feat: add get_session MCP tool to resolve memory_session_id to Claude Code session

[skypper]

Summary

• Adds a new get_session MCP tool that resolves one or more memory_session_id values to their corresponding Claude Code session metadata (content_session_id, custom_title, project, started_at, status)
• Backed by the existing /api/sdk-sessions/batch endpoint — no new backend code needed
• Closes Add MCP tool to resolve memory_session_id → Claude Code session ID and name #1638

Usage

After get_observations() returns results containing memory_session_id, Claude can now call:

get_session(memory_session_ids=["d308ed77-fefb-4abf-9566-b3e7711ed420"])

And get back the Claude Code session ID and name the user can actually recognise and navigate to.

Test plan

• Call search() + get_observations() to find observations with a known memory_session_id
• Call get_session(memory_session_ids=[...]) and verify content_session_id and custom_title are returned correctly
• Test with multiple IDs in one call
• Test with an unknown memory_session_id — should return an empty array

[coderabbitai]

Summary by CodeRabbit

• New Features
  • Added a batch session retrieval tool: users can query multiple sessions at once and receive consolidated session metadata (IDs, titles, project, start time, status) as formatted JSON.

Walkthrough

Added a new MCP tool, get_session, that accepts memory_session_ids and queries the Worker /api/sdk-sessions/batch endpoint to return mapped session metadata (content_session_id, custom_title, project, started_at, status) as MCP text content.

Changes

Cohort / File(s)	Summary
New MCP tool src/servers/mcp-server.ts	Registered get_session tool with input schema memory_session_ids: string[] (required, no additional props). Handler POSTs { memorySessionIds: memory_session_ids } to /api/sdk-sessions/batch, validates response, maps session fields into a metadata array, and returns it as a single MCP text content item.

Sequence Diagram(s)

sequenceDiagram
    participant Claude as Claude (MCP Client)
    participant MCP as MCP Server (get_session tool)
    participant Worker as Worker API (/api/sdk-sessions/batch)

    Claude->>MCP: call get_session(memory_session_ids)
    MCP->>Worker: POST /api/sdk-sessions/batch\n{ memorySessionIds: [...] }
    Worker-->>MCP: 200 OK with SDK sessions JSON array
    MCP->>MCP: map fields -> metadata array\n(content_session_id, custom_title, project, started_at, status)
    MCP-->>Claude: return MCP `text` with JSON.stringify(metadata)

Loading

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~8 minutes

Poem

🐰 Hop, hop — a lookup fresh and neat,
I fetch the sessions, link each heartbeat.
From memory IDs to titles bright,
I hop the path and shed some light,
Trail to sessions found — a rabbit's delight.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately summarizes the main change: adding a new MCP tool named get_session that resolves memory_session_id to Claude Code session information.
Description check	✅ Passed	The description is directly related to the changeset, explaining the purpose of the new get_session MCP tool and providing usage examples and test plans.
Linked Issues check	✅ Passed	The PR fully implements the requirement from issue #1638: a new MCP tool that resolves memory_session_id to Claude Code session metadata (content_session_id, custom_title, project, started_at, status) via the existing /api/sdk-sessions/batch endpoint.
Out of Scope Changes check	✅ Passed	All changes are in scope: only adds the get_session MCP tool handler and schema to mcp-server.ts as specified in #1638, with no unrelated modifications.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@src/servers/mcp-server.ts`:
- Around line 233-248: The handler currently returns the raw batch response from
callWorkerAPIPost('/api/sdk-sessions/batch', { memorySessionIds:
args.memory_session_ids }) which can include prompt content (e.g., user_prompt)
from SessionStore; change the handler to post-process the worker response and
return only metadata fields—filter out user_prompt and any prompt/text fields
and map each item to a safe metadata object (e.g., memory_session_id,
claude_session_id, name, created_at or other non-content columns) before
returning to the caller so the tool adheres to a metadata-only contract.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 0a3d520a-a2ad-4acf-a005-5b043ea17017

📥 Commits

Reviewing files that changed from the base of the PR and between 9f01228 and cbef129.

📒 Files selected for processing (2)

• plugin/scripts/mcp-server.cjs
• src/servers/mcp-server.ts

[coderabbitai]

🧹 Nitpick comments (1)

src/servers/mcp-server.ts (1)

237-244: Consider enforcing at least one memory_session_id in schema.

The PR objective says “one or more” IDs; adding minItems: 1 prevents empty no-op calls and tightens contract validation.

Suggested schema tweak

         memory_session_ids: {
           type: 'array',
           items: { type: 'string' },
+          minItems: 1,
           description: 'Array of memory_session_id values from get_observations results'
         }

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@src/servers/mcp-server.ts` around lines 237 - 244, The JSON schema for the
property memory_session_ids currently allows an empty array even though the PR
intends “one or more” IDs; update the schema for memory_session_ids (the object
with type: 'array' and items: { type: 'string' }) to enforce at least one
element by adding minItems: 1 so required: ['memory_session_ids'] combined with
additionalProperties: false prevents empty/no-op calls.

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Nitpick comments:
In `@src/servers/mcp-server.ts`:
- Around line 237-244: The JSON schema for the property memory_session_ids
currently allows an empty array even though the PR intends “one or more” IDs;
update the schema for memory_session_ids (the object with type: 'array' and
items: { type: 'string' }) to enforce at least one element by adding minItems: 1
so required: ['memory_session_ids'] combined with additionalProperties: false
prevents empty/no-op calls.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: e5393ee1-431c-4fc2-b16d-0b5a91aa4033

📥 Commits

Reviewing files that changed from the base of the PR and between cbef129 and 7c35d92.

📒 Files selected for processing (2)

• plugin/scripts/mcp-server.cjs
• src/servers/mcp-server.ts

[skypper]

@thedotmack hey, would love a review when you get a chance! 👋

[thedotmack]

Closed during the April 2026 backlog cleanup. The feature ask is now tracked in #1971, which is the single canonical issue for prioritization. Thanks for the suggestion — your context is captured there.