agentverity 0.2.0__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- agentverity-0.2.0/.github/workflows/ci.yml +105 -0
- agentverity-0.2.0/.github/workflows/release.yml +55 -0
- agentverity-0.2.0/.gitignore +8 -0
- agentverity-0.2.0/CHANGELOG.md +99 -0
- agentverity-0.2.0/CONTRIBUTING.md +32 -0
- agentverity-0.2.0/DESIGN.md +121 -0
- agentverity-0.2.0/LICENSE +202 -0
- agentverity-0.2.0/PKG-INFO +401 -0
- agentverity-0.2.0/README.md +367 -0
- agentverity-0.2.0/RELEASING.md +59 -0
- agentverity-0.2.0/agentverity/__init__.py +53 -0
- agentverity-0.2.0/agentverity/adapters/__init__.py +14 -0
- agentverity-0.2.0/agentverity/adapters/callable_adapter.py +43 -0
- agentverity-0.2.0/agentverity/adapters/strands.py +100 -0
- agentverity-0.2.0/agentverity/blindness.py +164 -0
- agentverity-0.2.0/agentverity/cli.py +113 -0
- agentverity-0.2.0/agentverity/meter.py +204 -0
- agentverity-0.2.0/agentverity/observation.py +64 -0
- agentverity-0.2.0/agentverity/py.typed +0 -0
- agentverity-0.2.0/agentverity/relations.py +142 -0
- agentverity-0.2.0/agentverity/runner.py +376 -0
- agentverity-0.2.0/examples/bugfix_pipeline.py +122 -0
- agentverity-0.2.0/examples/strands_example.py +61 -0
- agentverity-0.2.0/examples/toy_agent.py +102 -0
- agentverity-0.2.0/pyproject.toml +68 -0
- agentverity-0.2.0/tests/test_blindness.py +112 -0
- agentverity-0.2.0/tests/test_callable_adapter.py +64 -0
- agentverity-0.2.0/tests/test_cli.py +72 -0
- agentverity-0.2.0/tests/test_meter.py +123 -0
- agentverity-0.2.0/tests/test_observation.py +30 -0
- agentverity-0.2.0/tests/test_public_api.py +55 -0
- agentverity-0.2.0/tests/test_relations.py +102 -0
- agentverity-0.2.0/tests/test_runner.py +295 -0
- agentverity-0.2.0/tests/test_strands_adapter.py +101 -0
|
@@ -0,0 +1,105 @@
|
|
|
1
|
+
name: CI
|
|
2
|
+
|
|
3
|
+
on:
|
|
4
|
+
pull_request:
|
|
5
|
+
push:
|
|
6
|
+
branches:
|
|
7
|
+
- main
|
|
8
|
+
|
|
9
|
+
permissions:
|
|
10
|
+
contents: read
|
|
11
|
+
|
|
12
|
+
concurrency:
|
|
13
|
+
group: ci-${{ github.workflow }}-${{ github.ref }}
|
|
14
|
+
cancel-in-progress: true
|
|
15
|
+
|
|
16
|
+
jobs:
|
|
17
|
+
test:
|
|
18
|
+
name: Python ${{ matrix.python-version }}
|
|
19
|
+
runs-on: ubuntu-latest
|
|
20
|
+
strategy:
|
|
21
|
+
fail-fast: false
|
|
22
|
+
matrix:
|
|
23
|
+
python-version:
|
|
24
|
+
- "3.10"
|
|
25
|
+
- "3.11"
|
|
26
|
+
- "3.12"
|
|
27
|
+
- "3.13"
|
|
28
|
+
- "3.14"
|
|
29
|
+
|
|
30
|
+
steps:
|
|
31
|
+
- uses: actions/checkout@v4
|
|
32
|
+
- uses: actions/setup-python@v5
|
|
33
|
+
with:
|
|
34
|
+
python-version: ${{ matrix.python-version }}
|
|
35
|
+
cache: pip
|
|
36
|
+
- name: Install
|
|
37
|
+
run: python -m pip install -e ".[dev]"
|
|
38
|
+
- name: Test
|
|
39
|
+
run: python -m pytest -q
|
|
40
|
+
|
|
41
|
+
lint:
|
|
42
|
+
name: Lint
|
|
43
|
+
runs-on: ubuntu-latest
|
|
44
|
+
steps:
|
|
45
|
+
- uses: actions/checkout@v4
|
|
46
|
+
- uses: actions/setup-python@v5
|
|
47
|
+
with:
|
|
48
|
+
python-version: "3.12"
|
|
49
|
+
cache: pip
|
|
50
|
+
- name: Install
|
|
51
|
+
run: python -m pip install -e ".[dev]"
|
|
52
|
+
- name: Lint
|
|
53
|
+
run: ruff check .
