feat(p2p): additional commitments

content/p2p/ift/2026q2-nimlibp2p-kad-dht-optimizations.md

@@ -0,0 +1,77 @@
+---
+
+title: nim-libp2p Kad-DHT Optimizations
+tags:
+  - "2026q2"
+  - "p2p"
+  - "ift"
+draft: false
+description: Improve Kad-DHT privacy and provider record advertisement behavior
+
+---
+
+`ift-ts:p2p:ift:2026q2-nimlibp2p-kad-dht-optimizations`
+
+Improve nim-libp2p Kad-DHT privacy and provider record advertisement behavior.
+
+## Description
+
+This commitment covers two related Kad-DHT improvements. First, Kad-DHT response
+messages should avoid disclosing peer connection type because it can reveal
+network topology and increase eclipse-attack risk. This follows
+[vacp2p/nim-libp2p#2050](https://github.com/vacp2p/nim-libp2p/issues/2050)
+and the Logos extended Kad discovery privacy requirement that every Kad-DHT
+`Peer.connection` field be encoded as `NOT_CONNECTED`.
+
+Second, provider record advertisement traffic can overload nodes close to
+popular content. The libp2p community has discussed Coral-style rejection,
+spillover, and backtracking for provider records in
+[libp2p/specs#163](https://github.com/libp2p/specs/issues/163), and
+[go-libp2p-kad-dht#345](https://github.com/libp2p/go-libp2p-kad-dht/issues/345)
+tracks these as part of the critical path toward DHT efficiency and
+performance. This commitment brings equivalent optimization work to the
+nim-libp2p Kad-DHT implementation.
+
+## Task List
+
+### Kad-DHT Connection Field Privacy
+
+* fully qualified name: `ift-ts:p2p:ift:2026q2-nimlibp2p-kad-dht-optimizations:connection-field-privacy`
+* owner: gabe
+* status: not started
+* start-date: 2026/05/01
+* end-date: 2026/06/30
+
+#### Description
+Ensure Kad-DHT response messages can optionally not disclose local connection information.
+When encoding Kad-DHT `Peer` records, set the `connection` field to
+`NOT_CONNECTED`. Maintain backward compatibility by continuing to parse peers
+that send other `ConnectionType` values.
+
+#### Deliverables
+- Kad-DHT encoder always emits `NOT_CONNECTED` for peer connection type depending on kademlia setup option
+- Receive-side compatibility for non-compliant peers
+- Tests covering encoding and parsing of `ConnectionType` values
+- Documentation of the privacy behavior and compatibility policy
+
+
+### Provider Record Spillover
+
+* fully qualified name: `ift-ts:p2p:ift:2026q2-nimlibp2p-kad-dht-optimizations:provider-record-spillover`
+* owner: gabe
+* status: not started
+* start-date: 2026/04/01
+* end-date: 2026/06/30
+
+#### Description
+Add provider record overload protection and advertisement spillover behavior.
+Provider records should have local limits and throttling, and rejected
+advertisements should be able to spill over by backtracking through the lookup
+path rather than repeatedly targeting only the closest nodes to a popular key.
+
+#### Deliverables
+- Configurable provider record limits and throttling policy
+- Rejection response handling for provider advertisements
+- Spillover/backtracking behavior for popular provider records
+- Metrics or simulation coverage showing advertisement load distribution
+- Tests for provider record limits, rejection, spillover, and lookup behavior

