// STRYKE-AWS — ENGINEERING REPORT

>_EXECUTIVE SUMMARY

stryke-aws is one of the opt-in connector packages in the stryke ecosystem. AWS client for stryke across 17 services — S3, DynamoDB, SQS, Lambda, STS, SNS, SSM Parameter Store, Secrets Manager, SES, CloudWatch, CloudWatch Logs, EC2, KMS, IAM, Kinesis, ECR, and Step Functions. Opt-in package, kept out of the stryke core binary so the daily-driver install stays slim. The 17 SDKs combined are too much to bake into stryke core, so they ship as a Rust cdylib that stryke dlopens in-process on first use AWS.

The official aws-sdk-rust crates pull in tokio, hyper, rustls, and a fat chain of smithy / signing / endpoint-resolution support code. Ships once as an opt-in package. Credentials and region come from the standard AWS chain (env vars, ~/.aws/config|credentials, IMDS) — same as the aws CLI.

~ARCHITECTURE

In-process cdylib design: the stryke side is a thin .stk wrapper that calls FFI symbols on the dlopen'd library. No subprocess, no pipe — calls return on the same thread that made them. The v1 helper-binary fork-per-call model was replaced in v0.2.0.

@SERVICE COVERAGE

Each row is a stryke module (use AWS::<Service>) and the operations it currently wraps. The full surface lives in the Docs API reference.

Flat extras on base use AWS: s3_copy_object, s3_delete_objects (≤1000), ddb_update_item, ddb_batch_get_item (≤100), ddb_batch_write_item (puts ≤25). Pure no-credential helpers (ARN / S3-URI / endpoint parsing + validation) are also on base AWS — see the Docs pure-helpers table.

!DEFERRED OPERATIONS

These wrappers exist but die with a message naming the cdylib export that needs adding. Intentionally not yet wired so the cdylib surface stays auditable against stryke.toml's [ffi] exports table.

$WHY OPT-IN (NOT BUILTIN)

The official aws-sdk-rust crates pull in tokio, hyper, rustls, and a fat chain of smithy / signing / endpoint-resolution support code. The 17 SDKs combined are way too much to bake into stryke core. This package ships them once, opt-in.

The trade-off is intentional. The core stryke binary stays slim precisely because each connector ships separately. Daily-driver work (one-liners, awk replacement, data scripting) doesn't need AWS SDKs, MongoDB drivers, or Spark Connect bindings linked in. Core stryke is never linked against this package's deps.

&FFI ENVELOPE

JSON in / JSON out across the FFI boundary. Each AWS::* wrapper serializes its args to a JSON dict and calls the matching aws__* symbol resolved out of libstryke_aws.{dylib,so} (declared in stryke.toml's [ffi] exports); the cdylib returns a JSON CString the wrapper parses. The returned CString is freed via the cdylib-exported stryke_free_cstring, wired automatically by rust_ffi::load_cdylib.

# request shape (built by the .stk wrapper)
{"verb": "<op>", "...": ...}

# response shape
{"result": ...}            # success — per-fn shape
{"error": "<msg>"}        # failure — wrapper die()s with the message

Persistent state: a shared tokio runtime + an aws_config::SdkConfig cache per region held in OnceCell — no fork-per-call, no full IMDS/SSO/env creds chain on each call.

/SCOPE

See the README's "Why this is a package" + "API reference" sections for the authoritative scope. The package is intentionally narrower than its underlying SDKs — the goal is "useful from a shell pipeline", not "complete API coverage". DynamoDB items are plain JSON in / out (no AttributeValue wrappers); S3 put is buffered. Roadmap-later items: deferred ops above, typed-set passthrough, optimistic-locking helpers, and streaming multipart upload.

#PROJECT METADATA