Review and run model-generated code with effect tracking

Jacquard

Jacquard is a FriendMachine research project for running, reviewing, simulating,
and trusting programs written by models and reviewed by people. Start with the
human-friendly introduction to Jacquard.

Concretely, it is a small programming language where every function signature
lists the outside-world effects the function may perform - network, files,
clock, randomness - and the runtime refuses any effect you have not granted
on the command line. A reviewer reads the signature to learn what a change
can touch; the checker guarantees the signature is complete. The
implementation is an OCaml checker and interpreter, a compiler that accepts
public .jac or lower-level .jqd files and produces standalone native
binaries by emitting C, the jac command-line tool, a standard library written
in Jacquard itself, and a test framework called Warp. Version 0.2 works end to
end but is a research prototype, not a production language;
docs/release/0.2/LIMITS.md is the honest boundary.

Install the 0.2 release without OCaml or opam:

curl -fsSL https://raw.githubusercontent.com/jbwinters/jacquard-lang/jacquard-core-0.2.0/scripts/install.sh | sh
~/.local/bin/jac run ~/.local/share/jacquard/demos/basics/m1-fact.jac

The expected output is 120. Linux x86-64, macOS Intel, and macOS Apple
Silicon binaries are published; development from source is documented below.

Then run one policy under concrete and probabilistic telemetry worlds, followed
by sampled and exhaustive Warp checks:

sh ~/.local/share/jacquard/demos/case-studies/release-risk/run.sh

For Humans

Most languages tell you what a program computes. Jacquard also exposes which
effects it may perform, finite discrete uncertainty, and canonical program
identity. Tools can inspect all three because they live in the language rather
than only in comments, logs, or your memory of the codebase.

Things you can do here that most languages cannot offer:

• Read one line and see the effects a function may perform. A signature like
  (text) ->{net} text says the function may perform the net effect. The
  Jacquard runtime rejects unhandled world effects unless their authority is
  explicitly granted with --allow, including effects performed by dynamic
  code. This is language-level enforcement in a research runtime, not a
  substitute for an operating-system sandbox.
• Run one program against many worlds. The same code can run against the
  shipped deterministic network stub, a scripted response list, a recorded
  trace, or a probability model of how servers usually behave. A real network
  adapter does not ship yet; a host integration can implement the same effect
  boundary without changing the Jacquard program. A handler is the piece
  that answers a program's requests to the outside world; you swap the
  handler, and the code never changes. This can replace much conventional
  mocking at effect boundaries and makes "what would my agent do if the API
  went down?" an ordinary test. If this sounds like dependency injection: an
  injected dependency is the special case of a handler that resumes the
  program exactly once. A handler can also decline to resume, aborting the
  rest of the computation cleanly, or resume many times, forking the rest of
  the program to explore every outcome. That last case is what makes
  exhaustive testing and exact inference ordinary library code here.
• Enumerate exact probabilities for finite discrete models. A program can
  sample weighted choices and record evidence, and enumeration lists every
  reachable outcome with its exact probability. The repair demo below treats a
  failing test as evidence and computes which patches remain possible and how
  likely each is.
• Rename and reformat without changing canonical identity. Jacquard hashes
  canonical resolved structure rather than source bytes. Comments, formatting,
  provenance, and ordinary local or term renames are erased; pure tests rerun
  only when canonical code or dependency content changes. This is structural
  identity, not a proof that arbitrary programs are behaviorally equivalent.

The bet behind all of this: when most code is written by machines, the humans
reviewing it need the language itself to answer "what can this touch, and how
sure are we" without reading every line.

Jacquard also lets public call sites say what each argument means:

resize(image, scale: ratio) = (image, ratio)

resize(photo, scale: 2)

An unlabeled positional prefix may come first; labeled arguments may follow in
any order and still run left to right as written. Labels are explicit API, not
guesses from local binder names: top-level functions and effect operations
declare them, while constructors reuse their declared field labels. Calls are
still exact-arity and uncurried-there are no defaults, label puns, or named
calls through local and higher-order values. The lower-level .jqd carrier
remains positional.