content/p2p/ift/2026q2-nimlibp2p-maintenance.md

@@ -262,3 +262,48 @@ Some related work available is:
 - Agreed technical direction for `logos-delivery#3634`
 - Ownership decision between nim-libp2p and the relevant dependent components
 - Follow-up issue or PR breakdown, including Chronos work if needed
+
+
+### Large RSA key limit
+
+* fully qualified name: `ift-ts:p2p:ift:2026q2-nimlibp2p-maintenance:large-rsa-key-limit`
+* owner: not assigned yet
+* status: not started
+* start-date: 2026/05/01
+* end-date: 2026/06/30
+
+#### Description
+Fix [vacp2p/nim-libp2p#1327](https://github.com/vacp2p/nim-libp2p/issues/1327)
+by rejecting RSA keys that are too large to parse or verify safely. This should
+protect nodes from resource exhaustion attacks where a malicious peer sends an
+oversized RSA key and forces expensive signature verification. Use the
+go-libp2p fix in
+[libp2p/go-libp2p#2454](https://github.com/libp2p/go-libp2p/pull/2454) as
+guidance, including the `<= 8192` bit maximum.
+
+#### Deliverables
+- Maximum RSA key size enforced during key parsing and generation
+- Tests for accepted RSA keys and rejected oversized RSA keys
+- Documentation or changelog note describing the compatibility/security behavior
+
+
+### Sink performance exploration
+
+* fully qualified name: `ift-ts:p2p:ift:2026q2-nimlibp2p-maintenance:sink-performance-exploration`
+* owner: richard
+* status: not started
+* start-date: 2026/05/01
+* end-date: 2026/06/30
+
+#### Description
+Explore using Nim `sink` parameters in internal code paths where ownership
+transfer can reduce unnecessary copies and improve memory performance. This is
+tracked in
+[vacp2p/nim-libp2p#1641](https://github.com/vacp2p/nim-libp2p/issues/1641).
+Maintain backward compatibility for public APIs.
+
+#### Deliverables
+- Benchmarked list of candidate internal hot paths
+- PRs applying `sink` where the ownership semantics are clear and beneficial
+- Tests or benchmarks showing no behavioral regression
+- No public API break.

content/p2p/ift/2026q2-nimlibp2p-nat-traversal.md

@@ -0,0 +1,69 @@
+---
+
+title: nim-libp2p NAT Traversal
+tags:
+  - "2026q2"
+  - "p2p"
+  - "ift"
+draft: false
+description: Add reusable NAT traversal address setup to nim-libp2p
+
+---
+
+`ift-ts:p2p:ift:2026q2-nimlibp2p-nat-traversal`
+
+Add reusable NAT traversal and public address setup support to nim-libp2p.
+
+## Description
+
+nim-libp2p has long tracked NAT traversal support in
+[vacp2p/nim-libp2p#51](https://github.com/vacp2p/nim-libp2p/issues/51).
+Downstream users already carry NAT address setup logic, including a configuration
+option with `any`, `none`, `upnp`, `pmp`, and `extip:<IP>` modes,
+and Nimbus code that calls `setupAddress(config.nat, ...)` for TCP and UDP
+ports. This commitment extracts the reusable behavior into nim-libp2p so
+applications can configure public address discovery and port mapping through
+the library instead of duplicating efforts downstream.
+
+## Task List
+
+### NAT Address Setup API
+
+* fully qualified name: `ift-ts:p2p:ift:2026q2-nimlibp2p-nat-traversal:address-setup-api`
+* owner: not assigned yet
+* status: not started
+* start-date: 2026/05/01
+* end-date: 2026/06/30
+
+#### Description
+Provide a reusable nim-libp2p API and builder/config integration for NAT public
+address setup. The API should support the downstream modes already used by
+Logos and Nimbus: automatic discovery, disabled NAT setup, UPnP, NAT-PMP, and
+explicit external IP configuration.
+
+#### Deliverables
+- Builder/config option for NAT address setup
+- Support for `any`, `none`, `upnp`, `pmp`, and `extip:<IP>` modes or equivalent typed configuration
+- Public address and mapped port results exposed to transport/listen address handling
+- Tests for mode parsing, explicit external IP handling, and disabled NAT setup
+
+
+### UPnP and NAT-PMP Integration
+
+* fully qualified name: `ift-ts:p2p:ift:2026q2-nimlibp2p-nat-traversal:upnp-pmp`
+* owner: not assigned yet
+* status: not started
+* start-date: 2026/04/01
+* end-date: 2026/06/30
+
+#### Description
+Integrate UPnP and NAT-PMP address mapping through existing Nim NAT traversal
+libraries where possible, reusing the behavior already proven by downstream
+projects. The implementation should work for nim-libp2p TCP usage and be
+structured so UDP-capable transports can use the same setup path.
+
+#### Deliverables
+- UPnP and NAT-PMP mapping support available from nim-libp2p
+- Address propagation into identify/listen address reporting where applicable
+- Downstream migration guidance for Logos and Nimbus NAT configuration
+- Tests or examples covering mapped TCP address advertisement

