Skip to content

Commit 8fab17c

Browse files
authored
Integrate curator and banthis plugins (#354)
1 parent e42f403 commit 8fab17c

20 files changed

Lines changed: 285 additions & 50 deletions

.claude-plugin/marketplace.json

Lines changed: 41 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1,7 +1,7 @@
11
{
22
"name": "agentsys",
3-
"description": "21 specialized plugins for AI workflow automation - task orchestration, PR workflow, slop detection, code review, drift detection, enhancement analysis, documentation sync, unified static analysis, durable memory, perf investigations, topic research, agent config linting, cross-tool AI consultation, structured AI debate, workflow pattern learning, codebase onboarding, contributor guidance, and Zig language support",
4-
"version": "5.12.0",
3+
"description": "24 specialized plugins for AI workflow automation - task orchestration, PR workflow, slop detection, code review, drift detection, enhancement analysis, documentation sync, unified static analysis, durable memory, negative behavior memory, skill and system prompt curation, perf investigations, topic research, agent config linting, cross-tool AI consultation, structured AI debate, workflow pattern learning, codebase onboarding, contributor guidance, and Zig language support",
4+
"version": "5.13.0",
55
"owner": {
66
"name": "Avi Fenesh",
77
"url": "https://github.com/avifenesh"
@@ -122,6 +122,32 @@
122122
"category": "development",
123123
"homepage": "https://github.com/agent-sh/enhance"
124124
},
125+
{
126+
"name": "skill-curator",
127+
"source": {
128+
"source": "url",
129+
"url": "https://github.com/agent-sh/skill-curator.git",
130+
"ref": "v1.0.0",
131+
"commit": "e30ebfe3e7a5307e97fa4e6d59f538b12f8074c5"
132+
},
133+
"description": "Canonical guidance for writing, reviewing, and maintaining production-grade SKILL.md files across agent tools",
134+
"version": "1.0.0",
135+
"category": "development",
136+
"homepage": "https://github.com/agent-sh/skill-curator"
137+
},
138+
{
139+
"name": "system-prompt-curator",
140+
"source": {
141+
"source": "url",
142+
"url": "https://github.com/agent-sh/system-prompt-curator.git",
143+
"ref": "v2.0.0",
144+
"commit": "277bee4b8b3b796d2a627132c88fb132479cbde0"
145+
},
146+
"description": "Production-grade system prompt curation for autonomous coding agents and multi-agent orchestration",
147+
"version": "2.0.0",
148+
"category": "development",
149+
"homepage": "https://github.com/agent-sh/system-prompt-curator"
150+
},
125151
{
126152
"name": "sync-docs",
127153
"source": {
@@ -184,6 +210,19 @@
184210
"category": "productivity",
185211
"homepage": "https://github.com/agent-sh/axiom"
186212
},
213+
{
214+
"name": "banthis",
215+
"source": {
216+
"source": "url",
217+
"url": "https://github.com/agent-sh/banthis.git",
218+
"ref": "v0.3.0",
219+
"commit": "5820a197bba0f01ec4877a63c5d13c108d8c661d"
220+
},
221+
"description": "Durable negative behavior memory: persist user-defined banned agent behaviors into CLAUDE.md or AGENTS.md",
222+
"version": "0.3.0",
223+
"category": "productivity",
224+
"homepage": "https://github.com/agent-sh/banthis"
225+
},
187226
{
188227
"name": "agnix",
189228
"source": {

.claude-plugin/plugin.json

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
{
22
"name": "agentsys",
3-
"version": "5.12.0",
3+
"version": "5.13.0",
44
"description": "Professional-grade slash commands for Claude Code with cross-platform support",
55
"keywords": [
66
"workflow",

.codex-plugin/plugin.json

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -4,8 +4,8 @@
44
"skills": "./adapters/codex/skills",
55
"interface": {
66
"displayName": "agentsys",
7-
"shortDescription": "AI agent orchestration with 42 skills and 49 agents",
8-
"longDescription": "Professional-grade slash commands for AI-powered development workflows. Includes /next-task (task discovery to production), /axiom (durable agent memory), /ship (commit to deploy), /audit-project (multi-agent code review), /deslop (AI slop cleanup), /perf (performance investigation), /enhance (config analysis), /consult (cross-tool AI consultation), and more.",
7+
"shortDescription": "AI agent orchestration with 45 skills and 49 agents",
8+
"longDescription": "Professional-grade slash commands for AI-powered development workflows. Includes /next-task (task discovery to production), /axiom (durable agent memory), /banthis (negative behavior memory), /skill-curator and /system-prompt-curator (agent configuration curation), /ship (commit to deploy), /audit-project (multi-agent code review), /deslop (AI slop cleanup), /perf (performance investigation), /enhance (config analysis), /consult (cross-tool AI consultation), and more.",
99
"developerName": "Avi Fenesh",
1010
"category": "developer-tools",
1111
"capabilities": [

