IPC Infrastructure

Named pipes.
Forged for throughput.

PipeForge provisions high-performance named communication channels between processes — zero temp files, zero polling, zero compromise. First-in, first-out data flow at the speed your stack demands.

terminal
$ pipeforge data_channel
$ ls -l data_channel
prw-r--r-- 1 user staff 0 Apr 7 09:12 data_channel
$ echo "payload" > data_channel &
$ cat < data_channel
payload
99.97%
Server Adoption Rate
<1ms
Channel Provisioning
0
Temp Files Generated
Concurrent Channels

Every feature your pipeline needs. Nothing it doesn't.

PipeForge ships the primitives that matter — battle-tested IPC infrastructure, purpose-built for processes that need to talk.

Instant Pipeline Provisioning

Spin up a named communication channel with a single invocation. pipeforge my_pipe — done. Your FIFO is live, discoverable in the filesystem, and ready for data.

pipeforge NAME
🔐

Precision Access Control

Define permission bits at creation time. Lock down who reads and who writes with fine-grained mode settings. No retroactive patching required.

-m MODE / --mode=MODE
🔄

Smart Queue Synchronization

Readers automatically block until data arrives. No polling loops, no CPU waste, no missed messages. The pipe handles orchestration so you don't have to.

blocking read
⏸️

Backpressure Management

Writers pause until a consumer is ready. Built-in backpressure prevents buffer overflow and data loss — the pipe knows when you're ready.

blocking write
📂

Filesystem-Native Discovery

Every pipe appears as a real file in your directory tree. Standard tools can discover, inspect, and manage your communication infrastructure.

FIFO special file
🌐

Fan-Out / Fan-In Architecture

Multiple producers and consumers on a single channel. No broker required, no message queue overhead. Pure peer-to-peer IPC at the filesystem level.

multiple readers/writers
🗑️

Zero-Artifact Cleanup

When you're done, remove the channel with a single operation. No orphaned temp files, no leaked resources, no ghost processes. Clean as the day you started.

rm cleanup
🛡️

Security Context Engine

Integrated SELinux and SMACK security contexts. Tag every pipe with a policy label at creation time for zero-trust IPC that compliance teams will love.

-Z / --context

Three steps to production-grade IPC

PipeForge reduces the entire inter-process communication lifecycle to three high-leverage operations.

01

Forge the Channel

Provision a named pipe with a single command. PipeForge allocates a FIFO special file in your filesystem — a persistent, discoverable IPC endpoint that any process can connect to.

$ pipeforge --mode=0644 data_bus
02

Connect Producers & Consumers

Any process can open the channel for reading or writing. PipeForge's synchronization layer handles blocking — writers wait for readers, readers wait for data. Zero orchestration code required.

$ producer_service > data_bus &
$ consumer_service < data_bus
03

Ship & Scale

Data flows first-in, first-out at kernel speed. No polling, no temp files, no serialization overhead. When you're done, remove the channel — zero artifacts left behind.

$ rm data_bus # clean teardown

IPC Architecture, Visualized

Understand how PipeForge channels sit between your processes — no broker, no overhead, just pure kernel-level data flow.

Producers
Process A PID 1024
Process B PID 1025
Process C PID 1026
PipeForge Channel
FIFO Buffer
Mode: 0644 Type: FIFO Blocking: ON
Consumers
Worker X PID 2048
Worker Y PID 2049

Pipe Flow Visualizer

Watch data packets traverse a PipeForge channel in real time. Click "Send" to push data through the pipe and observe FIFO ordering, blocking behavior, and throughput.

Sent: 0 Received: 0 In-flight: 0
Producer
PipeForge Channel
Consumer
00:00.000 Channel initialized. Waiting for data...

Built for every IPC pattern

From microservice coordination to real-time log streaming, PipeForge channels are the high-leverage primitive your infrastructure is missing.

📊

Real-Time Log Aggregation

Stream logs from multiple services into a single PipeForge channel. One consumer process aggregates, filters, and forwards — no log broker required.

$ pipeforge /var/log/app_bus
$ service_a >> /var/log/app_bus &
$ service_b >> /var/log/app_bus &
$ aggregator < /var/log/app_bus
🔗

Pipeline Orchestration

Chain producer and consumer processes through named channels for complex ETL workflows. Each stage reads from one pipe and writes to the next.

$ pipeforge stage_1 stage_2
$ extract > stage_1 &
$ transform < stage_1 > stage_2 &
$ load < stage_2
🛡️

Privilege-Separated I/O

Use permission-controlled pipes to safely pass data between processes running at different privilege levels. IPC with built-in access control.

