スキル一覧に戻る
LPDigital-Agent

context-manager

by LPDigital-Agent

Galderma TrackWise AI Autopilot Demo - 9-agent mesh on AWS Bedrock AgentCore

0🍴 0📅 2026年1月21日
GitHubで見るManusで実行

SKILL.md


name: context-manager description: Context management specialist for multi-agent workflows and long-running tasks. Use PROACTIVELY for complex projects, session coordination, and when context preservation is needed across multiple agents. (project) allowed-tools: Read, Write, Edit, TodoWrite, mcp__memory__create_entities, mcp__memory__add_observations, mcp__memory__search_nodes

Context Manager Skill

Context management specialist for Faiston NEXO multi-agent workflows and long-running tasks.

Faiston NEXO Architecture Context

LayerComponentsKey Files
FrontendReact, TypeScript, TanStack Queryclient/
BackendFastAPI, Python, Mangumserver/main.py
AI AgentsGoogle ADK, Gemini 3.0 Proserver/agentcore/
InfrastructureTerraform, AWS Lambda, S3, DynamoDBterraform/
CI/CDGitHub Actions.github/workflows/

Primary Functions

Context Capture

  1. Extract key decisions and rationale from agent outputs
  2. Identify reusable patterns and solutions
  3. Document integration points between components
  4. Track unresolved issues and TODOs
  5. Faiston-specific: Track AgentCore session IDs and Cognito auth state

Context Distribution

  1. Prepare minimal, relevant context for each agent
  2. Create agent-specific briefings
  3. Maintain a context index for quick retrieval
  4. Prune outdated or irrelevant information

Memory Management (MCP Memory)

Use MCP Memory tools to persist critical context:

mcp__memory__create_entities    → Store new decisions/patterns
mcp__memory__add_observations   → Add details to existing entities
mcp__memory__search_nodes       → Retrieve relevant context

Faiston NEXO Context Templates

Quick Context (< 500 tokens)

## Current Task
[What we're working on]

## Recent Decisions
- [Decision 1 with rationale]
- [Decision 2 with rationale]

## Active Blockers
- [Blocker and owner]

## Key Files Modified
- `path/to/file.ts` - [change description]

Full Context (< 2000 tokens)

## Project State
- **Frontend**: [current status]
- **Backend**: [current status]
- **AgentCore**: [current status]
- **Infrastructure**: [current status]

## Architecture Decisions
| Decision | Rationale | Date |
|----------|-----------|------|
| Use Gemini 3.0 Pro | Claude was collapsing structures | 2025-12 |
| S3 regional endpoint | Fix 307 CORS redirect | 2025-12 |

## Integration Points
- Frontend → AgentCore: Cognito JWT + `agentcore.ts`
- AgentCore → Gemini: Google ADK Agent class
- Backend → S3: Presigned URLs with s3v4

## Active Work Streams
1. [Stream 1]
2. [Stream 2]

AgentCore Session Context

## AgentCore State
- **Runtime**: faiston_nexo_agents-WNYXe1CyLz
- **Memory ID**: faiston_nexo_agents_mem-2LaTp8COvj
- **Auth**: Cognito JWT (pool: us-east-2_6Vzhr0J6M)

## Active Agents
| Agent | Last Invoked | Status |
|-------|--------------|--------|
| NEXOAgent | [time] | [status] |
| FlashcardsAgent | [time] | [status] |

Workflow Integration

When activated:

  1. Review current conversation and agent outputs
  2. Extract important context using templates above
  3. Store in MCP Memory if critical decision
  4. Create summary for next agent/session
  5. Update docs/KNOWN_ISSUES.md if bugs found
  6. Suggest context compression when needed

Cross-Skill Context Handoff

From SkillTo SkillContext Needed
bug-detectivetest-engineerError reproduction steps
ai-engineerfrontend-builderNew agent actions/responses
backend-architectprompt-engineerAPI contract changes
workflow-architectcontext-managerTask decomposition

Memory Entities to Track

Entity: FaistonNEXO_Decision
Type: architectural_decision
Observations:
  - "2025-12: Migrated to Gemini 3.0 Pro for better structure"
  - "2025-12: S3 presigned URLs use regional endpoint"

Entity: FaistonNEXO_KnownIssue
Type: bug
Observations:
  - "S3 307 redirect: Use s3v4 signature with virtual addressing"
  - "Session ID must be >= 33 characters"

Always optimize for relevance over completeness. Good context accelerates work; bad context creates confusion.

スコア

総合スコア

40/100

リポジトリの品質指標に基づく評価

SKILL.md

SKILL.mdファイルが含まれている

+20
LICENSE

ライセンスが設定されている

0/10
説明文

100文字以上の説明がある

0/10
人気

GitHub Stars 100以上

0/15
最近の活動

3ヶ月以内に更新がある

0/10
フォーク

10回以上フォークされている

0/5
Issue管理

オープンIssueが50未満

+5
言語

プログラミング言語が設定されている

+5
タグ

1つ以上のタグが設定されている

0/5

レビュー

💬

レビュー機能は近日公開予定です