|
|
54
|
+
|
|
55
|
+
package:
|
|
56
|
+
name: Package
|
|
57
|
+
runs-on: ubuntu-latest
|
|
58
|
+
steps:
|
|
59
|
+
- uses: actions/checkout@v4
|
|
60
|
+
- uses: actions/setup-python@v5
|
|
61
|
+
with:
|
|
62
|
+
python-version: "3.12"
|
|
63
|
+
cache: pip
|
|
64
|
+
- name: Install build tools
|
|
65
|
+
run: python -m pip install build twine
|
|
66
|
+
- name: Build distributions
|
|
67
|
+
run: python -m build
|
|
68
|
+
- name: Check distribution metadata
|
|
69
|
+
run: python -m twine check dist/*
|
|
70
|
+
- name: Install wheel
|
|
71
|
+
run: python -m pip install --force-reinstall dist/*.whl
|
|
72
|
+
- name: Smoke-test installed package
|
|
73
|
+
run: |
|
|
74
|
+
export PACKAGE_VERSION
|
|
75
|
+
PACKAGE_VERSION="$(python -c "import tomllib; print(tomllib.load(open('pyproject.toml', 'rb'))['project']['version'])")"
|
|
76
|
+
cd "${RUNNER_TEMP}"
|
|
77
|
+
python -c "import importlib.metadata as m, os; import agentverity; assert m.version('agentverity') == os.environ['PACKAGE_VERSION']"
|
|
78
|
+
agentverity --help
|
|
79
|
+
|
|
80
|
+
gate:
|
|
81
|
+
name: CI gate
|
|
82
|
+
runs-on: ubuntu-latest
|
|
83
|
+
needs:
