effect-log
Semantic side-effect tracking for AI agents.
When an AI agent crashes mid-task, what happens on restart? Without effect-log, irreversible actions (emails, payments, deployments) get repeated. With effect-log, the system knows what completed and what didn't — it returns sealed results for finished work and resumes from where it left off.
The Core Idea
Every tool declares its effect kind at registration time. This drives all recovery behavior:
| EffectKind | Recovery (completed) | Recovery (crashed) |
|---|---|---|
| ReadOnly | Replay for fresh data | Replay safely |
| IdempotentWrite | Return sealed result | Replay with same key |
| Compensatable | Return sealed result | Compensate, then replay |
| IrreversibleWrite | Return sealed result | Escalate to human |
| ReadThenWrite | Return sealed result | Escalate to human |
Quick Start
from effect_log import EffectKind, EffectLog, ToolDef
def send_email(args):
return smtp.send(args["to"], args["subject"], args["body"])
tools = [
ToolDef("read_file", EffectKind.ReadOnly, read_file),
ToolDef("send_email", EffectKind.IrreversibleWrite, send_email),
ToolDef("upsert", EffectKind.IdempotentWrite, upsert_record),
]
log = EffectLog(execution_id="task-001", tools=tools, storage="sqlite:///effects.db")
log.execute("read_file", {"path": "/tmp/report.csv"})
log.execute("send_email", {"to": "ceo@co.com", "subject": "Report", "body": "..."})
log.execute("upsert", {"id": "report-001", "data": data})
Recovery — just add recover=True, re-run the same steps:
log = EffectLog(execution_id="task-001", tools=tools, storage="sqlite:///effects.db", recover=True)
log.execute("read_file", {"path": "/tmp/report.csv"}) # Replayed (fresh data)
log.execute("send_email", {"to": "ceo@co.com", ...}) # Sealed — NOT re-sent
log.execute("upsert", {"id": "report-001", ...}) # Replayed (idempotent)
Framework Integration
Built-in middleware for major agent frameworks:
| Framework | Middleware | Entry Point |
|---|---|---|
| LangGraph | effect_log.middleware.langgraph |
EffectLogToolNode, effect_logged_tools |
| OpenAI Agents SDK | effect_log.middleware.openai_agents |
effect_logged_agent, wrap_function_tool |
| CrewAI | effect_log.middleware.crewai |
effect_logged_crew, effect_logged_tool |
See examples/ for runnable demos:
crash_recovery.py— Core crash recovery demo (Phase 1 milestone)langgraph_integration.py— LangGraph ToolNode + tool wrappingopenai_agents_integration.py— OpenAI Agents SDK wrappingcrewai_integration.py— CrewAI tool + crew wrapping
How It Works
A write-ahead log with two record types:
- Intent — written before execution (tool name, effect kind, input, cursor)
- Completion — written after execution (outcome, sealed response)
Intent without completion = crash detected. The recovery engine uses the effect kind to decide: replay, return sealed, compensate, or escalate.
Building from Source
# Rust
cargo build --release
cargo test --workspace --all-features
# Python
cd bindings/python
pip install maturin
maturin develop --all-features
pytest tests/ -v
Roadmap
- Core library — WAL engine, recovery engine, SQLite + in-memory backends
- Python bindings — PyO3 + maturin
- Framework middleware — LangGraph, OpenAI Agents SDK, CrewAI
- TypeScript bindings — napi-rs, Vercel AI SDK
- Additional backends — RocksDB, S3, Restate journal
- Auto-classification — infer effect kind from HTTP methods / API metadata
Inspiration
The idea behind this project was inspired by a blog post from Guanlan Dai. He introduced the concepts of effect log and semantic correctness.
License
Apache-2.0
1 maintainer
Metadata
1 maintainer
Release files for ai-effectlog 0.1.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | |
|---|---|---|---|---|
| ai_effectlog-0.1.0.tar.gz | 41.4 kB | With provenance | Details |
Built distributions (wheels)
Total release size: 14.6 MB
Release files / ai_effectlog-0.1.0.tar.gz
| Download URL | ai_effectlog-0.1.0.tar.gz |
|---|---|
| Size | 41.4 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
3fe20c1c633b070ef3e573b95a1260be68fb398e6473a9201ac21f48727172b7
|
|
BLAKE2b-256 checksum How to use checksums |
1b53ac072b5d0d99625d8dce9dee7111a70f22668d5deb987ee6acbd0538ea67
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Mar 15, 2026.
Transparency logRelease files / ai_effectlog-0.1.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
| Download URL | ai_effectlog-0.1.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl |
|---|---|
| Size | 1.6 MB |
| Tags | CPython 3.13 Linux glibc 2.17+ ARM64 |
|
SHA-256 checksum How to use checksums |
2383ee19431e3da47113cfdf56dfc52d2856e0265e1588f98fed7a1e34415fc0
|
|
BLAKE2b-256 checksum How to use checksums |
8195378d9fedbe37229380d3a200be401255a54d8e48ed04f58d297672ab08d6
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Mar 15, 2026.
