@peonai/swarm 0.1.0

package/skill/GEMINI.md

@@ -0,0 +1,38 @@
+# Swarm AI Integration
+
+You have access to a shared user profile API at `http://localhost:3777`. Use it to personalize responses and record what you learn.
+
+## Session Start
+
+Read the user profile and use it as context:
+
+```bash
+curl -s http://localhost:3777/api/v1/profile \
+  -H "Authorization: Bearer $SWARM_API_KEY"
+```
+
+## During Interaction
+
+When you learn something new about the user, record it:
+
+```bash
+curl -s -X POST http://localhost:3777/api/v1/profile/observe \
+  -H "Content-Type: application/json" \
+  -H "Authorization: Bearer $SWARM_API_KEY" \
+  -d '{"observations":[{"layer":"LAYER","key":"KEY","value":"VALUE","confidence":0.8}]}'
+```
+
+Layers: `identity`, `work`, `preferences`, `context` (24h TTL).
+
+## Memory
+
+```bash
+# Search
+curl -s "http://localhost:3777/api/v1/memory?q=QUERY" -H "Authorization: Bearer $SWARM_API_KEY"
+
+# Write
+curl -s -X POST http://localhost:3777/api/v1/memory \
+  -H "Content-Type: application/json" \
+  -H "Authorization: Bearer $SWARM_API_KEY" \
+  -d '{"content":"CONTENT","tags":["tag1"]}'
+```

package/skill/IFLOW.md

@@ -0,0 +1,38 @@
+# Swarm AI Integration
+
+You have access to a shared user profile API at `http://localhost:3777`. Use it to personalize responses and record what you learn.
+
+## Session Start
+
+Read the user profile and use it as context:
+
+```bash
+curl -s http://localhost:3777/api/v1/profile \
+  -H "Authorization: Bearer $SWARM_API_KEY"
+```
+
+## During Interaction
+
+When you learn something new about the user, record it:
+
+```bash
+curl -s -X POST http://localhost:3777/api/v1/profile/observe \
+  -H "Content-Type: application/json" \
+  -H "Authorization: Bearer $SWARM_API_KEY" \
+  -d '{"observations":[{"layer":"LAYER","key":"KEY","value":"VALUE","confidence":0.8}]}'
+```
+
+Layers: `identity`, `work`, `preferences`, `context` (24h TTL).
+
+## Memory
+
+```bash
+# Search
+curl -s "http://localhost:3777/api/v1/memory?q=QUERY" -H "Authorization: Bearer $SWARM_API_KEY"
+
+# Write
+curl -s -X POST http://localhost:3777/api/v1/memory \
+  -H "Content-Type: application/json" \
+  -H "Authorization: Bearer $SWARM_API_KEY" \
+  -d '{"content":"CONTENT","tags":["tag1"]}'
+```

package/skill/OPENCODE.md

@@ -0,0 +1,38 @@
+# Swarm AI Integration
+
+You have access to a shared user profile API at `http://localhost:3777`. Use it to personalize responses and record what you learn.
+
+## Session Start
+
+Read the user profile and use it as context:
+
+```bash
+curl -s http://localhost:3777/api/v1/profile \
+  -H "Authorization: Bearer $SWARM_API_KEY"
+```
+
+## During Interaction
+
+When you learn something new about the user, record it:
+
+```bash
+curl -s -X POST http://localhost:3777/api/v1/profile/observe \
+  -H "Content-Type: application/json" \
+  -H "Authorization: Bearer $SWARM_API_KEY" \
+  -d '{"observations":[{"layer":"LAYER","key":"KEY","value":"VALUE","confidence":0.8}]}'
+```
+
+Layers: `identity`, `work`, `preferences`, `context` (24h TTL).
+
+## Memory
+
+```bash
+# Search
+curl -s "http://localhost:3777/api/v1/memory?q=QUERY" -H "Authorization: Bearer $SWARM_API_KEY"
+
+# Write
+curl -s -X POST http://localhost:3777/api/v1/memory \
+  -H "Content-Type: application/json" \
+  -H "Authorization: Bearer $SWARM_API_KEY" \
+  -d '{"content":"CONTENT","tags":["tag1"]}'
+```