AGENTS.md

Lines changed: 8 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -76,7 +76,7 @@
7676
<!-- GEN:START:claude-architecture -->
7777
```
7878
lib/ → Shared library (vendored to plugins)
79-
plugins/ → 21 plugins, 49 agents (39 file-based + 10 role-based), 42 skills
79+
plugins/ → 24 plugins, 49 agents (39 file-based + 10 role-based), 45 skills
8080
adapters/ → Platform adapters (opencode-plugin/, opencode/, codex/)
8181
checklists/ → Action checklists (9 files)
8282
bin/cli.js → npm CLI installer
@@ -95,12 +95,15 @@ bin/cli.js → npm CLI installer
9595
| sync-docs | 1 | 1 | Documentation sync |
9696
| repo-intel | 1 | 1 | Unified static analysis |
9797
| axiom | 0 | 1 | Durable agent-native memory |
98+
| banthis | 0 | 1 | Durable negative behavior memory |
9899
| perf | 6 | 8 | Performance investigation |
99100
| learn | 1 | 1 | Topic research and learning guides |
100101
| agnix | 1 | 1 | Agent config linting |
101102
| consult | 1 | 1 | Cross-tool AI consultation |
102103
| debate | 1 | 1 | Multi-perspective debate analysis |
103104
| web-ctl | 1 | 2 | Browser automation for AI agents |
105+
| skill-curator | 0 | 1 | Skill authoring and review |
106+
| system-prompt-curator | 0 | 1 | System prompt curation |
104107
| skillers | 2 | 2 | Workflow pattern learning |
105108
| onboard | 1 | 1 | Codebase onboarding |
106109
| can-i-help | 1 | 1 | Contributor guidance |
@@ -173,7 +176,7 @@ agentsys # Run installer
173176
<agents>
174177
## Agents
175178

176-
49 agents across 21 plugins (18 have agents; gate-and-ship is commands-only; axiom is skill/command-only; zig-lsp is config-only with no commands or agents). Key agents by model:
179+
49 agents across 24 plugins (18 have agents; gate-and-ship is commands-only; axiom, banthis, skill-curator, and system-prompt-curator are skill/command-only; zig-lsp is config-only with no commands or agents). Key agents by model:
177180

178181
| Model | Agents | Use Case |
179182
|-------|--------|----------|
@@ -187,15 +190,15 @@ See [README.md](./README.md#command-details) and [docs/reference/AGENTS.md](./do
187190
<skills>
188191
## Skills
189192

190-
42 skills across plugins. Agents invoke skills for reusable implementation.
193+
45 skills across plugins. Agents invoke skills for reusable implementation.
191194

192195
| Category | Key Skills |
193196
|----------|------------|
194197
| Workflow | `orchestrate-review`, `discover-tasks`, `validate-delivery` |
195-
| Enhancement | `enhance-*` (9 skills for plugins, agents, docs, prompts, hooks) |
198+
| Enhancement | `enhance-*` (9 skills for plugins, agents, docs, prompts, hooks), `skill-curator`, `system-prompt-curator` |
196199
| Performance | `baseline`, `benchmark`, `profile`, `theory-tester` |
197200
| Cleanup | `deslop`, `sync-docs`, `drift-analysis`, `repo-intel` |
198-
| Memory | `axiom` |
201+
| Memory | `axiom`, `banthis` |
199202

200203
See [README.md](./README.md#skills) for full skill list.
201204
</skills>

CHANGELOG.md

Lines changed: 8 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -9,8 +9,16 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
99

1010
## [Unreleased]
1111

12+
## [5.13.0] - 2026-05-17
13+
1214
### Added
1315
- Added the `axiom` plugin to the agentsys marketplace and static docs metadata. Axiom provides durable agent-native memory, project context loading, scoped querying, project scaffolding, and human-approved record proposals.
16+
- Added `banthis` as a published standalone plugin pinned to `v0.3.0`. It provides durable negative behavior memory through a tiny CLI, skill, and slash command.
17+
- Added `skill-curator` as a published standalone plugin pinned to `v1.0.0` for production-grade `SKILL.md` authoring and review.
18+
- Added `system-prompt-curator` as a published standalone plugin pinned to `v2.0.0` for autonomous coding-agent system prompt curation.
19+
20+
### Changed
21+
- Updated marketplace, Codex metadata, docs, and website counts to 24 plugins, 49 agents, and 45 skills.
1422

1523
## [5.12.0] - 2026-04-26
1624

README.md

Lines changed: 66 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -19,7 +19,7 @@
