What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
AI agents may call MCP tools/list repeatedly because discovering available tools is a separate step from invoking them, and a client cannot safely assume that the list is unchanged. SEP-2549 gives list responses optional freshness and sharing metadata: ttlMs and cacheScope. That lets clients and gateways avoid needless rediscovery when reuse is safe, without treating a cached list as permanent or sharing user-specific tools across users.
The proposal is marked final and is included in the MCP specification revision dated July 28, 2026. Support still depends on the protocol revision and software versions deployed by a particular server, client, or gateway.
As an Amazon Associate I earn from qualifying purchases.
What SEP-2549 changes
SEP-2549, “TTL for List Results,” adds two optional fields to results from tools/list, prompts/list, resources/list, resources/read, and resources/templates/list: ttlMs and cacheScope. The server returns these alongside the current result; a client or intermediary then decides whether it may retain and reuse that exact response.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minutettlMsis a freshness estimate, expressed in milliseconds from when the response is received. A positive value is the period the client should treat the response as fresh; zero means it is immediately stale.cacheScopeindicates whether reuse may cross user boundaries.publicpermits shared-cache reuse only when the response is not user-specific.privateconfines caching to the requesting user’s client; a shared cache must not serve that response to another user.
As the proposal puts it, “A TTL is a freshness estimate, not a guarantee.” The underlying data can change before the TTL expires.
#1 Best Overall
- More for the money with this high quality Product
- Offers premium quality at outstanding saving
- Excellent product
- 100% satisfaction
How long should a client cache tools/list?
SEP-2549 does not prescribe one universal duration. The server should provide a TTL that reflects its estimate of how long the returned data is likely to remain fresh; clients should honor it as guidance rather than as proof that nothing changed. A short TTL reduces the time a stale definition might be reused but leads to more rediscovery. A longer TTL can reduce repeated list requests, but increases the window in which a changed tool definition may go unnoticed.
| Choice | Benefit | Trade-off |
|---|---|---|
| Shorter TTL | Limits time relying on a possibly outdated result. | More frequent refetches. |
| Longer TTL | Fewer refetches while the result remains stable. | Greater chance that a change is not seen until expiry or invalidation. |
Choose based on how often the authorized tool set changes and the cost of acting on a stale description. The proposal does not establish a performance target or a numeric reduction in requests, latency, or tokens.
Can a cached tool list be shared between users?
Only when the result is genuinely independent of the user and its scope is public. If available tools depend on the authorization presented with the request, the list is user-dependent and must not be served to a different user through a shared cache. Use private for that case.
Cache identity must preserve every factor that determines the response, including the authorization context. A server-provided scope is not a substitute for correct cache keys, and labelling a response public cannot make a user-specific result safe to share. MCP’s tools documentation notes that tools/list may reflect request authorization and recommends deterministic ordering when the underlying set is unchanged.
Rank #3
- Product type: Screw kit
- Made by Super Micro
- Manufacturer part number: MCP-410-00005-0N
- Supermicro MCP-410-00005-0N Screw Bag(100PCS) and Label for 24x Hot swap
- Mfr Part Number: MCP-410-00005-0N
Deterministic ordering makes equivalent lists serialize consistently, which can help reuse cached responses and improve LLM prompt-cache hits when tool definitions are included in context. It does not establish that two users are authorized for the same set.
Use notifications as well as TTL
Time-based expiry is not the only freshness mechanism. A server that advertises list-change notifications should send the corresponding notification when its data changes before expiry. Clients may also refetch early when they have reason to believe a cached result is stale—for example, when a tool call fails in a way that suggests the cached definition is obsolete.
SEP-2549 states: “TTL supplements rather than replaces the existing notification mechanism — both can coexist.” In practice, expiry provides a fallback freshness bound while notifications allow faster reaction to known changes.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Handle pagination without assuming a snapshot
Cache paginated results page by page. Each page has its own TTL and freshness clock; the proposal does not guarantee that separate pages form one coherent snapshot. A client that needs a newly consistent view should restart from the beginning rather than combine pages fetched at different times.
- Track freshness for each page independently.
- If a cursor is rejected or invalidated, discard the cached pages and fetch again from the start.
- If polling for updates, use jitter and backoff to avoid synchronized request bursts.
Separate protocol metadata from SDK cache behavior
SEP-2549 describes server-provided freshness and scope metadata. An SDK’s in-memory cache is a distinct client implementation choice. For example, the OpenAI Agents SDK documentation says an agent may call list_tools() on each run and offers optional in-memory caching with explicit invalidation. Its guidance is to enable that option when tool definitions are unlikely to change. That opt-in behavior is not a protocol-wide default and does not replace handling server TTLs, notifications, authorization, or pagination.
Stale fallback when refresh fails
If a refresh fails because of a network or server error, SEP-2549 permits a client to serve a stale response as a resilience fallback. This trades freshness for availability: the cached definitions may no longer match the current tools. Treat it as an explicit fallback policy, not the normal rule for deciding whether an entry is fresh.
Quick Recap
Implementation checklist
- Return the authorized result. Build each list from the authorization context in the request, then include the applicable
ttlMsandcacheScope. - Choose scope conservatively. Mark a result
publiconly if it is truly the same across users; otherwise useprivateand prevent shared caches from crossing user boundaries. - Keep stable ordering. Return a deterministic order when the underlying set is unchanged.
- Honor both freshness paths. Expire entries according to their TTL and invalidate promptly on applicable change notifications or other evidence of staleness.
- Cache pages independently. Handle invalid cursors by discarding the page set and restarting from the beginning.
- Define failure behavior. Decide whether to use stale data when refresh fails, and make the freshness-versus-availability trade-off explicit.
- Verify deployed support. Confirm the protocol revision and SDK versions used by the actual server, client, and intermediary; final status in the specification does not by itself establish universal deployment.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