$ pipeforge --mode=0620 secure_chan
$ high_priv_writer > secure_chan &
$ low_priv_reader < secure_chan

Producer-Consumer Decoupling

Decouple fast producers from slow consumers with built-in backpressure. The pipe blocks writers when the consumer isn't ready — no data loss, no overflow.

$ pipeforge event_queue
$ fast_emitter > event_queue &
$ slow_processor < event_queue

See PipeForge in action

A complete inter-process communication workflow — from channel creation to data flow to cleanup.

pipeforge-demo — zsh
$ pipeforge --mode=0644 ipc_channel
$ ls -l ipc_channel
prw-r--r-- 1 deploy staff 0 Apr 7 09:15 ipc_channel
$ file ipc_channel
ipc_channel: fifo (named pipe)
$ stat -c '%F %a %n' ipc_channel
fifo 644 ipc_channel
# Producer writes, consumer reads — FIFO ordering guaranteed
$ echo "first payload" > ipc_channel &
[1] 48201
$ cat < ipc_channel
first payload
[1]+ Done echo "first payload" > ipc_channel
# Clean teardown — zero artifacts
$ rm ipc_channel
$ ls ipc_channel
ls: cannot access 'ipc_channel': No such file or directory

Pipelines at every scale

From solo developers prototyping IPC patterns to enterprise teams running mission-critical data buses.

Starter
$0/mo

One channel. One direction. Get the feel.

  • 1 named pipe
  • Default permissions only
  • Single reader, single writer
  • Blocking I/O
  • Custom --mode
  • Security contexts
  • Multiple readers/writers
Team
$49/seat/mo

SELinux contexts. Audit trails. Compliance-ready.

  • Everything in Pro
  • SELinux / SMACK contexts
  • --context=CTX integration
  • Pipeline audit logs
  • SSO & RBAC
  • Priority support
  • Dedicated infra
Enterprise
Custom

Dedicated pipeline infrastructure built for your org.

  • Everything in Team
  • Dedicated FIFO clusters
  • Custom security policies
  • 99.999% SLA
  • On-prem deployment
  • Dedicated solutions engineer
  • SOC 2 & HIPAA compliance

Trusted by teams shipping real-time infrastructure

★★★★★

"PipeForge replaced our entire message queue layer. We went from managing a Kafka cluster to three named pipes and a shell script. Our IPC latency dropped to sub-millisecond. This is the 10x improvement we've been chasing."

Elena Whitfield
Jess Kim
VP of Infrastructure, DataMesh
★★★★★

"The blocking I/O model is genuinely category-defining. Writers wait for readers. Readers wait for data. No orchestration code, no race conditions, no lost messages. We deleted 2,000 lines of synchronization logic."

Yara Whitford
Marcus Adeyemi
Principal Engineer, ScaleForge
★★★★★

"We built our entire ETL pipeline on PipeForge channels. Each stage reads from one pipe, processes, and writes to the next. It's pipelines as code, and it compounds. Our data team ships 3x faster now."

Dina Whitmore
Sara Reyes
CTO, PipelineOps
★★★★★

"As a solo dev, I needed lightweight IPC without the overhead of a message broker. PipeForge is the answer. One command, one file, and my processes just talk. Ramen-profitable infrastructure at its finest."

Eliot Grant
Toby Lindqvist
Indie Developer & Build-in-Public Creator

Frequently asked questions

PipeForge's Smart Queue Synchronization handles this automatically. The reader blocks — pausing execution with zero CPU usage — until a writer connects to the channel. This is by design: it guarantees that no read ever returns empty and no data is ever missed.

Yes. PipeForge supports fan-in architecture out of the box. Multiple producers can write to a single named pipe. Data arrives at the consumer in the order it's written — strict FIFO ordering is always maintained within each write operation.

Use the --mode flag to set file permission bits at creation time. For example, pipeforge --mode=0640 my_pipe creates a channel where the owner can read and write, the group can read, and others have no access. For enterprise deployments, the -Z and --context flags enable SELinux and SMACK security labeling.

PipeForge channels operate at the kernel level — data never touches the network or disk. It flows directly between process memory spaces through a kernel-managed buffer. There's nothing to intercept because there's no transport layer. It's the most secure IPC pattern possible: no network, no files, no attack surface.

No. PipeForge channels are completely brokerless. There's no daemon, no server process, no runtime dependency. The named pipe exists as a file in your filesystem — processes connect to it directly. This is why PipeForge is the highest-leverage IPC primitive available: zero operational overhead.

See mandō's portfolio