The Review Case In Miniature

Suppose a model hands you this one-line change in Python:

def normalize_name(name):
    return lookup_alias(name).strip().lower()

To learn whether the change can reach the network, you read lookup_alias,
then everything it calls. The answer lives in the transitive closure of the
diff, and nothing checks whatever answer you settle on.

The same change in Jacquard arrives with this checked signature:

normalize-name : (text) ->{net} text

Some function below lookup-alias performs a net operation, so net
surfaces in the row of every caller until a handler discharges it. The
checker computes the row; a signature that omits an effect is a type error.
The reviewer's first question about generated code - what can this touch -
is answered on the first line of the diff, before reading any body. At run
time the same row is enforced: jac run refuses the program without
--allow net, and that includes effects performed by dynamically loaded
code.

Effect rows are also what separates this from an ordinary type system: they
propagate through the call graph without hand annotation, and they are tied
to runtime authority. Ordinary types describe the values a function handles;
the row describes what running it may do to the world, and the runtime holds
it to that.

For Agents

Read docs/SKILL.md first. It compresses the kernel, the CLI, the prelude,
Warp testing, and the known gotchas into one file, and it loads as a project
skill from docs/SKILL.md. The language is deliberately small enough that an
agent with no Jacquard in its training data can work from that one file.
Operating rules are in AGENTS.md. What will save you time:

• Behavior is pinned by evidence: cram transcripts under test/cli/, corpus
  goldens, demo scripts, and docs/release/0.2/CLAIMS.md. If a pin fails,
  treat it as information about your change, and never weaken a pin to make a
  diff pass.
• The kernel is 27 forms (docs/ast.md); .jac is a projection onto those
  forms, and bootstrap .jqd remains permanently supported. Treat the shipped
  surface boundary and its parked follow-ups as release evidence, not as a
  frozen grammar; do not add out-of-scope features (AGENTS.md lists them).
• The development gate is dune build @all && dune runtest && dune fmt
  followed by a clean git diff --exit-code.

Core Ingredients

For readers who speak programming languages:

• One uniform representation: every form is a (head, meta, args) triple, and
  the kernel grammar has 27 forms. Quoted code is ordinary data.
• Algebraic effects with deep, mode-aware handlers. A multi operation has a
  reusable continuation and can resume zero, one, or many times, which makes
  exhaustive search and exact inference ordinary library code. A once
  operation instead binds an affine Resume: the checker reports E0816 when
  one possible path consumes it twice, and the runtime retains E0906 as a
  repeated-resume backstop for each captured instance.
• Explicit capability grants. The runtime installs handlers for the outside
  world only for effects you pass with --allow; there is no ambient
  authority.
• Type-and-effect rows. Every arrow carries the set of effects the function
  may perform, so a program's inferred row is its authority manifest.
• Discrete probabilistic programming as a library: sample and observe are
  effect operations, and each inference algorithm is a handler.
• Content-addressed definitions. Identity is a hash of canonical resolved
  structure with non-identity metadata erased, so formatting, comments, and
  ordinary local or term renames change nothing downstream. Explicit external
  call labels are a separately stored, hash-bound API contract; changing those
  labels for the same callable identity is rejected.
• Tooling that leans on the above: formatter, structure-aware differ, Warp
  tests with a content-addressed cache, record/replay, and a reproducible
  release evidence pack.
• A native AOT path that emits C, specializes and caches units by content hash,
  and is differential-tested against the interpreter under clang and gcc.

Design Lineage

The design borrows deliberately from languages whose ASTs and semantics were
studied during planning; docs/ast.md records each debt in detail:

• Unison: effects carried on function arrows, operations as ordinary
  functions, content-addressed definitions, and cycle hashing.
• Koka: effect rows, uncurried arrows, the tail-resumptive handler
  discipline, and a warning heeded about row-inference ergonomics.
• Racket: scope-set hygiene for quoted code.
• Haskell: strict evaluation as the verdict on laziness, and exhaustive
  matching as a checker obligation rather than a lint.
• OCaml: the host language, plus negative lessons on builtin structural
  equality and on deferring ad-hoc polymorphism.
• Roslyn (C#): full-fidelity syntax metadata so tools can round-trip source
  without losing comments or formatting.