|
|
84
|
+
- test
|
|
85
|
+
- lint
|
|
86
|
+
- package
|
|
87
|
+
# Runs even when a needed job fails, so the gate reports a real failure
|
|
88
|
+
# instead of being skipped and leaving the required check pending forever.
|
|
89
|
+
if: always()
|
|
90
|
+
steps:
|
|
91
|
+
- name: Require every upstream job to succeed
|
|
92
|
+
env:
|
|
93
|
+
TEST_RESULT: ${{ needs.test.result }}
|
|
94
|
+
LINT_RESULT: ${{ needs.lint.result }}
|
|
95
|
+
PACKAGE_RESULT: ${{ needs.package.result }}
|
|
96
|
+
run: |
|
|
97
|
+
echo "test: ${TEST_RESULT}"
|
|
98
|
+
echo "lint: ${LINT_RESULT}"
|
|
99
|
+
echo "package: ${PACKAGE_RESULT}"
|
|
100
|
+
for result in "${TEST_RESULT}" "${LINT_RESULT}" "${PACKAGE_RESULT}"; do
|
|
101
|
+
if [ "${result}" != "success" ]; then
|
|
102
|
+
echo "::error::a required job did not succeed"
|
|
103
|
+
exit 1
|
|
104
|
+
fi
|
|
105
|
+
done
|
|
@@ -0,0 +1,55 @@
|
|
|
1
|
+
name: Release
|
|
2
|
+
|
|
3
|
+
on:
|
|
4
|
+
release:
|
|
5
|
+
types:
|
|
6
|
+
- published
|
|
7
|
+
|
|
8
|
+
permissions:
|
|
9
|
+
contents: read
|
|
10
|
+
|
|
11
|
+
jobs:
|
|
12
|
+
build:
|
|
13
|
+
name: Build distributions
|
|
14
|
+
runs-on: ubuntu-latest
|
|
15
|
+
steps:
|
|
16
|
+
- uses: actions/checkout@v4
|
|
17
|
+
with:
|
|
18
|
+
ref: ${{ github.event.release.tag_name }}
|
|
19
|
+
- uses: actions/setup-python@v5
|
|
20
|
+
with:
|
|
21
|
+
python-version: "3.12"
|
|
22
|
+
cache: pip
|
|
23
|
+
- name: Verify release tag
|
|
24
|
+
env:
|
|
25
|
+
RELEASE_TAG: ${{ github.event.release.tag_name }}
|
|
26
|
+
run: |
|
|
27
|
+
PACKAGE_VERSION="$(python -c "import tomllib; print(tomllib.load(open('pyproject.toml', 'rb'))['project']['version'])")"
|
|
28
|
+
test "${RELEASE_TAG}" = "v${PACKAGE_VERSION}"
|
|
29
|
+
- name: Install build tools
|
|
30
|
+
run: python -m pip install build twine
|
|
31
|
+
- name: Build distributions
|
|
32
|
+
run: python -m build
|
|
33
|
+
- name: Check distribution metadata
|
|
34
|
+
run: python -m twine check dist/*
|
|
35
|
+
- uses: actions/upload-artifact@v4
|
|
36
|
+
with:
|
|
37
|
+
name: python-package-distributions
|
|
38
|
+
path: dist/
|
|
39
|
+
|
|
40
|
+
publish:
|
|
41
|
+
name: Publish to PyPI
|
|
42
|
+
needs: build
|
|
43
|
+
runs-on: ubuntu-latest
|
|
44
|
+
environment:
|
|
45
|
+
name: pypi
|
|
46
|
+
url: https://pypi.org/p/agentverity
|
|
47
|
+
permissions:
|
|
48
|
+
id-token: write
|
|
49
|
+
steps:
|
|
50
|
+
- uses: actions/download-artifact@v4
|
|
51
|
+
with:
|
|
52
|
+
name: python-package-distributions
|
|
53
|
+
path: dist/
|
|
54
|
+
- name: Publish distributions
|
|
55
|
+
uses: pypa/gh-action-pypi-publish@release/v1
|
|
@@ -0,0 +1,99 @@
|
|
|
1
|
+
# Changelog
|
|
2
|
+
|
|
3
|
+
All notable changes to this project are documented in this file.
|
|
4
|
+
|
|
5
|
+
The format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/).
|
|
6
|
+
This project follows [Semantic Versioning](https://semver.org/) once it
|
|
7
|
+
reaches 1.0.0; before that, minor versions may include breaking changes.
|
|
8
|
+
|
|
9
|
+
## [Unreleased]
|
|
10
|
+
|
|
11
|
+
## [0.2.0] - 2026-07-25
|
|
12
|
+
|
|
13
|
+
### Added
|
|
14
|
+
|
|
15
|
+
- Distribution build and clean-install validation in CI.
|
|
16
|
+
- PyPI Trusted Publishing workflow and a documented maintainer release
|
|
17
|
+
procedure.
|
|
18
|
+
- `RelationResult.skipped`, `.exercised`, and `.is_vacuous`, plus
|
|
19
|
+
`RunResult.vacuous_relations`, so a relation whose transform never changed
|
|
20
|
+
an input is reported instead of counted as a pass.