1919
</p>
2020

2121
<p align="center">
22-
<b>21 plugins · 49 agents · 42 skills (across all repos) · 30k lines of lib code · 3,513 tests · 5 platforms</b><br>
22+
<b>24 plugins · 49 agents · 45 skills (across all repos) · 30k lines of lib code · 3,513 tests · 5 platforms</b><br>
2323
<em>Plugins distributed as standalone repos under <a href="https://github.com/agent-sh">agent-sh</a> org - agentsys is the marketplace &amp; installer</em>
2424
</p>
2525

@@ -45,7 +45,7 @@ AI models can write code. That's not the hard part anymore. The hard part is eve
4545
4646
## What This Is
4747

48-
An agent orchestration system - 21 plugins, 49 agents (39 file-based + 10 role-based specialists in audit-project), and 42 skills that compose into structured pipelines for software development. Each plugin lives in its own standalone repo under the [agent-sh](https://github.com/agent-sh) org. agentsys is the marketplace and installer that ties them together.
48+
An agent orchestration system - 24 plugins, 49 agents (39 file-based + 10 role-based specialists in audit-project), and 45 skills that compose into structured pipelines for software development. Each plugin lives in its own standalone repo under the [agent-sh](https://github.com/agent-sh) org. agentsys is the marketplace and installer that ties them together.
4949

5050
Each agent has a single responsibility, a specific model assignment, and defined inputs/outputs. Pipelines enforce phase gates so agents can't skip steps. State persists across sessions so work survives interruptions.
5151

@@ -119,6 +119,7 @@ The investment shifts from model spend to pipeline design. Better prompts, riche
119119
| [`/prepare-delivery`](#prepare-delivery) | Pre-ship quality gates: deslop, review, validation, docs sync |
120120
| [`/gate-and-ship`](#gate-and-ship) | Quality gates then ship (/prepare-delivery + /ship) |
121121
| [`/axiom`](#axiom) | Durable memory: load, query, list, bootstrap projects, and record approved knowledge |
122+
| [`/banthis`](#banthis) | Durable negative memory: persist banned agent behaviors |
122123
| [`/agnix`](#agnix) | Lint agent configurations (399 rules) |
123124
| [`/ship`](#ship) | PR creation, CI monitoring, merge |
124125
| [`/deslop`](#deslop) | Clean AI slop patterns |
@@ -134,6 +135,8 @@ The investment shifts from model spend to pipeline design. Better prompts, riche
134135
| [`/web-ctl`](#web-ctl) | Browser automation for AI agents |
135136
| [`/release`](#release) | Versioned release with ecosystem detection |
136137
| [`/skillers`](#skillers) | Workflow pattern learning and automation |
138+
| [`/skill-curator`](#skill-curator) | Create and improve reliable SKILL.md files |
139+
| [`/system-prompt-curator`](#system-prompt-curator) | Create and improve autonomous agent system prompts |
137140
| [`/onboard`](#onboard) | Codebase orientation for newcomers |
138141
| [`/can-i-help`](#can-i-help) | Match contributor skills to project needs |
139142

@@ -143,13 +146,13 @@ Each command works standalone. Together, they compose into end-to-end pipelines.
143146

144147
## Skills
145148

146-
42 skills included across the plugins:
149+
45 skills included across the plugins:
147150

148151
| Category | Skills |
149152
|----------|--------|
150153
| **Workflow** | `discover-tasks`, `prepare-delivery`, `check-test-coverage`, `orchestrate-review`, `validate-delivery` |
151154
| **Message Queues** | `glide-mq-migrate-bee`, `glide-mq-migrate-bullmq`, `glide-mq` |
152-
| **Enhancement** | `enhance-agent-prompts`, `enhance-claude-memory`, `enhance-cross-file`, `enhance-docs`, `enhance-hooks`, `enhance-orchestrator`, `enhance-plugins`, `enhance-prompts`, `enhance-skills` |
155+
| **Enhancement** | `enhance-agent-prompts`, `enhance-claude-memory`, `enhance-cross-file`, `enhance-docs`, `enhance-hooks`, `enhance-orchestrator`, `enhance-plugins`, `enhance-prompts`, `enhance-skills`, `skill-curator`, `system-prompt-curator` |
153156
| **Performance** | `baseline`, `benchmark`, `code-paths`, `investigation-logger`, `perf-analyzer`, `profile`, `theory-gatherer`, `theory-tester` |
