fix(slots): wire pinned slot injection into mem::context

[wyh0626]

Summary

Closes #286. renderPinnedContext and listPinnedSlots were introduced in #182 and tested in test/slots.test.ts, but no code path actually invokes them. mem::context — the function /agentmemory/session/start calls and that session-start.mjs reads back into Claude Code — reads profile / sessions / observations but never reads slots, so pinned content stays in KV and never reaches the conversation.

#182's description hedged with "pinned slots can be injected into SessionStart context" and "pinned slots are candidates for SessionStart injection". The helpers and unit tests landed; the wiring didn't.

Fix

Add the listPinnedSlots → renderPinnedContext call at the top of mem::context, gated by isSlotsEnabled():

if (isSlotsEnabled()) {
  const slotContent = renderPinnedContext(
    await listPinnedSlots(kv).catch(() => []),
  );
  if (slotContent) {
    blocks.push({
      type: "memory",
      content: slotContent,
      tokens: estimateTokens(slotContent),
      recency: Date.now(),
    });
  }
}

recency: Date.now() so the pinned block sorts to the top of the budget-bounded selection. Mirrors how isGraphExtractionEnabled() guards mem::graph-extract after #215.

AGENTMEMORY_SLOTS=false (the default) and AGENTMEMORY_INJECT_CONTEXT=false (also the default) both keep the existing behaviour untouched — the gate composition matches the established opt-in convention (#138, #143, #160, #179).

On cross-project visibility

This PR uses the same listPinnedSlots(kv) reader as the existing mem::slot-list / memory_slot_list / memory_slot_get APIs and the mem::slot-reflect writer. Project-scoped slot storage (KV.slots = "mem:slots") is keyed by label without a project prefix today, so cross-project visibility already exists via the slot list/get tools and via reflect. This PR doesn't change the underlying isolation model; if tighter per-project keying is wanted, that's an orthogonal concern in the slot storage layer.

Verify

# server: AGENTMEMORY_SLOTS=true AGENTMEMORY_INJECT_CONTEXT=true
curl -X POST localhost:3111/agentmemory/slot/replace \
  -H 'Content-Type: application/json' \
  -d '{"label":"tool_guidelines","content":"rule-alpha"}'

curl -X POST localhost:3111/agentmemory/session/start \
  -H 'Content-Type: application/json' \
  -d '{"sessionId":"s1","project":"/tmp","cwd":"/tmp"}'

Before: "context": "" (0 chars).
After: context contains the tool_guidelines slot wrapped in <agentmemory-context>.

Tests

6 cases in test/context-slots.test.ts:

• pinned global slot lands in returned context
• multiple pinned slots render in label-sorted order
• unpinned slots are excluded even with content
• empty pinned slots (the seeded defaults) are excluded
• project-scoped slot shadows global slot with the same label
• AGENTMEMORY_SLOTS=off path emits nothing

npm test for the file: 6/6 pass. Pre-existing failures in test/mcp-standalone.test.ts and test/fs-watcher.test.ts are unaffected by this change.

Summary by CodeRabbit

Release Notes

• New Features

  • Support for pinned slots in context memory: when enabled, pinned slot content is injected into memory blocks, sorted by label, with project-scoped slots overriding global ones. If disabled, no pinned content is included.

• Tests

  • Added tests verifying pinned slot injection behavior across configurations and precedence rules.

[vercel]

@wyh0626 is attempting to deploy a commit to the rohitg00's projects Team on Vercel.

A member of the Team first needs to authorize it.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 4c88aba6-62d2-47a0-8eb2-b0ed1d352aba

📥 Commits

Reviewing files that changed from the base of the PR and between 0572cbb and b22e256.

📒 Files selected for processing (1)

• src/functions/context.ts

🚧 Files skipped from review as they are similar to previous changes (1)

• src/functions/context.ts

---

📝 Walkthrough

Walkthrough

Conditionally fetches and renders pinned slot content into the mem::context output when AGENTMEMORY_SLOTS is enabled; adds a memory block with token count and recency only if rendered content is non-empty. Tests verify enabled/disabled behavior and precedence between project and global slots.

Changes

Pinned Slots Context Integration

Layer / File(s)	Summary
Slot Feature Imports src/functions/context.ts	Adds imports for isSlotsEnabled, listPinnedSlots, and renderPinnedContext to enable conditional pinned-slot context generation.
Pinned Slots Context Generation src/functions/context.ts	mem::context handler now checks if slots are enabled, fetches pinned slot data from KV, renders it as a single context block, and adds it to memory blocks with token estimation and current timestamp if content is non-empty.
Test Infrastructure and Mocking test/context-slots.test.ts	Establishes in-memory KV mock and registers mem::context handler for isolated testing of pinned slot injection behavior.
Test Fixture Helpers test/context-slots.test.ts	Provides seedPinnedSlot helper function to seed consistent pinned KV slot records into global or project-scoped storage with standard metadata (pinned/readOnly flags, timestamps, limits).
Pinned Slots Enabled Tests test/context-slots.test.ts	Verifies that when AGENTMEMORY_SLOTS=true, context includes pinned global slot content sorted by label, excludes unpinned and empty slots, and applies project-scoped slot precedence over global slots when labels collide.
Pinned Slots Disabled Tests test/context-slots.test.ts	Confirms backward compatibility: when AGENTMEMORY_SLOTS is absent or false, context returns without any injected slot content.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~12 minutes

Possibly related issues

• Pinned slot content never reaches SessionStart context — renderPinnedContext has zero callers #286: Directly addresses the unused-callers problem by adding calls to listPinnedSlots and renderPinnedContext inside src/functions/context.ts.