package/skill/swarm-ai-skill/SKILL.md

@@ -0,0 +1,74 @@
+# Swarm AI — Agent Skill
+
+跨 Agent 用户画像共享。让每个 agent 都了解用户，不用反复教育。
+
+## 安装
+
+### 1. 配置 API Key
+
+编辑 `scripts/env.sh`：
+```bash
+export SWARM_API_URL="http://localhost:3777"
+export SWARM_API_KEY="your_api_key_here"
+```
+
+### 2. 接入你的 Agent 平台
+
+#### OpenClaw
+在 `AGENTS.md` 的「Every Session」部分加一行：
+```
+- Run `bash ~/.openclaw/workspace/skills/swarm-ai-skill/scripts/bootstrap.sh` to load user profile context
+```
+
+#### Claude Code
+在项目根目录 `CLAUDE.md` 加：
+```
+At the start of every task, run: bash /path/to/skills/swarm-ai-skill/scripts/bootstrap.sh
+When you learn something new about the user, run: bash /path/to/skills/swarm-ai-skill/scripts/observe.sh '[{"key":"...","value":"..."}]'
+```
+
+#### Cursor
+在 `.cursorrules` 加：
+```
+At session start, execute: bash /path/to/skills/swarm-ai-skill/scripts/bootstrap.sh
+Use the output as user context for personalization.
+```
+
+#### 其他平台
+任何能执行 shell 的 agent，在 system prompt 里加：
+```
+Before responding, run `scripts/bootstrap.sh` and use the output as user context.
+```
+
+## 脚本
+
+`scripts/` 目录下，全部 bash，零依赖（只需 curl）：
+
+| 脚本 | 用途 | 调用时机 |
+|------|------|----------|
+| `bootstrap.sh` | 读画像，输出上下文文本 | 会话开始 |
+| `profile-read.sh [layer] [tag]` | 读画像 JSON | 需要查具体字段 |
+| `profile-update.sh <layer> <json>` | 更新画像 | 确认了新的用户信息 |
+| `observe.sh <json_array>` | 提交观察 | 发现用户偏好/习惯 |
+| `memory-read.sh [query]` | 搜索记忆 | 需要历史上下文 |
+| `memory-write.sh <content> [tags]` | 写入记忆 | 记录重要事件 |
+
+## 什么时候该写入
+
+Agent 应该在以下场景调用 `observe.sh`：
+- 用户明确表达偏好（「我喜欢用 TypeScript」）
+- 用户纠正了你的假设（「不是 Vue，是 React」）
+- 发现用户的工作习惯（沟通风格、时区、角色）
+- 项目上下文变化（新项目、技术栈变更）
+
+**不要**记录：敏感信息、临时情绪、一次性指令。
+
+## 画像结构
+
+按 layer 组织，名称自由定义：
+- `identity` — 姓名、语言、时区
+- `work` — 技术栈、角色、项目
+- `communication` — 沟通风格
+- `context` — 当前状态（默认 24h 过期）
+
+每条有 `confidence`（0-1），observe 时高 confidence 覆盖低的，不同 agent 的观察互不干扰。

package/skill/swarm-ai-skill/env.sh

@@ -0,0 +1,4 @@
+#!/bin/bash
+# Swarm AI Skill - env config
+export SWARM_API_URL="http://localhost:3777"
+export SWARM_API_KEY="${SWARM_API_KEY:-}"

package/skill/swarm-ai-skill/scripts/bootstrap.sh

@@ -0,0 +1,21 @@
+#!/bin/bash
+# swarm-ai bootstrap — 会话开始时调用，输出用户画像上下文
+# 输出纯文本，直接注入 system prompt / 上下文
+set -euo pipefail
+source "$(dirname "$0")/env.sh"
+
+PROFILE=$(curl -sf "${SWARM_API_URL}/api/v1/profile" \
+  -H "Authorization: Bearer ${SWARM_API_KEY}" 2>/dev/null || echo '{}')
+
+# 空画像不输出
+[ "$PROFILE" = "{}" ] && exit 0
+
+cat <<EOF
+[Swarm AI — User Profile Context]
+The following is known about the user from cross-agent shared memory.
+Use this to personalize your responses. When you learn new information
+about the user, call the observe script to record it.
+
+$PROFILE
+[End Profile Context]
+EOF

package/skill/swarm-ai-skill/scripts/env.sh

@@ -0,0 +1,4 @@
+#!/bin/bash
+# Swarm AI env config — edit these values for your agent
+export SWARM_API_URL="${SWARM_API_URL:-http://localhost:3777}"
+export SWARM_API_KEY="${SWARM_API_KEY:-REPLACE_ME}"

package/skill/swarm-ai-skill/scripts/env.sh.example

@@ -0,0 +1,3 @@
+#!/bin/bash
+export SWARM_API_URL="${SWARM_API_URL:-http://localhost:3777}"
+export SWARM_API_KEY="${SWARM_API_KEY:-REPLACE_ME}"

package/skill/swarm-ai-skill/scripts/memory-read.sh

@@ -0,0 +1,8 @@
+#!/bin/bash
+# swarm-ai memory search/list
+# Usage: ./memory-read.sh [search_query]
+set -euo pipefail
+source "$(dirname "$0")/env.sh"
+Q="${1:-}"
+[ "$Q" ] && P="?q=$Q" || P=""
+curl -sf "${SWARM_API_URL}/api/v1/memory${P}" -H "Authorization: Bearer ${SWARM_API_KEY}"

