Sign inSign up

t2fn/chapel-claude-abliterated

By t2fn

Updated 22 days ago

Chapel parallel programming compiler + gdb + valgrind + OpenMPI, tweakcc optimized for HPC

Image
Machine learning & AI
Developer tools
0

490

t2fn/chapel-claude-abliterated repository overview

chapel-claude-abliterated — Chapel Parallel Programming with Claude Code

Cost savings: use the tweakcc reduced system prompt + specialized Chapel tools for the best results. A default Claude Code prompt injects 30K+ characters per turn -- that's $90/MTurn on Sonnet and $150 on Opus for tokens you never read. The tweakcc thinkingVerbs and ablated prompts trim this to ~10K, saving up to 90% on cached reads and ~$4.50/hour in input for a coding session at 100 requests/hour.

Drop a Chapel project into a container and ask Claude to compile, parallelize, debug, and ship it. This image combines Claude Code (abliterated) with Chapel 2.9.0, gdb, valgrind, cmake, make, LLD linker, and full HPC tooling -- a complete Chapel development workstation for parallel computing.

What It Does

This container is a focused Chapel development environment. Mount whatever Chapel projects, parallel applications, or scientific computing code you want to work on, start the container, and Claude automatically gets:

  • Chapel 2.9.0 -- High-productivity programming language for parallel computing (Cray/Intel)
  • chpl -- Chapel compiler with -nl (network-level), -ll (language-level), and -lt (task-level) parallelism support
  • gdb -- Full-screen TUI debugger with native Chapel expression evaluation
  • valgrind -- Memory debugging, profiling, and leak detection for Chapel programs
  • LLD linker -- High-performance LLVM-based linker for fast Chapel builds
  • cmake / make -- Build system support for complex Chapel/C/Fortran projects
  • OpenMP / MPI-ready toolchain -- Compile parallel code with -ll=openmp or link against libmpi
  • Abliterated Claude -- lean ~10K system prompt, 200+ skills, subagent support, and low token cost

No configuration, no setup. Give it a Chapel project and ask Claude to build, parallelize, debug, and ship.

# First, navigate to your project directory.
# cd $HOME/git/my-chapel-project

# Build a Chapel project with Claude working together
docker run -u $(id -u):$(id -g) -e ANTHROPIC_API_KEY=sk-... \
    -v "$PWD:/workdir:rw" \
    t2fn/chapel-claude-abliterated:latest

# Or with podman:
podman run -u $(id -u):$(id -g) --userns=keep-id \
    -e ANTHROPIC_API_KEY=sk-... \
    -v "$PWD:/workdir:rw" \
    t2fn/chapel-claude-abliterated:latest

# Then send a prompt:
# "chpl my code.chpl, run it with 4 locales, and profile with valgrind"

# Or with a local Ollama model:
docker run -u $(id -u):$(id -g) -e OLLAMA_MODEL=gemma4:31b \
    -e OLLAMA_HOST=10.12.2.4 \
    -e ANTHROPIC_API_KEY=sk-... \
    -v "$PWD:/workdir:rw" \
    t2fn/chapel-claude-abliterated:latest

# Or with podman:
podman run -u $(id -u):$(id -g) --userns=keep-id \
    -e OLLAMA_MODEL=gemma4:31b \
    -e OLLAMA_HOST=10.12.2.4 \
    -e ANTHROPIC_API_KEY=sk-... \
    -v "$PWD:/workdir:rw" \
    t2fn/chapel-claude-abliterated:latest

# Claude-tools shell -- invoke a shell with claude-shell capabilities
docker run -u $(id -u):$(id -g) -e ANTHROPIC_API_KEY=sk-... \
    -v "$PWD:/workdir:rw" \
    t2fn/chapel-claude-abliterated:latest /bin/bash

# Or with podman:
podman run -u $(id -u):$(id -g) --userns=keep-id \
    -e ANTHROPIC_API_KEY=sk-... \
    -v "$PWD:/workdir:rw" \
    t2fn/chapel-claude-abliterated:latest /bin/bash

Claude traces the compilation pipeline, calls chpl, gdb, valgrind, and cmake, and returns a natural-language report -- all from one prompt.