Possibly related PRs

• rohitg00/agentmemory#182: Introduced the slots subsystem functions (isSlotsEnabled, listPinnedSlots, renderPinnedContext) that this PR integrates into mem::context.

Poem

I nibble through code, a rabbit so spry,
Pinned memories gathered where tokens lie.
When the flag blooms, I stitch them in place,
Project-first whispers, global finds their space.
Hooray for context — hopping with grace. 🐇

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title 'fix(slots): wire pinned slot injection into mem::context' clearly summarizes the main change: integrating slot injection into the mem::context handler. It is concise, specific, and directly reflects the core modification across both the implementation and test files.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

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

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

Tip

💬 Introducing Slack Agent: The best way for teams to turn conversations into code.

Slack Agent is built on CodeRabbit's deep understanding of your code, so your team can collaborate across the entire SDLC without losing context.

• Generate code and open pull requests
• Plan features and break down work
• Investigate incidents and troubleshoot customer tickets together
• Automate recurring tasks and respond to alerts with triggers
• Summarize progress and report instantly

Built for teams:

• Shared memory across your entire org—no repeating context
• Per-thread sandboxes to safely plan and execute work
• Governance built-in—scoped access, auditability, and budget controls

One agent for your entire SDLC. Right inside Slack.

👉 Get started

---

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]

🧹 Nitpick comments (1)

src/functions/context.ts (1)

41-52: ⚡ Quick win

Consider parallelizing independent KV reads.

The listPinnedSlots call and the profile fetch (line 54-56) are independent operations reading from different KV scopes. Per the coding guideline, these could be executed in parallel using Promise.all to reduce latency.

⚡ Proposed refactor to parallelize KV reads

-    if (isSlotsEnabled()) {
-      const pinned = await listPinnedSlots(kv).catch(() => []);
-      const slotContent = renderPinnedContext(pinned);
-      if (slotContent) {
-        blocks.push({
-          type: "memory",
-          content: slotContent,
-          tokens: estimateTokens(slotContent),
-          recency: Date.now(),
-        });
-      }
-    }
-
-    const profile = await kv
-      .get<ProjectProfile>(KV.profiles, data.project)
-      .catch(() => null);
+    const [pinnedSlots, profile] = await Promise.all([
+      isSlotsEnabled() 
+        ? listPinnedSlots(kv).catch(() => [])
+        : Promise.resolve([]),
+      kv.get<ProjectProfile>(KV.profiles, data.project).catch(() => null),
+    ]);
+
+    if (pinnedSlots.length > 0) {
+      const slotContent = renderPinnedContext(pinnedSlots);
+      if (slotContent) {
+        blocks.push({
+          type: "memory",
+          content: slotContent,
+          tokens: estimateTokens(slotContent),
+          recency: Date.now(),
+        });
+      }
+    }
+
     if (profile) {

As per coding guidelines: "Use parallel operations where possible with Promise.all for independent kv writes/reads" for files in src/functions/**/*.ts.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@src/functions/context.ts` around lines 41 - 52, The two independent KV reads
(listPinnedSlots and the profile fetch used just after this block) should run in
parallel using Promise.all to reduce latency: call
Promise.all([listPinnedSlots(kv), fetchProfile(kvOrScope)]) (replace
fetchProfile with the actual profile-fetch function used in the file), then
destructure the results into pinned and profile; keep the existing fallback for
pinned (empty array) and handle a missing profile as before, then call
renderPinnedContext(pinned) and push the memory block if slotContent exists;
ensure you preserve estimateTokens(slotContent) and recency timestamp and
maintain any existing error handling for each read (e.g., map the pinned promise
to .catch(() => []) if you want that specific fallback).

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Nitpick comments:
In `@src/functions/context.ts`:
- Around line 41-52: The two independent KV reads (listPinnedSlots and the
profile fetch used just after this block) should run in parallel using
Promise.all to reduce latency: call Promise.all([listPinnedSlots(kv),
fetchProfile(kvOrScope)]) (replace fetchProfile with the actual profile-fetch
function used in the file), then destructure the results into pinned and
profile; keep the existing fallback for pinned (empty array) and handle a
missing profile as before, then call renderPinnedContext(pinned) and push the
memory block if slotContent exists; ensure you preserve
estimateTokens(slotContent) and recency timestamp and maintain any existing
error handling for each read (e.g., map the pinned promise to .catch(() => [])
if you want that specific fallback).

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 3c97c886-7838-44a3-81bd-989a6f59df9c

📥 Commits

Reviewing files that changed from the base of the PR and between ec84ff8 and 0572cbb.

📒 Files selected for processing (2)

• src/functions/context.ts
• test/context-slots.test.ts

[rohitg00]

Thanks @wyh0626