openport
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| OPENPORT_MCP_KEY | No | Optional legacy static path key | |
| OPENPORT_SETTINGS_KEY | No | Used for personal auth mode to mint connect URLs | |
| OPENPORT_WRITE_GUARDS | No | strict = headings + live-state bans; relaxed = light checks | strict |
| OPENPORT_SESSION_RETENTION_DAYS | No | How long to keep _session notes | 14 |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| prompts | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| pingC | Health check. |
| start_sessionA | GATE — first tool every work session. Unlocks a skill, loads must-load + one local + _session, returns next_action. Default skill: bootstrap (seed) or resume-work. Prefer this over bare get_context. |
| finish_sessionA | Save a final handoff and close this work session. REQUIRED when finishing work; use update_context for checkpoints while continuing. |
| get_contextA | Discover available scopes with get_context(), or load memory by scope/scopes (must-load and shared scopes auto-included). A scope shorthand allows one local. An explicit scopes list may name several locals. The index returns IDs, kinds, and timestamps only. Prefer start_session once you know the local. |
| learn_workflowA | Unlock a skill (how-to) for this session — the OpenPortable gate. Prefer start_session which does this automatically. Day one seed: bootstrap; later: resume-work. |
| get_docA | Read a markdown doc from memory (getting-started, tools, context, …). |
| update_docC | Replace or append a doc in memory. |
| get_skillB | Read an instruction skill by id, or list with skill="*". |
| update_skillB | Create or update a skill. Available before session setup for installation/recovery; editing a loaded skill requires clients to reload it. |
| update_contextB | Write durable memory or a handoff checkpoint using session_id from start_session. At session end use finish_session to save the final handoff and close the gate. |
| collapse_contextC | Prune |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| resume | Session-start: call start_session, obey next_action, continue from handoff. |
| handoff | Session-end contract: write a session_note for the next client. |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 11 tools
The tools separate sessions, context, docs, and skills, but several boundaries overlap: start_session vs learn_workflow vs get_context, and finish_session vs update_context. The descriptions explicitly tell agents to prefer one tool over another, which helps, but also confirms there is real ambiguity in selection.
Most names follow a clear snake_case action_resource pattern: start_session, get_doc, update_skill, collapse_context. The main deviations are ping as a conventional health-check name and a mix of verbs like start/finish/learn, but the convention remains predictable overall.
Eleven tools is a reasonable size for a session/memory/workflow server, covering health, sessions, context, docs, and skills without becoming oversized. However, some tools appear redundant enough that consolidation could reduce the count slightly.
The surface covers session lifecycle, context reading/writing/pruning, docs reading/writing, skills reading/writing, and health checks. There are minor gaps such as no explicit delete/remove operation for docs, skills, or context entries, but core workflows are covered.