| name | workspace-navigation |
| model | sonnet |
| description | Produces correct file placement, frontmatter, and navigation paths for shared agent workspaces. Use when: 'organize this in the workspace', 'where does this belong in the structure', 'find the decision that answers this', 'set up a collaboration repo'. |
| triggers | ["organize this in the workspace","where does this belong in the structure","navigate to the relevant research","find the decision that answers this","set up a collaboration repo"] |
| metadata | {"version":"1.1","created":"2026-02-02","author":"Manus","tool_dependencies":["file_system"],"portable":true,"tier":2,"agents":["agent-orchestration"]} |
| category | agent-orchestration |
| inputs | [{"name":"artifact","type":"string","description":"The file, decision, or content item to place or locate in the workspace","required":true},{"name":"workspace_root","type":"string","description":"Root path of the shared agent workspace","required":false}] |
| outputs | [{"name":"placement_decision","type":"string","description":"Correct file placement, frontmatter, and navigation paths for the workspace artifact"}] |
Agent Workspace Navigator Skill
I. Philosophy
A shared workspace is not a file dump — it is a thinking room. Structure enables clarity; clarity enables collaboration. This skill encodes best practices for reading, writing, and organizing content in shared agent repositories so that any agent can orient in under 1,000 tokens and contribute without creating disorder.
II. When to Use
- Working in a shared private repository with other agents
- Contributing to or reading from a collaborative discussion space
- Organizing thoughts, specifications, or research in a structured way
- Coordinating work across multiple agents or sessions
- Building shared context without unstructured file chaos
III. Workspace Structure
Standard Directory Layout
workspace/
├── README.md # Workspace overview and navigation guide
├── 00_Active/ # Current work in progress
│ ├── discussions/ # Active discussions and threads
│ ├── drafts/ # Work in progress (not final)
│ └── decisions/ # Decisions made (with rationale)
├── 01_Specifications/ # Finalized specs and designs
│ ├── architecture/
│ ├── features/
│ └── protocols/
├── 02_Research/ # Research findings and synthesis
│ ├── deep-dives/
│ ├── landscape-scans/
│ └── references/
├── 03_Memory/ # Shared memory and knowledge base
│ ├── seeds/
│ ├── learnings/
│ └── context/
├── 04_Artifacts/ # Generated artifacts and outputs
│ ├── code/
│ ├── diagrams/
│ └── documents/
└── 05_Archive/ # Completed or deprecated work
├── by-date/
└── by-topic/
File Naming Convention
Format: YYYY-MM-DD_topic-name_author.md
Examples: 2026-02-02_memory-garden-design_manus.md, 2026-02-02_collaboration-protocol_manus-cipher.md
Date prefix enables sorting; topic name provides context; author attribution enables tracking.
IV. Reading Workflow
Step 1: Read README.md first to understand workspace purpose, active discussions, and how to contribute.
Step 2: Navigate by purpose, not time. Identify your goal, go to the relevant directory, read the most recent relevant file.
Step 3: Use Grep for targeted search:
grep -r "memory compression" workspace/ --include="*.md"
Step 4: Read frontmatter before body. Check status, date, and tags to decide if the file is worth reading in full.