The prototype is complete against its original core plan and has since added
the public surface syntax, ringed standard library, Warp properties and cache,
native compilation, packaged binaries, and product-scale case studies. The RC1
semantic boundary remains historical; the current successor is pinned by 984
Alcotest/QCheck cases, 60 cram transcripts, 28 documentation examples, native
sanitizer/leak/fuzz lanes, and fresh-clone evidence workflows. RC2 repaired
binary-demo packaging; RC3 adds an explicit
runtime/output license exception and packages the native runtime. The current
successor distribution relicenses Jacquard under Apache License 2.0 and keeps
that runtime/output permission as an explicit clarification. These licensing
and packaging changes do not change the language semantics pinned at RC1.

What It Looks Like

Here is one handler resuming one continuation twice. The block is copied
byte-for-byte to test/docs-doctest/fixtures/readme-multishot.jac and run by
the documentation test lane:

multi effect Choice where {
  choose : () -> Bool
}

handle {
  match choose() {
    | True -> 1
    | False -> 2
  }
} {
  | return x -> x
  | choose() resume continue -> add(continue(True), continue(False))
}

$ jac run test/docs-doctest/fixtures/readme-multishot.jac
3

Reading it line by line: multi effect Choice declares an effect with one
operation, choose, which takes nothing and answers a boolean. multi means
its continuation may be resumed more than once. The handle
block runs the code in the first braces. When that code calls choose(),
control jumps to the matching clause below, which receives the paused
rest-of-the-computation as continue. The clause calls continue twice,
once per answer, so the match runs once with True (producing 1) and once
with False (producing 2), and add combines the two runs into 3. The
return x -> x clause says finished runs pass through unchanged.

That ability to resume more than once is why exact Bayesian inference is a
library handler here rather than a runtime feature. The repair demo builds on
it: mutate a buggy program's quoted AST into candidate patches, treat a failing
test as an observation, and read off the updated probabilities. Running
candidate code is an authority, so the pure
step still runs (it counts eight candidate patches) and then the demo refuses
until you grant the rest:

$ jac run demos/tooling/repair.jac
8
error[E0814]: The program requires an effect that was not granted
  Cause: This program requires eval [meta/high] — run code constructed or loaded at runtime, which is not granted (performed via `posterior-over-patches`).
  Next step: grant it with --allow eval, or handle the effect in the program
$ jac run demos/tooling/repair.jac --allow eval

Under the grant, one failing test leaves two surviving patches: the intended
fix at 0.75 and a patch that games the suite at 0.25. Adding one regression
test prunes the impostor, and the surviving fix prints as a one-line canonical
diff: - sub + add. See sh demos/tooling/repair.sh for the full transcript.

Install A Release Binary

Most users do not need OCaml or opam. Install the reviewed 0.2 binary with:

curl -fsSL https://raw.githubusercontent.com/jbwinters/jacquard-lang/jacquard-core-0.2.0/scripts/install.sh | sh

The installer detects your OS and CPU, downloads the matching archive and
SHA-256 checksum, refuses a checksum mismatch, and installs under ~/.local
by default. Make sure ~/.local/bin is on PATH, then run:

jacquard --version
jac --version

jac is the short alias for jacquard. Both commands set JACQUARD_PRELUDE
from the installed package, so ordinary runs do not need an environment variable:

jac run ~/.local/share/jacquard/demos/basics/m1-fact.jac

Narrative demos ship with launchers that choose the installed binary and
prelude automatically. They do not require Dune:

DEMO_ROOT="$HOME/.local/share/jacquard/demos"
sh "$DEMO_ROOT/case-studies/release-risk/run.sh"
sh "$DEMO_ROOT/worlds/agent-dream.sh"
sh "$DEMO_ROOT/worlds/escrow/run.sh"

Use these launchers rather than directly running a probabilistic model or a
multi-file entrypoint. The launcher selects infer where observation requires
it and assembles related files in isolated scratch space.