154157
| **Cleanup** | `deslop`, `sync-docs` |
155158
| **Code Review** | `audit-project` |
@@ -158,7 +161,7 @@ Each command works standalone. Together, they compose into end-to-end pipelines.
158161
| **Web** | `web-auth`, `web-browse` |
159162
| **Release** | `release` |
160163
| **Analysis** | `drift-analysis`, `repo-intel` |
161-
| **Memory** | `axiom` |
164+
| **Memory** | `axiom`, `banthis` |
162165
| **Linting** | `agnix` |
163166

164167
**External skill plugins** (standalone repos, installed separately):
@@ -177,8 +180,8 @@ Skills are the reusable implementation units. Agents invoke skills; commands orc
177180
|---------|--------------|
178181
| [The Approach](#the-approach) | Why it's built this way |
179182
| [Benchmarks](#benchmarks) | Sonnet + agentsys vs raw Opus |
180-
| [Commands](#commands) | All 21 commands overview |
181-
| [Skills](#skills) | 42 skills across plugins |
183+
| [Commands](#commands) | All 24 commands overview |
184+
| [Skills](#skills) | 45 skills across plugins |
182185
| [Skill-Only Plugins](#skill-only-plugins) | glide-mq and other non-command plugins |
183186
| [Command Details](#command-details) | Deep dive into each command |
184187
| [How Commands Work Together](#how-commands-work-together) | Standalone vs integrated |
@@ -338,6 +341,32 @@ Each piece runs independently - use `/prepare-delivery` alone to review before d
338341

339342
---
340343

344+
### /banthis
345+
346+
**Purpose:** Durable negative memory for repeated agent mistakes. Turn a user's "stop doing this" correction into a persistent rule in `CLAUDE.md` or `AGENTS.md`.
347+
348+
**[banthis](https://github.com/agent-sh/banthis)** is a tiny standalone CLI plus skill. It maintains a managed banned-behaviors section, supports project or global targets, and includes an `init` meta-rule so agents learn when to invoke it automatically.
349+
350+
**What it does:**
351+
352+
| Command | Use |
353+
|---------|-----|
354+
| `banthis add "<title>" "<rule>"` | Add or update a banned behavior |
355+
| `banthis list` | List current bans |
356+
| `banthis show` | Print the managed section |
357+
| `banthis remove "<title>"` | Remove a ban |
358+
| `banthis init` | Install the meta-rule that teaches agents to call `banthis` |
359+
360+
**Usage:**
361+
362+
```bash
363+
/banthis "stop ending with vague optional follow-up offers"
364+
banthis add "No vague endings" "Do not end with vague optional follow-up offers."
365+
banthis init --file AGENTS.md
366+
```
367+
368+
---
369+
341370
### /agnix
342371

343372
**Purpose:** Lint agent configurations before they break your workflow. The first dedicated linter for AI agent configs.
@@ -984,6 +1013,36 @@ No per-turn overhead - it reads transcripts that Claude Code already saves.
9841013

9851014
---
9861015

1016+
### /skill-curator
1017+
1018+
> Create and improve reliable `SKILL.md` files
1019+
1020+
```bash
1021+
/skill-curator "create a skill for reviewing background jobs"
1022+
/skill-curator --improve path/to/SKILL.md --category review
1023+
```
1024+
1025+
The skill curator focuses on trigger quality, clear scope, router patterns, concrete `Skip unless:` gates, token budgets, and agnix-ready structure across Claude Code, Codex, OpenCode, Cursor, Kiro, and similar tools.
1026+
1027+
**Skill:** skill-curator
1028+
1029+
---
1030+
1031+
### /system-prompt-curator
1032+
1033+
> Create and improve autonomous coding-agent system prompts
1034+
1035+
```bash
1036+
/system-prompt-curator "GitHub issue resolver"
1037+
/system-prompt-curator --improve path/to/prompt.md --for-orchestrator
1038+
```
1039+
1040+
The system prompt curator rewrites prompts around task-matched identity, phased workflow, explicit tools, evidence-based completion criteria, and realistic error recovery examples. It separates prompt guidance from harness-level checks that belong in code.
1041+
1042+
**Skill:** system-prompt-curator
1043+
1044+
---
1045+
9871046
### /onboard
9881047

9891048
**Purpose:** Get oriented in any codebase in under 3 minutes.

0 commit comments

Comments
 (0)