Hacktoberfest 2026: le issue che i maintainer hanno segnato per ottobre, aperte e adatte ai principianti. Sfoglia le issue Hacktoberfest

`sentient mcp`: let Claude Code and other agents use Sentient's memory, tasks and tools over MCP

Aperta
  • Qualcosa non va?
#443 0 commenti 0 reazioni 0 assegnatari Vedi su GitHub

I maintainer di solito rispondono entro 1 giorno

Nessuno ha ancora preso questa issue.

Valutazione

Difficoltà
5/5
Tempo stimato
Più di una settimana
Idoneità per principianti
35/100
Tipo di issue
Funzionalità
Chiarezza
Abbastanza chiara
Stato di attività
Attiva
Stack tecnologico
python
Ambito
backend-api-design, cli, security

Direzione di ricerca

Start with sentient/cli.py, packaging/engine_entry.py and the existing ADRs 0016 and 0018; the issue also names a new sentient/mcp_server/, Settings → Integrations and docs/API.md. Add an ADR and check the test enforcing _COMMANDS. Done means Claude Code can search memories and create a task for approval, while risky requests are refused or ask on the desktop, verified with a real Claude Code test.

Scritto dal modello di indicizzazione a partire dal testo della issue.

Descrizione

area: engine area: integrations enhancement help wanted priority: P2 size: M

Sentient can use MCP servers, but other agents can't use Sentient. People who live in Claude Code, Cursor or another MCP client want their assistant's memory and tasks there too, the way Hermes and OpenClaw expose theirs.

Idea: a sentient mcp command (stdio, plus optionally streamable HTTP on localhost) that exposes a small, safe set of tools backed by the running engine's REST API and token:

  • memory_search, memory_add (lands in Review unless the user allows otherwise), memory_get;
  • task_create (goes through the normal plan-and-approve flow), task_list, task_status;
  • notify_me (a Sentient notification);
  • optionally ask_sentient, one chat turn whose tools run through Sentient's own approvals.

Safety, non-negotiable:

  • Every action still goes through Sentient's approvals, rules (ADR 0016) and outside-content checks (ADR 0018). The MCP client is outside content, and anything it asks Sentient to send or delete asks the user on the desktop.
  • There is no direct send/delete/exec tool.
  • It only works on this computer with the engine's token, and it is off until the user turns it on in Settings, which shows the config snippet to paste into Claude Code (claude mcp add sentient -- sentient mcp) or other clients.
  • Add an ADR.

Where: sentient/cli.py, a new sentient/mcp_server/, Settings → Integrations, docs/API.md. Also add mcp to packaging/engine_entry.py's _COMMANDS, which a test enforces.

Done when: Claude Code with sentient added can search Maya's memories, create a task that then shows up in Sentient for approval, and is refused or asks on the desktop for anything risky. Real test with Claude Code itself. Related: #211 (OpenAI-compatible API).

Lingua principale
Python
Stelle
702
Fork
113
Merge medio
5h 8m
PR unite (30g)
171

Preparare l'ambiente

Come iniziare

  1. Leggi tutta la issue e poi la guida ai contributi del progetto.
  2. Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
  3. Fai un fork del repository e lavora su un branch.
  4. Apri una pull request che faccia riferimento al numero della issue.

Altre issue di existence-master/Sentient

Tutte le issue di existence-master/Sentient

Issue simili

Altre issue su Python

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.