To install under a different user-owned prefix:

curl -fsSL https://raw.githubusercontent.com/jbwinters/jacquard-lang/jacquard-core-0.2.0/scripts/install.sh \
  | JACQUARD_INSTALL_PREFIX="$HOME/.jacquard" sh

Set JACQUARD_INSTALL_VERSION to install a different release tag. Supported
binary targets are linux-x86_64, macos-x86_64, and macos-arm64; other
platforms currently require the development setup.

Release archives are attached to jacquard-core-* GitHub releases. Each
archive contains bin/jacquard, bin/jac, libexec/jacquard/jacquard,
share/jacquard/prelude, share/jacquard/demos, the native C runtime, and the
license, notice, exception, and trademark documents.

Development Quick Start

These commands assume a fresh clone and asdf available for installing opam.
If you already have opam 2.5.x, start at the local switch step. If opam
is already initialized on your machine, skip opam init.

git clone https://github.com/jbwinters/jacquard-lang.git
cd jacquard-lang

asdf plugin add opam https://github.com/asdf-community/asdf-opam.git
asdf install opam 2.5.1
asdf set opam 2.5.1
asdf reshim opam 2.5.1

opam init -y --no-setup --bare
opam switch create . ocaml-base-compiler.5.1.1 -y
eval "$(opam env)"

opam install --deps-only . --with-test --with-dev-setup --with-doc -y
opam exec -- dune build @all
opam exec -- dune runtest
opam exec -- dune fmt
git diff --exit-code

The switch step compiles OCaml 5.1.1 from source, so expect the first setup to
take around ten minutes.

The final git diff --exit-code is part of the development contract: formatting
must leave the worktree clean unless you intentionally commit the formatting
diff.

Expected versions after setup:

• opam 2.5.1 from .tool-versions
• OCaml 5.1.1 from the repo-local _opam/ switch
• dune, ocamlformat, alcotest, qcheck, digestif, menhir,
  cmdliner, odoc, utop, and ocaml-lsp-server from jacquard.opam

In a new shell inside an existing checkout, run:

eval "$(opam env)"

_opam/ is intentionally ignored. It is a local build artifact, not source.

Running Jacquard

During development, use the built binary through Dune:

opam exec -- dune exec jac -- --help
opam exec -- dune exec jac -- --version

Many direct CLI commands need the prelude. From the repository root:

export JACQUARD_PRELUDE=$PWD/prelude
opam exec -- dune exec jac -- run demos/basics/m1-fact.jac

The main commands are:

jac run FILE.jac [--allow fs] [--allow net] [--dry-run]
jac relate FILE.jac --vary schedule=N --seed S [--allow EFFECT ...]
jac relate FILE.jac --vary secret=NAME --seed S [--allow EFFECT ...]
jac relate FILE.jac --vary grant=net|infer|dist --seed S
jac check FILE.jac [--print-sigs] [--manifest fs,net,console]
jac hash FILE.jac
jac fmt FILE.jac
jac diff FILE_A.jac FILE_B.jac
jac diff STORE_A STORE_B
jac infer enumerate MODEL.jac
jac infer lw MODEL.jac --seed 42 --samples 100000
jac replay TRACE.jqd PROGRAM.jqd [--fork '1=(response 500 "down")']
jac test TESTS.jac [TESTS.jqd ...] [--exhaustive] [--schedules N --seed S] [--cache-dir CACHE]
jac build FILE.jac -o PROG
jac export FILE.jac -o FILE.jqd
jac governance check FILE.jac [--output-format text|json-v1]
jac governance verify-run RUN_BUNDLE.jqd
jac governance reconcile RECONCILIATION_BUNDLE.jqd
jac governance explain PROPOSAL_ID --bundle RECONCILIATION_BUNDLE.jqd [--output-format text|json-v1]
jac why-effect EFFECT --source FILE.jac [--output-format text|json-v1]
jac host worker --store DIR

jac host worker is the opt-in serial carrier for the experimental
jacquard-host-v0 protocol: a trusted host process invokes one checked stored
term and answers its typed root operations over length-prefixed JSON frames on
stdin/stdout (docs/host-worker-v0.md). Ordinary commands never use it.