With the /bin/bash argument, the container starts an interactive shell where claude-tools are available on $PATH and Claude Code is configured for development. From the shell you can:

  • Run claude-fix to analyze errors and propose fixes
  • Run claude-bash for bash sessions with smarter timeouts
  • Run claude-agent to launch Claude as a persistent agent
  • Send slash commands (/fix, /ask, /plan) directly to Claude
  • Use claude-fix --comment to post inline PR comments or --fix to apply findings directly

For claude-tools examples and full reference, see mijuny/claude-tools.

Why This Container?

Chapel development without the workstation

Building Chapel applications typically means installing the Chapel toolchain, configuring parallelism backends (locales, threads, MPI), and setting up debugging/profiling tools. This container bundles it all and makes it accessible to Claude via a single -p prompt.

Without this containerWith chapel-claude-abliterated
Install Chapel from source with correct dependenciesdocker run -- everything is ready
Compile Chapel code with -nl, -ll=openmp, or MPI flagsClaude knows the parallelism model flags automatically
Debug Chapel programs with gdb and valgrindPre-configured debugger chain ready to go
One tool per terminal sessionchpl, gdb, valgrind, cmake all available
Manual parallelism configuration (locales, threads)Ask Claude -- it knows Chapel's locale/thread model
What Claude can do

The container works with any Chapel project under /workdir:

  • Chapel source files -- compile, run, and debug with chpl/gdb/valgrind
  • Parallel programs -- compile with -nl (locales), -ll=openmp (OpenMP threads), or MPI backends
  • Locale programming -- multi-locale Chapel code with network-level parallelism
  • Task parallelism -- forall, cobegin, coparallel constructs compiled and profiled
  • C/Fortran interop -- link Chapel programs against C libraries via cmake
  • Debugging -- gdb with breakpoints, step through parallel loops, inspect locale state
  • Memory analysis -- valgrind memcheck for Chapel memory management issues
  • Build systems -- cmake-based multi-language Chapel/C/Chapel/Fortran projects
  • Performance profiling -- compile with -g -O2 and profile hot spots in parallel code

Toolkit

Core Compiler Tools
ComponentSourceDescription
Chapelchapel-lang/chapelHigh-productivity parallel programming language (v2.9.0)
chplchapel-lang/chapelChapel compiler with -nl/-ll-openmp/-ll=omp flags
Claude Codeclaude-abliterated/Abliterated CLI with tweakcc, 200+ skills, rules
gdbrockylinux.devFull-screen TUI debugger with Chapel expression support
valgrindrockylinux.devMemory leak detection and profiling (memcheck, cachegrind)
LLDllvm.orgHigh-performance LLVM linker for fast chpl builds
cmakerockylinux.devMulti-language CMake builds for Chapel/C/Fortran projects
makerockylinux.devBuild automation for Makefile-based Chapel projects
Parallelism Models Supported

Chapel supports three levels of parallelism -- Claude knows how to compile and debug each:

ModelFlagDescription
-nl (Network Level)chpl -nl N code.chplMulti-locale execution across networked nodes
-ll=openmpchpl -ll=openmp code.chplOpenMP thread-level parallelism via LLVM
-ll=ompchpl -ll=omp code.chplOpenMP via GCC's libgomp backend
Pre-installed Libraries
CategoryLibraries
Core runtimechapel-runtime, libc, libstdc++, libpthread
Parallelismlibomp (OpenMP), libdl, librt
Math / numericlibm, libblacs, libscalapack (via system packages)
I/Ozlib, libbz2, liblz4, libzstd, libpng
Debugginggdb, valgrind, libasan, libtsan
Why this container is better suited for development

Prompt Claude to use standard Chapel parallelism patterns. You can say "use the chapel skill to compile my code with 4 locales and profile it" and Claude will apply Chapel-specific compilation rules, set locale counts correctly, and verify each segment works.