Transparency logRelease files / ai_effectlog-0.1.0-cp312-cp312-win_amd64.whl
| Download URL | ai_effectlog-0.1.0-cp312-cp312-win_amd64.whl |
|---|---|
| Size | 1.5 MB |
| Tags | CPython 3.12 Windows x86-64 |
|
SHA-256 checksum How to use checksums |
66a5e203c4a4ade8d58ddbbadd37c2fc0f64919238b766e67dadf372e7dd9ddb
|
|
BLAKE2b-256 checksum How to use checksums |
4d251eadff93db93e8c424f088b2dedcaf8d08a46f3e9d7abfb4a4ca9476f52a
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Mar 15, 2026.
Transparency logRelease files / ai_effectlog-0.1.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
| Download URL | ai_effectlog-0.1.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl |
|---|---|
| Size | 1.6 MB |
| Tags | CPython 3.12 Linux glibc 2.17+ ARM64 |
|
SHA-256 checksum How to use checksums |
655b2e73f8ef6a75eb9ad810fc7aa474d65c030ca6654e2c8020049c35c6e0e3
|
|
BLAKE2b-256 checksum How to use checksums |
0ca18333d628a7bb4d8cc3af8ae8de93326e1a148d082841ec939d460cbbf038
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Mar 15, 2026.
Transparency logRelease files / ai_effectlog-0.1.0-cp312-cp312-macosx_11_0_arm64.whl
| Download URL | ai_effectlog-0.1.0-cp312-cp312-macosx_11_0_arm64.whl |
|---|---|
| Size | 1.7 MB |
| Tags | CPython 3.12 macOS 11.0+ ARM64 |
|
SHA-256 checksum How to use checksums |
3a126a3e28371e393030eb1bec6bb8e7eef0aa699f2bd703a893553af29bb97f
|
|
BLAKE2b-256 checksum How to use checksums |
4fb556a727c1616152dd831f36d802d3aa0757d777e2c5fd94a369fe5a6c7f8e
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Mar 15, 2026.
Transparency logRelease files / ai_effectlog-0.1.0-cp312-cp312-macosx_10_12_x86_64.whl
| Download URL | ai_effectlog-0.1.0-cp312-cp312-macosx_10_12_x86_64.whl |
|---|---|
| Size | 1.7 MB |
| Tags | CPython 3.12 macOS 10.12+ x86-64 |
|
SHA-256 checksum How to use checksums |
66bc7136a71db1f989762b6e04f881882f8ec2ef9ee4136a1be0bc22c0da3a62
|
|
BLAKE2b-256 checksum How to use checksums |
60b8e1fe1390f5c2cd09ffdd12229764065f68c10e8b11b313e91867bd8f8341
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Mar 15, 2026.
Transparency logRelease files / ai_effectlog-0.1.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
| Download URL | ai_effectlog-0.1.0-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl |
|---|---|
| Size | 1.6 MB |
| Tags | CPython 3.11 Linux glibc 2.17+ ARM64 |
|
SHA-256 checksum How to use checksums |
82fbe4483409a74d0b9fd9da508bc6b42b58588fe2f7b48e1c0beae879426990
|
|
BLAKE2b-256 checksum How to use checksums |
ef4f26130f8db5abaa21317fcce48d5cb2f21cbb8037c626c03468401b1b950d
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Mar 15, 2026.
Transparency logRelease files / ai_effectlog-0.1.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
| Download URL | ai_effectlog-0.1.0-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl |
|---|---|
| Size | 1.6 MB |
| Tags | CPython 3.10 Linux glibc 2.17+ ARM64 |
|
SHA-256 checksum How to use checksums |
d80f418a0fba704a80fa1e4a73c8aa4e3ccd4e018ff2076f2665bd1ca38c1ae2
|
|
BLAKE2b-256 checksum How to use checksums |
986b3183824ffba0004512cd4824d9ed367bf9741bb63da21ecae55db7ace305
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Mar 15, 2026.
Transparency logRelease files / ai_effectlog-0.1.0-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
| Download URL | ai_effectlog-0.1.0-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl |
|---|---|
| Size | 1.6 MB |
| Tags | CPython 3.9 Linux glibc 2.17+ ARM64 |
|
SHA-256 checksum How to use checksums |
1a1dd708ed74de3cf4eacc97286f48a5c83deca50d151761440ab2e5f104471d
|
|
BLAKE2b-256 checksum How to use checksums |
9e6a82f4a8c2d1f4987fcee43e18f54ab4c465f1694587909e57c44769d91cd4
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Mar 15, 2026.
Transparency logRelease files / ai_effectlog-0.1.0-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
| Download URL | ai_effectlog-0.1.0-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl |
|---|---|
| Size | 1.9 MB |
| Tags | CPython 3.8 Linux glibc 2.17+ x86-64 |
|
SHA-256 checksum How to use checksums |
33f689768483d3ddffff1d25ac25abce06354948fc18cae9886d12103db42e5d
|
|
BLAKE2b-256 checksum How to use checksums |
d2c7712a7ddc39b357b8c570b0cb205b3c34cf0159a6cd2c3ab4d2ee9815d530
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Mar 15, 2026.
Transparency logSecurity for ai-effectlog 0.1.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
All 10 files in this release have been uploaded with provenance metadata.
Attestation for ai_effectlog-0.1.0.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Mar 15, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.