.jac is the source format people and agents write. .jqd is the lower-level
format that .jac files reduce to - a small fixed grammar of 27 forms, called
the kernel - and it remains fully supported as the internal/debug syntax,
quote notation, and format of record. run, check, hash, fmt, diff,
infer, and test select surface syntax by extension. Native build accepts
either format without writing an intermediate twin; replay programs, the
prelude, and many internal fixtures continue to use .jqd.

Ordinary programs and demos need only a .jac source file. Do not hand-author
a .jqd twin unless a conformance test specifically needs to prove that both
formats lower to the same kernel and hash. The paired files retained in the
corpus and selected demos are evidence fixtures, not an authoring requirement.

Native compilation

jacquard build accepts a public .jac program directly (or a retained kernel
.jqd carrier) and compiles it and its reachable declarations to a
standalone binary. Within the documented native subset, its output is
byte-identical to jacquard run - stdout, stderr, and exit codes, pinned by a
differential harness in CI (scripts/native-diff.sh). The effect-and-handler
kernel compiles, including capturing and multi-shot handlers, and code values
compile since task 73 - quotes, splices, and the structural code ops. Dynamic
Eval, interpreted Task scheduling, and typed Channels stay on the interpreter
tier.

export JACQUARD_PRELUDE=$PWD/prelude
export JACQUARD_RUNTIME=$PWD/runtime
jac build demos/tooling/word-count.jac -o word-count
echo "some words some" | ./word-count --allow console

Build uses the same surface parse/lower/resolution pipeline as check and hash
and does not create a .jqd twin. Use jac export INPUT.jac -o OUTPUT.jqd
only when conformance evidence or kernel debugging needs an explicit canonical
carrier. Export is deterministic and exclusive/atomic; it preserves semantic
member hashes and quote namespace markers, while intentionally erasing
comments, formatting, spans, documentation, and provenance metadata.
Export resolves and canonicalizes input but does not typecheck it; use
jac check, jac run, or jac build when typechecking is required.

Requirements and knobs:

• Release binaries discover their packaged prelude and C runtime
  automatically. Source checkouts may set the two variables shown above.
• A C toolchain: clang (any recent) or gcc. Tail calls are O(1) stack on
  every toolchain: musttail on clang and gcc 15+, a trampoline below
  them (the emitted C is identical either way).
• The binary parses --allow EFFECT for its implemented root grants
  (console, clock, fs, dist, and infer), plus --seed N for the
  sampling grant. It rejects unsupported grants such as net, eval, and
  secret, and refuses --infer-cache and --dry-run (interpreter tooling)
  with pointed errors.
• JACQUARD_STACK_MB sizes the program stack (default 1024): deep
  non-tail recursion is real C recursion in this backend.
• Compiled units cache under .jacquard-native/, keyed by content, so
  an unchanged program relinks without recompiling.
• Measured performance lives in docs/benchmarks.md - nine scenarios
  with interpreter, native (both toolchains), Python, and hand-C
  columns - with the claim boundaries in docs/native-compilation.md
  (reproduce with scripts/native-bench.sh).