package/skill/swarm-ai-skill/scripts/memory-write.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+# swarm-ai memory write
+# Usage: ./memory-write.sh <content> [tags_comma_separated]
+set -euo pipefail
+source "$(dirname "$0")/env.sh"
+TAGS="[]"
+[ "${2:-}" ] && TAGS="[$(echo "$2" | sed 's/[^,]*/\"&\"/g')]"
+curl -sf -X POST "${SWARM_API_URL}/api/v1/memory" \
+  -H "Content-Type: application/json" -H "Authorization: Bearer ${SWARM_API_KEY}" \
+  -d "{\"content\":$(echo "$1" | python3 -c 'import json,sys;print(json.dumps(sys.stdin.read().strip()))'),\"tags\":$TAGS}"

package/skill/swarm-ai-skill/scripts/observe.sh

@@ -0,0 +1,9 @@
+#!/bin/bash
+# swarm-ai observe
+# Usage: ./observe.sh <json_observations>
+# Example: ./observe.sh '[{"key":"mood","value":"focused","confidence":0.8}]'
+set -euo pipefail
+source "$(dirname "$0")/env.sh"
+curl -sf -X POST "${SWARM_API_URL}/api/v1/profile/observe" \
+  -H "Content-Type: application/json" -H "Authorization: Bearer ${SWARM_API_KEY}" \
+  -d "{\"observations\":$1}"

package/skill/swarm-ai-skill/scripts/profile-read.sh

@@ -0,0 +1,9 @@
+#!/bin/bash
+# swarm-ai profile read
+# Usage: ./profile-read.sh [layer] [tag]
+set -euo pipefail
+source "$(dirname "$0")/env.sh"
+PARAMS=""
+[ "${1:-}" ] && PARAMS="?layer=$1"
+[ "${2:-}" ] && PARAMS="${PARAMS:+$PARAMS&}${PARAMS:+}${PARAMS:-?}tag=$2"
+curl -sf "${SWARM_API_URL}/api/v1/profile${PARAMS}" -H "Authorization: Bearer ${SWARM_API_KEY}"

package/skill/swarm-ai-skill/scripts/profile-update.sh

@@ -0,0 +1,9 @@
+#!/bin/bash
+# swarm-ai profile update
+# Usage: ./profile-update.sh <layer> <json_entries>
+# Example: ./profile-update.sh work '{"tech_stack":{"value":["TS","React"],"tags":["dev"]}}'
+set -euo pipefail
+source "$(dirname "$0")/env.sh"
+curl -sf -X PATCH "${SWARM_API_URL}/api/v1/profile" \
+  -H "Content-Type: application/json" -H "Authorization: Bearer ${SWARM_API_KEY}" \
+  -d "{\"layer\":\"$1\",\"entries\":$2}"

package/skill/swarm-ai-skill/scripts/session-start.sh

@@ -0,0 +1,19 @@
+#!/bin/bash
+# swarm-ai session-start — 会话开始时调用
+# 记录会话启动时间，用于后续 reflect 计算"本次会话"的记忆范围
+set -euo pipefail
+source "$(dirname "$0")/env.sh"
+
+SESSION_FILE="${SWARM_SESSION_FILE:-/tmp/swarm-session-${SWARM_AGENT_ID:-unknown}.json}"
+
+# 写入会话起始标记
+cat > "$SESSION_FILE" <<EOF
+{
+  "agent_id": "${SWARM_AGENT_ID:-unknown}",
+  "started_at": "$(date -u +%Y-%m-%dT%H:%M:%SZ)",
+  "pid": $$
+}
+EOF
+
+# 同时输出画像上下文（保持 bootstrap 的原有功能）
+"$(dirname "$0")/bootstrap.sh"

package/tsconfig.json

@@ -0,0 +1,21 @@
+{
+  "compilerOptions": {
+    "target": "ES2022",
+    "lib": ["dom", "dom.iterable", "esnext"],
+    "allowJs": true,
+    "skipLibCheck": true,
+    "strict": true,
+    "noEmit": true,
+    "esModuleInterop": true,
+    "module": "esnext",
+    "moduleResolution": "bundler",
+    "resolveJsonModule": true,
+    "isolatedModules": true,
+    "jsx": "preserve",
+    "incremental": true,
+    "plugins": [{ "name": "next" }],
+    "paths": { "@/*": ["./*"] }
+  },
+  "include": ["next-env.d.ts", "**/*.ts", "**/*.tsx", ".next/types/**/*.ts"],
+  "exclude": ["node_modules"]
+}