Add gmem to Developer Tools

[yksanjo]

What

Adds gmem — a project memory MCP server for Solana AI agents — to the Developer Tools section.

Why this fits the list

The MCP slot in the current Developer Tools is well covered for reference knowledge — Solana Developer MCP queries docs, DFlow MCP exposes docs/recipes, Deside MCP handles agent messaging. What's missing is a layer for project state: when an agent works on a Solana program across sessions, it forgets the program IDs it deployed, the PDA seeds it chose, the architectural decisions made yesterday. gmem fills that slot.

It is opinionated about Solana primitives (Programs, Accounts/PDAs, Instructions, IDLs, Anchor configs, audit Findings) rather than being a generic key-value store, which is why a dedicated entry makes sense rather than pointing to mem0 or basic-memory.

What's in the repo today

• Open spec (SPEC.md) defining 4 MCP tools and 6 first-class entity kinds
• JSON Schema (draft 2020-12) per entity, in schema/
• A reference TypeScript MCP server that boots over stdio, registers all 4 tools, validates writes against the schemas, and passes an end-to-end wire test
• MIT licensed; public roadmap to v1.0 in ROADMAP.md

Checklist (per CONTRIBUTING.md)

• Related to AI and Solana development
• Added to the appropriate category (Developer Tools, MCP cluster)
• Format: [name](link) - Brief description.
• Working code, usable today; not a token, not a paid promo, not vaporware
• MIT-licensed, documented, public roadmap

[greptile-apps]

Greptile Summary

Adds a single entry for gmem, a Solana-opinionated project-memory MCP server, to the Developer Tools section of the list. The entry follows the required format and slots in after the other MCP-server entries already grouped together.

• The GitHub link points to the PR author's own public repository (yksanjo/gmem), which is typical for awesome-list submissions and not itself a concern.
• The description is notably longer than the Brief description wording in CONTRIBUTING.md suggests — it spans two sentences and enumerates entity types plus four individual tool names. The second sentence ("Ships an open spec…") describes implementation status rather than what the tool does and could be trimmed without losing the key value proposition.

Confidence Score: 5/5

Single README line addition with a correctly formatted entry and a reachable public GitHub link — no functional code is touched.

The change is a one-line addition to a markdown list. The link resolves to a real public repository, the format matches the required pattern, and placement in the Developer Tools MCP cluster is appropriate. The only observation is that the description is longer than the contributing guide's 'Brief description' wording implies, which is a minor style point and not a blocker.

No files require special attention.

Important Files Changed

Filename	Overview
README.md	Adds a single gmem entry to the Developer Tools section; follows the required [name](link) - Brief description. format with a working GitHub link.

Sequence Diagram

sequenceDiagram
    participant Agent as AI Agent
    participant gmem as gmem MCP Server
    participant Store as Local JSON Store

    Agent->>gmem: "gmem.write(entity: Program, data: {programId, ...})"
    gmem->>Store: Validate against JSON Schema (draft 2020-12)
    Store-->>gmem: OK
    gmem-->>Agent: ack

    Agent->>gmem: gmem.recall(query: "deployed program IDs")
    gmem->>Store: Lookup entities
    Store-->>gmem: [Program records]
    gmem-->>Agent: recalled context

    Agent->>gmem: gmem.diff(since: "yesterday")
    gmem->>Store: Diff entity snapshots
    Store-->>gmem: changed entities
    gmem-->>Agent: diff result

    Agent->>gmem: gmem.list_decisions()
    gmem->>Store: Fetch Decision entities
    Store-->>gmem: [architectural Decisions]
    gmem-->>Agent: decision log

Loading

_{Reviews (1): Last reviewed commit: "Add gmem to Developer Tools" | Re-trigger Greptile}

[greptile-apps]

The second sentence ("Ships an open spec…") describes implementation status/internals rather than what the tool does for the user. CONTRIBUTING.md asks for a brief description, and the other MCP entries in this section stick to a single descriptive sentence. Trimming to the core value proposition keeps the entry consistent.

Suggested change

	- [gmem](https://github.com/yksanjo/gmem) - Project memory MCP server for Solana AI agents, with first-class entities for Programs, Accounts, PDAs and seeds, Instructions, architectural Decisions, and audit Findings. Ships an open spec, JSON Schema (draft 2020-12) per entity, and a reference TypeScript server with `gmem.recall`, `gmem.write`, `gmem.diff`, and `gmem.list_decisions` tools.
	- [gmem](https://github.com/yksanjo/gmem) - Project memory MCP server for Solana AI agents with first-class entities for Programs, Accounts/PDAs, Instructions, architectural Decisions, and audit Findings, enabling persistent cross-session context for Solana program development.

Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

[yksanjo]

Thanks for the review! Tightened the description per the brief-description guideline. The single line now mentions all six first-class entity kinds plus the three new auto-ingest paths (Anchor / Solana CLI / git) that shipped in v0.3–v0.5.

The repo is now at v1.0.0 with three worked examples covering the DeFi vault, cNFT mint, and AI-agent use cases the entity model was designed for.