Demos

Start with these from the repo root after dune build @all. The same scripts
also work in an installed bundle without opam or Dune:

opam exec -- sh demos/case-studies/stormglass/run.sh
opam exec -- sh demos/case-studies/release-risk/run.sh
opam exec -- sh demos/basics/m1.sh
opam exec -- sh demos/inference/m3.sh
opam exec -- sh demos/worlds/agent-dream.sh
opam exec -- sh demos/worlds/preflight.sh
opam exec -- sh demos/tooling/repair.sh
opam exec -- sh demos/concurrency/run.sh

What they show:

• case-studies/stormglass/: one checkout policy under simulated network and
  clock laws, exact incident forecasts, and Warp proofs over all 27 worlds.
• case-studies/release-risk/: one release policy under concrete and
  probabilistic telemetry, plus a Warp safety proof over all 18 worlds.
• basics/m1.sh: factorial, multi-shot choice, and gated eval.
• inference/m3.sh: one model under exact enumeration and likelihood weighting; same
  model hash, different inference handler.
• inference/clarifying-question.sh: an agent computes whether asking the user a
  question is worth the interruption (value of information).
• worlds/agent-dream.sh: one policy under scripted and probabilistic world handlers.
• worlds/preflight.sh: candidate agent plans scored under alternate worlds; the
  live policy still needs a Net grant after the dreams pass.
• inference/ambiguity-pipeline.sh: an extraction pipeline that keeps its uncertainty;
  the user's click becomes an observe.
• tooling/showcase-warp-tests.sh: Warp checks for the clarifying-question,
  dream-mode, and ambiguity demos.
• tooling/repair.sh: program repair as Bayesian inference; a bug report is an
  observation over computed single-edit patches, and the most likely patch
  prints as a one-line canonical-structure diff.
• concurrency/run.sh: one task program under FIFO, seeded, exhaustive, and
  strict replay scheduling, with exact child-authority signatures and eight
  replayable schedule worlds. This developer evidence demo requires a source
  checkout built with Dune.
• worlds/m4-hostile.sh: generated-looking code that reaches for net; signatures and
  manifests expose the authority.
• worlds/escrow/run.sh: product-shaped generated workflow with manifest, dry-run,
  Warp tests, fault exploration, replay, canonical diff, and approval by hash.

Demo paths are canonical within the categorized directories; there are no
flat compatibility aliases. The full catalog is in demos/README.md.

All public demo outputs are pinned by cram tests (recorded command-line
transcripts that fail on any drift), especially test/cli/demos.t,
test/cli/hostile-demo.t, test/cli/escrow.t, test/cli/showcase.t, and
test/cli/repair.t, test/cli/preflight.t, plus test/cli/case-studies.t for
the larger applications.

Release Evidence

The current release evidence pack lives in docs/release/0.2/. Historical 0.1
evidence remains byte-preserved under docs/release/0.1/.

To reproduce the release evidence from this checkout:

JACQUARD_RELEASE_REF=HEAD JACQUARD_RELEASE_BASE=c0f570501b751865c0c0584d9b15be08b6ec1cde scripts/release/reproduce-0.2.sh

The script installs dependencies, builds, runs the full test suite, checks
formatting, runs public demos, runs gauntlet tests, records jacquard --version,
and writes generated evidence under .scratch/release/0.2/. It also checks
the complete release-diff manifest, historical publications, parser-depth and
GM.12B evidence, and native memory/differential/leak/fuzz lanes under both
Clang and GCC.

Key release docs:

• docs/release/0.2/EVIDENCE.md: artifact, inventory, evidence lineage, and gate
• docs/release/0.2/CLAIMS.md: integrated claims with adjacent caveats
• docs/release/0.2/REPRO.md: fresh-clone reproduction and promotion steps
• docs/release/0.2/FREEZE.md: distribution and retained semantic identities
• docs/release/0.2/GAUNTLET.md: adversarial classes present and omitted
• docs/release/0.2/LIMITS.md: explicit non-goals and trusted boundaries
• docs/release/0.2/DECISION.md: RC1 and same-commit final decision
• docs/release/0.2/RELEASE-NOTES.md: public contents and install command
• docs/release/named-call-arguments/DECISION.md and EVIDENCE.md: post-0.2
  direct named-call contract, hash-bound ABI, proving tests, and non-claims
• docs/release/structured-concurrency/EVIDENCE.md: successor C0-C2 publication
  claims plus the shipped interpreted C3 Channel runtime, exact counts, demo,
  and proving tests
• docs/release/structured-concurrency/LIMITS.md: structured-concurrency
  caveats and explicit C4 non-claims
• docs/release/governed-membranes/DECISION.md: bounded decision to advertise
  deterministic governance for the frozen typed Workspace v0 facade as an
  evidence-backed research reference implementation
• docs/release/governed-membranes/CLAIMS.md: D61-D73 claims mapped to exact
  executable evidence and adjacent negative boundaries
• docs/release/governed-membranes/LIMITS.md: trusted-host, authority,
  recovery, simulation, secret, Audit, and production-readiness limits

