, and the domain families that make up the catalogue.
A skill is a self-contained, invocable capability — a named unit of procedural knowledge with its own instructions, references, and optional assets. Skills are how LISA turns a recurring task ("classify this note", "plan the sprint", "commit the session to memory") into a repeatable, versioned protocol rather than an ad-hoc improvisation.
The catalogue
As of the latest registry reconciliation the catalogue holds 96 custom skills, authoritatively counted in the skills registry and reconciled against the LisaOS map. Plugin-provided skills (design, frontend, workflow helpers) are catalogued separately under plugins; they are not part of the custom-skill tally.
How a skill is invoked
- Model-invoked — LISA (or a dispatched agent) recognises that a task matches a skill's trigger and invokes it. Most skills work this way; the skill's description is the matching surface.
- Operator-invoked — some skills are marked operator-only and never fire autonomously; they run only when the operator names them.
- Paired with automation — a handful of skills have a scheduled counterpart that runs the same protocol unattended (for example, nightly memory consolidation and weekly reconciliation). See automation.
Every skill carries a name and a description; agents discover skills by domain through a shared tooling index, so domain defines activation priority, not access. Any agent can reach any skill.
Domain families
The catalogue is grouped by function. The families below are the shape of the ecosystem, not an exhaustive list.
| Family | What it covers | Representative skills |
|---|---|---|
| Session & system | Activating and closing sessions, daily-note processing, disk hygiene | session activation, session close, memory consolidation, reconciliation |
| Governance & pipeline | Change scoping and agent-definition governance | change-scoping (the Fuda gate), agent-definition tuning |
| Planning & operations | Missions, sprints, seeds, invoicing | mission init, sprint planning, seed lifecycle, invoice engine |
| Writing & language | Voice, style, and prompt quality | voice replication, style rewriting, text and prompt improvement |
| Creative & visual | Image, video, brand, and character design | image generation, brand kits, character/face/body design, video prompts |
| Vault & knowledge | Classification, coherence, skill lifecycle, branded documents | file classification, coherence auditing, the skill-lifecycle tools, document design |
| Data & analytics | Turning media and channels into structured notes | meeting/chat ingestion, video and transcript extraction |
| Intelligence & monitoring | Situational awareness sweeps | multi-channel intelligence ingestion, standing monitors |
| Ideation & critique | Structuring and stress-testing ideas | idea structuring, idea critique |
| Skill lifecycle | Building and improving skills themselves | loop composition, grading, comparison, analysis |
| Code & engineering | Codebase-facing helpers | whole-codebase audit, feature scoping, issue-tracker integration |
Where skills come from and how they improve
- Lifecycle — how a skill is created, gated for approval, graded, and optimised.
- Improvement — the telemetry, overlay, and curator machinery that lets the skill set improve autonomously over time.