content/p2p/ift/2026q2-nimlibp2p-stream-budgets.md

@@ -0,0 +1,118 @@
+---
+
+title: nim-libp2p Stream Budgets
+tags:
+  - "2026q2"
+  - "p2p"
+  - "ift"
+draft: false
+description: Per-protocol stream budgets in nim-libp2p
+
+---
+
+`ift-ts:p2p:ift:2026q2-nimlibp2p-stream-budgets`
+
+Add opt-in per-protocol stream budgets to nim-libp2p so stream-count abuse can be rejected before a protocol handler is overloaded.
+
+## Description
+
+nim-libp2p currently exposes `LPProtocol.maxIncomingStreams` as an inbound
+per-peer limit for a protocol, but it does not provide a complete budget model
+for total inbound streams, outbound streams, or protocol-wide stream pressure.
+This leaves protocol handlers vulnerable to overload when one peer or many
+peers consume too much stream capacity.
+
+This commitment extends the existing `LPProtocol.maxIncomingStreams` model into
+a small per-protocol stream budget matrix. The work should keep the V1 scope
+narrow: it limits stream counts only, does not attempt memory accounting, idle
+time tracking, or peer reputation, and keeps new limits opt-in unless a separate
+compatibility decision changes existing defaults.
+
+## Task List
+
+### Budget API And Accounting
+
+* fully qualified name: `ift-ts:p2p:ift:2026q2-nimlibp2p-stream-budgets:budget-api`
+* owner: not assigned yet
+* status: not started
+* start-date: 2026/05/01
+* end-date: 2026/06/30
+
+#### Description
+Extend `LPProtocol` with stream budget settings for inbound total streams,
+outbound per-peer streams, and outbound total streams, while preserving
+`maxIncomingStreams` as the existing inbound per-peer cap. Track open stream
+counters per mounted protocol handler, direction, and peer where per-peer caps
+need it. If one protocol handler exposes multiple compatible codec strings, all
+compatible codec strings should share the same budget.
+
+#### Deliverables
+- `maxTotalIncomingStreams`, `maxOutgoingStreams`, and `maxTotalOutgoingStreams`
+  added to `LPProtocol`
+- Existing `maxIncomingStreams` behavior preserved as the inbound per-peer cap
+- Open stream accounting scoped by mounted protocol handler and direction
+- Shared budget behavior for compatible codec strings mounted by the same protocol handler
+
+
+### Inbound Stream Enforcement
+
+* fully qualified name: `ift-ts:p2p:ift:2026q2-nimlibp2p-stream-budgets:inbound-enforcement`
+* owner: not assigned yet
+* status: not started
+* start-date: 2026/05/01
+* end-date: 2026/06/30
+
+#### Description
+Enforce inbound per-peer and total protocol caps in the multistream accept path
+after protocol negotiation, next to the existing `maxIncomingStreams` check. On
+breach, reject or close the new stream while keeping the peer connection open.
+Counters must be decremented on clean close, error close, connection drop,
+handler crash, and muxer close paths.
+
+#### Deliverables
+- Inbound per-peer stream limit enforced through the existing `maxIncomingStreams` path
+- Inbound total stream limit enforced per mounted protocol handler
+- Rejection behavior that closes only the new stream and keeps the connection up
+- Tests covering clean close, error close, connection drop, handler crash, and muxer close
+
+
+### Outbound Stream Enforcement
+
+* fully qualified name: `ift-ts:p2p:ift:2026q2-nimlibp2p-stream-budgets:outbound-enforcement`
+* owner: not assigned yet
+* status: not started
+* start-date: 2026/05/01
+* end-date: 2026/06/30
+
+#### Description
+Enforce outbound per-peer and total protocol caps after outbound protocol
+negotiation, because the muxer and connection manager stream-open path does not
+know which protocol will be selected before negotiation completes. Failed
+negotiation, cancellation, stream close or reset, and muxer close must roll back
+any counter reservation.
+
+#### Deliverables
+- Outbound per-peer stream limit enforced after protocol negotiation
+- Outbound total stream limit enforced per mounted protocol handler
+- Counter rollback for negotiation failure, cancellation, stream close/reset, and muxer close
+- Tests covering outbound negotiation failure and outbound cancellation
+
+
+### Metrics And Validation
+
+* fully qualified name: `ift-ts:p2p:ift:2026q2-nimlibp2p-stream-budgets:metrics-validation`
+* owner: not assigned yet
+* status: not started
+* start-date: 2026/05/01
+* end-date: 2026/06/30
+
+#### Description
+Expose protocol stream budget metrics and validate the feature against overload
+and lifecycle scenarios. Metrics should include protocol, direction, and scope
+labels where useful, but should not add a per-peer label.
+
+#### Deliverables
+- `libp2p_protocol_stream_cap_rejections_total{protocol, direction, scope}` metric
+- `libp2p_protocol_streams_open{protocol, direction}` metric
+- Integration tests for inbound and outbound cap breaches
+- Documentation describing the opt-in V1 behavior and compatibility expectations