Repository Map

• .github/: CI, release evidence workflow, and PR template.
• AGENTS.md: operating notes for future coding agents.
• bin/: jacquard CLI entry point.
• corpus/: conformance corpus and golden outputs.
• demos/: runnable examples and product-shaped demos.
• docs/: design docs, tutorial, CI/CD, Warp, stdlib, errors, release evidence.
• prelude/: Jacquard standard library and effect declarations.
• scripts/release/: reproducible release evidence script.
• spec/: kernel AST, canonical serialization, and frozen host-protocol specs.
• src/: OCaml implementation.
• test/: Alcotest/QCheck suites plus cram CLI transcripts.
• jacquard.opam, dune-project: package and build metadata.

Implementation Map

• src/form.ml, src/meta.ml, src/span.ml: uniform triple and metadata.
• src/reader.ml, src/printer.ml: bootstrap .jqd notation and formatter.
• src/kernel.ml: validator and typed kernel AST.
• src/resolve.ml: names to content-addressed references.
• src/canon.ml, src/hash.ml: HASH_V0 canonical serialization and hashing.
• src/store.ml: object store and mutable name index.
• src/value.ml, src/eval.ml: CPS evaluator and mode-aware deep handlers.
• src/types.ml, src/check.ml: type/effect inference, rows, manifests,
  exhaustiveness.
• src/prelude.ml: prelude loader, builtin wiring, and root grants.
• src/infer_dist.ml: exact enumeration and likelihood weighting.
• src/diff.ml: canonical-structure diff over stores.
• src/warp.ml: Warp test discovery, running, cache, and properties.
• src/host_protocol_v0.ml: strict framing, JSON, limit-selection, shutdown,
  bounded first-order type/value codecs, invoke preflight, and serial session
  accounting for the experimental host protocol.
• src/host_worker.ml: the opt-in jac host worker process carrier that
  evaluates one preflighted invocation over stdin/stdout frames.

Documentation Map

Read these in order if you are new:

1. docs/README.md: documentation index and suggested reading paths.
2. docs/tutorial.md: runnable user-facing examples.
3. demos/README.md: demo catalog and what each demo proves.
4. docs/ci-cd.md: GitHub checks and release evidence process.
5. docs/release/0.2/EVIDENCE.md: current release evidence overview.

Deeper design references:

• docs/whitepaper.tex: historical initial design thesis, motivation, risks,
  and related work; its roadmap and implementation-status sections are
  outdated.
• docs/ast.md: implemented kernel AST contract and retained design reasoning.
• spec/jacquard-kernel-ast-m0.md: implemented kernel source-of-truth spec.
• spec/serialization.md: canonical byte format.
• docs/host-boundary.md: host/Core ownership and trust boundary.
• spec/host-protocol-v0.md: frozen experimental host envelopes, process
  framing, limits, failures, and conformance-vector contract. The library now
  validates one exact checked invocation-including its store closure, pinned
  interface, typed arguments, grants, and closed once-operation registry-but
  does not yet run it or expose a process worker.
• docs/stdlib.md: prelude and ringed standard library.
• docs/warp-testing.md: Warp testing model.
• docs/errors.md: diagnostic catalog.
• docs/development-plan.md: completed historical implementation plan.

