Most platforms let you add tools. This one lets you add whole new abilities to how your crew works: a writing style, a memory upgrade, a scheduling behavior, an entire prebuilt team. Browse the catalog, install what fits, and it's ready the next time an agent runs.
AllSkillsMemoryAgent packsChannelsProvidersTools64 published · 9 new this week
Caveman
Prompt skill · v2.1.0
Output brevity. Appends terseness guidance so answers come back dramatically shorter, while keeping code, commands, error messages and URLs byte-for-byte exact.
fewer tokens · same accuracy
Ponytail
Prompt skill · v1.4.2
Code minimization. Appends a YAGNI decision ladder so the model reaches for reuse, the standard library and native features before writing anything new.
smallest correct solution
Pi Compactor
Memory upgrade · v0.9.7
Replaces the built-in summarizer. Rewrites history into Goal / Progress / Key Decisions / Next Steps and refines that summary on every later compaction.
structured memory
Ship's Company
Agent pack · v3.0.1
Seven specialists in one install — orchestrator, planner, deep worker, architect, researcher, navigator, visual analyst — plus an orchestration doctrine and five slash commands.
a crew, not a blank agent
Quiet Hours
Scheduling skill · v1.0.4
Cancels scheduled fires outside working hours and annotates the ones it lets through, so overnight missions do not page anyone.
intercepts the fire
Matrix Bridge
Channel · v2.2.0
Routes a Matrix room to an agent, with per-room agent selection and thread-aware replies.
one more way in
Ledger
Tool · v1.1.0
Gives an agent read access to your cost data so it can budget its own runs and refuse work that would blow a quota.
spend-aware agents
Local Bedrock
Model provider · v0.6.3
Adds a provider for models behind your own gateway, with per-token pricing you define.
your endpoint, your price
WHAT YOU CAN ADD
Five kinds of upgrade, in order of how much they change.
The first three are the ones nobody else offers — most platforms stop at the last two.
Prompt skillsChange how every answer reads, for every agent that has it installed. This is how Caveman and Ponytail work.
Memory upgradesChange what happens when a long session's context fills, so a mission doesn't lose the thread.
Scheduling behaviorShape what a standing mission does when it fires and when it finishes — annotate it, or cancel it outright.
Agent packsProvision a whole crew — souls, permissions and slash commands as a unit — and keep it healthy on every session start.
Tools · providers · channels · commandsThe ordinary extensions, for when you just need one more capability.
INSTALLING
Browse, install, done.
Every listing tells you plainly what it gives an agent before you commit — no config file to hand-edit, no service to restart, no coordination with anyone else on the account. Confirm the install and it's ready the next time an agent runs.
Versions are immutable and semantic. An install pins a version; an upgrade is a decision you make, with a diff of what changed.
search by capability · filter by kind · see what an install changes before you commit
02install
one click · nothing to configure by hand · nothing to restart
03live
your crew has it starting with the very next session
what changes is on the card before you install, not in a changelog after
SUPPLY SIDE
You can publish, too.
The catalog is only as good as the people building for it. Authoring a plugin is a TypeScript project against a typed SDK, published under an immutable version, and installed by anyone who wants it.