content/p2p/ift/2026q2-nimlibp2p-stream-connection-refactor.md

@@ -0,0 +1,55 @@
+---
+
+title: nim-libp2p Stream and Connection Refactor
+tags:
+  - "2026q2"
+  - "p2p"
+  - "ift"
+draft: false
+description: Refactor nim-libp2p stream and connection abstractions
+
+---
+
+`ift-ts:p2p:ift:2026q2-nimlibp2p-stream-connection-refactor`
+
+Refactor nim-libp2p stream and connection abstractions so peer connections and protocol streams are represented by distinct types.
+
+## Description
+
+The current nim-libp2p implementation uses `Connection` both for actual peer
+connections and for streams opened over multiplexed connections. In the current
+code shape `Connection` inherits from `LPStream`, muxers return `Connection`
+from `newStream`, and protocol handlers receive `Connection` even when they are
+handling a single protocol stream. This makes the abstraction harder to
+understand, increases the cost of onboarding contributors, and complicates
+resource lifecycle work such as close/reset behavior and stream ownership.
+
+This commitment separates the concepts so `Connection` refers to peer-level
+connections and a new stream abstraction is used for protocol streams. The
+refactor should preserve downstream compatibility during the migration and
+reduce ambiguity in code, tests, metrics, and documentation.
+
+## Task List
+
+### Stream and Connection Abstractions
+
+* fully qualified name: `ift-ts:p2p:ift:2026q2-nimlibp2p-stream-connection-refactor:abstractions`
+* owner: not assigned yet
+* status: not started
+* start-date: 2026/05/01
+* end-date: 2026/06/30
+
+#### Description
+Introduce a clear stream abstraction for protocol streams and narrow
+`Connection` to represent actual peer connections. Update muxer and connection
+manager interfaces so opening a stream returns the stream type, while peer-level
+connection state remains on connection objects.
+
+
+#### Deliverables
+- New stream-level abstraction used by muxers and protocol handlers
+- `Connection` narrowed to peer-level connection state and behavior
+- Compatibility shims or migration aliases for existing downstream users
+- Updated close/reset lifecycle behavior for streams and connections
+- Tests covering stream creation, close/reset semantics, and protocol handler usage
+- Documentation and examples using the new names consistently

content/p2p/index.md

@@ -21,8 +21,12 @@ Major future work items on nim-libp2p (also beyond this reporting period are cap
 * [[p2p/ift/2026q2-nimlibp2p-cbindings|nimlibp2p-cbindings]]
 * [[p2p/ift/2026q2-nimlibp2p-connection-management|nimlibp2p-connection-management]]
 * [[p2p/ift/2026q2-nimlibp2p-gossipsub-queues|nimlibp2p-gossipsub-queues]]
+* [[p2p/ift/2026q2-nimlibp2p-kad-dht-optimizations|nimlibp2p-kad-dht-optimizations]]
 * [[p2p/ift/2026q2-nimlibp2p-logos-core|nimlibp2p-logos-core]]
 * [[p2p/ift/2026q2-nimlibp2p-maintenance|nimlibp2p-maintenance]]
+* [[p2p/ift/2026q2-nimlibp2p-nat-traversal|nimlibp2p-nat-traversal]]
+* [[p2p/ift/2026q2-nimlibp2p-stream-budgets|nimlibp2p-stream-budgets]]
+* [[p2p/ift/2026q2-nimlibp2p-stream-connection-refactor|nimlibp2p-stream-connection-refactor]]
 
 
 ## 2026q1 `ift-ts:p2p:`
@@ -94,5 +98,3 @@ Major future work items on nim-libp2p (also beyond this reporting period are cap
 
 #### `acz:`
 * [[p2p/acz/2025q1-acz-consulting|2025q1-acz-consulting]]
-
-