Development Workflow

Before opening a PR:

eval "$(opam env)"
opam exec -- dune build @all
opam exec -- dune runtest
opam exec -- dune fmt
git diff --exit-code

When adding valid corpus files, regenerate golden hashes:

opam exec -- dune exec test/gen_goldens.exe

When touching release-facing demos, claims, CI, or semantics, also run:

JACQUARD_RELEASE_REF=HEAD JACQUARD_RELEASE_BASE=c0f570501b751865c0c0584d9b15be08b6ec1cde scripts/release/reproduce-0.2.sh

CI/CD

GitHub Actions separates independently retryable evidence:

• CI / Development gate: build, full tests, clean formatting, version smoke,
  and release-doc presence on PRs, main, and release/**.
• CI / Native parity (clang|gcc): runtime memory, differential, leak, and
  seeded fuzz evidence for both supported C compilers.
• Governance / Governance playground: lint, types, unit/accessibility tests,
  production build, and browser/keyboard/offline-network checks.
• GM12B / GM12B exhaustive forwarding evidence: the scoped 50,000-case
  forwarding proof, with a successful no-op result outside its dependency
  closure.
• Release Evidence / Reproduce 0.2 evidence: release branches, jacquard-core-*
  tags, and manual dispatch; runs scripts/release/reproduce-0.2.sh and uploads
  transcripts.
• Release Binaries: jacquard-core-* tags and manual dispatch; builds
  Linux/macOS tarballs with jacquard, jac, the prelude, demos, and native
  runtime sources.

See docs/ci-cd.md for branch protection recommendations.

Current Limits

Jacquard core is a research prototype, not a production platform. The .jac
surface is implemented and supported but remains an evolving v0 projection
onto the permanent 27-form kernel. Native AOT compilation and C-toolchain
optimization ship. parallel.map and parallel.both remain pure, sequential
optimization hints. The interpreted runtime now supports opaque scoped Tasks,
cooperative cancellation, fail-fast language scopes, an OCaml-only Collect
policy seam, FIFO and seeded scheduling,
versioned strict replay, bounded exhaustive schedule enumeration, and scoped
typed channels with rendezvous and buffered FIFO behavior, close, cancellation,
and exact run/scope ownership. It does not provide native root scheduling or
native Channel execution, preemptive cancellation, finalizers, shared memory,
channel select or timeouts, actors/supervision, host scheduling, or real
asynchronous host I/O at this evidence base. A VM/JIT, continuous
distributions, gradients,
typed staging, language package management, self-hosting, and formal soundness
proofs also do not ship. World grants remain coarse. See
docs/release/0.2/LIMITS.md for the current integrated boundary,
docs/release/0.1/LIMITS.md for the historical Core 0.1 boundary, and
docs/release/structured-concurrency/LIMITS.md for the successor C0-C3 boundary.
docs/host-boundary.md freezes the ownership and trust model, and
spec/host-protocol-v0.md freezes the experimental language-neutral envelopes,
process framing, limits, and schema/state vectors. jac host worker is the
experimental serial carrier for that protocol; no stable ABI, adapter,
cross-language conformance kit, or HTTP server ships in this repository today.
The deterministic Workspace v0 governance boundary is separately advertised
as an evidence-backed research reference implementation, not as a sandbox or
production security system; its exact claim and trusted-host limits are in
docs/release/governed-membranes/DECISION.md and
docs/release/governed-membranes/LIMITS.md.

Troubleshooting

• opam: command not found: install opam with asdf using .tool-versions, or
  install a compatible opam manually.
• Dune cannot find packages: run eval "$(opam env)" in this shell, then
  reinstall deps with opam install --deps-only . --with-test --with-dev-setup --with-doc -y.
• jacquard cannot find names from the prelude: set JACQUARD_PRELUDE=$PWD/prelude or
  run through Dune from the repo root.
• Formatting changed files: run opam exec -- dune fmt, inspect the diff, and
  commit the formatting changes if they are intended.
• Release reproduction writes generated evidence under .scratch/release/0.2/
  by default. Set JACQUARD_RELEASE_OUT to use another disposable output path.