Directed skills improve code quality and resilience. With abliterated Claude and directed skills:

  • Ask Claude to parallelize: "chpl -nl 8 my-code.chpl with OpenMP threads and run valgrind" -- Claude compiles with correct flags and sets locale/thread counts automatically
  • Break tasks down: instead of a monolithic parallel refactoring, Claude splits the work into subagent tasks -- each module gets compiled and tested independently before integration
  • Ensure debugged efficacy: Claude runs gdb in TUI mode to verify variable evaluation and memory correctness with -g debug symbols
  • Parallelism focused: Claude knows Chapel's locale/thread/task model and can use -nl, -ll=openmp flags correctly without manual configuration

Say "use the chapel skill to..." to directly task Claude with Chapel-specific skills -- chpl compilation with parallelism flags, gdb debugging of multi-locale code, valgrind profiling, and cmake-based builds.

claude-tools

Extends Claude Code with specialized command-line tools. Available scripts are automatically in $PATH and activate via slash commands or Claude's built-in tool use:

claude-fix    # Fix code issues with context-aware edits
claude-ask    # Ask questions, get answers without side effects
claude-bash   # Run bash with smarter timeouts and session persistence
claude-agent  # Launch Claude as a persistent agent

To invoke the claude-tools development shell, add /bin/bash to your docker run or podman run command. This starts an interactive shell where claude-tools are available on $PATH and Claude Code is configured for development:

# claude-tools development shell with Docker
docker run -u $(id -u):$(id -g) -e ANTHROPIC_API_KEY=sk-... \
    -v "$PWD:/workdir:rw" \
    t2fn/chapel-claude-abliterated:latest /bin/bash

# Or with podman:
podman run -u $(id -u):$(id -g) --userns=keep-id \
    -e ANTHROPIC_API_KEY=sk-... \
    -v "$PWD:/workdir:rw" \
    t2fn/chapel-claude-abliterated:latest /bin/bash

From the shell you can run claude-fix, claude-ask, claude-bash, claude-agent, and send slash commands (/fix, /ask, /plan) directly to Claude. Use claude-fix --comment to post inline PR comments or --fix to apply findings directly.

More examples: mijuny/claude-tools/tree/main/README.md

Cost Savings with tweakcc

This container runs on claude-abliterated:rocky10, which uses the tweakcc (tweakcc-fixed) reduced system prompt to cut ~67% off daily-flow prompts -- from ~30K to ~10K characters. For Chapel development, this means:

  • Focused tools, focused rules -- Chapel-specific rules (chpl parallelism flags, locale configuration) are loaded on-demand, not every turn
  • tweakcc thinkingVerbs -- 207 safe movement-only verbs reduce cognitive drift on every turn
  • Cost-per-hour comparison: ~$4.50/hour savings on input tokens alone for a typical 100-request/hour session
  • Better output -- fewer contradictory always-on rules means less "overcorrection" waste on each turn

"Claude" and "Claude Code" are registered trademarks of Anthropic, PBC. This project uses the name "Claude" in a nominative (descriptive) sense -- it identifies the Anthropic product (Claude Code) that this project is built to work with, consistent with the nominative fair use doctrine. The use is necessary: a third-party project cannot refer to "Claude Code" without using the mark itself.

This project is not affiliated with, endorsed by, or sponsored by Anthropic, PBC. The name "chapel-claude-abliterated" is a third-party project name where:

  • "chapel" -- descriptive reference to the Chapel programming language (nominative fair use)
  • "claude" -- descriptive reference to Anthropic's Claude Code (nominative fair use)
  • "abliterated" -- coined portmanteau ("ablation" + "obliteration") representing the reduction of system prompt bloat

The project uses both "Chapel" and "Claude" to describe what it works with, not who it is from.

All other trademarks belong to their respective owners. No implication of endorsement is made by the use of these marks.

Each language-specific container adds a focused toolchain and domain skills on top of claude-abliterated. They are all interchangeable -- use the one that matches your project, or combine them in the same workspace.

Open Source

Source code and suggestions/issues/pull requests can be made at:

https://github.com/t2fn/claude-abliterated

Tag summary

Content type

Image

Digest

sha256:308cf0cba

Size

972.6 MB

Last updated

22 days ago

docker pull t2fn/chapel-claude-abliterated