|
|
21
|
+
- `RunConfig.reuse_unchanged_calls` (default on) and
|
|
22
|
+
`agentverity.blindness.score`, which scores observations another phase has
|
|
23
|
+
already collected.
|
|
24
|
+
- A reproducible supervisor-pattern example that contrasts a blind triage step
|
|
25
|
+
with a stochastic full pipeline.
|
|
26
|
+
- A pull-request-only contribution workflow and CI across Python 3.10 to 3.14.
|
|
27
|
+
|
|
28
|
+
### Changed
|
|
29
|
+
|
|
30
|
+
- The verdict-stochasticity interval now uses disjoint repeat pairs. The
|
|
31
|
+
earlier all-pairs calculation reused the same calls across comparisons and
|
|
32
|
+
overstated the effective sample size.
|
|
33
|
+
- The runner now executes both diagnostics before any metamorphic relation.
|
|
34
|
+
- The built-in accent/whitespace transform is named
|
|
35
|
+
`normalisation-invariance`, matching what it actually does.
|
|
36
|
+
- `suite_is_meaningful` now reports whether relation passes may be vacuous
|
|
37
|
+
under the skew scan. Meter calls remain separate oracle guidance.
|
|
38
|
+
- Public novelty claims now distinguish repeated-trial and calibration tools
|
|
39
|
+
from AgentVerity's narrower oracle-selection and verdict-skew diagnostics.
|
|
40
|
+
- `violation_rate` is now measured over exercised pairs rather than over all
|
|
41
|
+
inputs, so inputs the transform left untouched no longer dilute it.
|
|
42
|
+
- A run reuses the meter's first draw per input for the blindness scan and for
|
|
43
|
+
each relation's source side. Agent calls drop from `n * (k + 1 + 2r)` to
|
|
44
|
+
`n * (k + r)`, halving them on the default configuration.
|
|
45
|
+
|
|
46
|
+
### Fixed
|
|
47
|
+
|
|
48
|
+
- Empty probe sets and invalid meter/blindness thresholds now fail with clear
|
|
49
|
+
`ValueError`s.
|
|
50
|
+
- The CLI no longer invokes an agent with a hidden probe input before the
|
|
51
|
+
configured suite, avoiding side effects and wasted model calls.
|
|
52
|
+
- Passing `relations=[]` now runs no relations instead of restoring built-ins.
|
|
53
|
+
- `normalisation-invariance` and `tool-selection-invariance` normalise accents
|
|
54
|
+
and whitespace, so on plain ASCII input the follow-up string was identical to
|
|
55
|
+
the source. Both relations reported a perfect pass without the agent ever
|
|
56
|
+
being asked a different question. Those inputs are now skipped, the rate
|
|
57
|
+
reads `n/a`, and the report names the relation as not exercised.
|
|
58
|
+
|
|
59
|
+
## [0.1.0] - 2026-07-24
|
|
60
|
+
|
|
61
|
+
Initial public release.
|
|
62
|
+
|
|
63
|
+
### Added
|
|
64
|
+
|
|
65
|
+
- Verdict-stochasticity meter (`agentverity.meter`): a Wilson-CI-backed
|
|
66
|
+
tri-state call (`verdict-stochastic` / `verdict-deterministic` /
|
|
67
|
+
`undecided`) on whether an agent's decision is stable across identical
|
|
68
|
+
reruns, refusing to label an underpowered probe "deterministic."
|
|
69
|
+
- Constant-gate-blindness detector (`agentverity.blindness`): flags when
|
|
70
|
+
a passing test suite is trivially satisfied because the agent returns
|
|
71
|
+
a near-constant verdict regardless of input.
|
|
72
|
+
- Typed metamorphic relations (`agentverity.relations`): invariant,
|
|
73
|
+
monotone, and directional, with four built-in relations (paraphrase,
|
|
74
|
+
case, whitespace, and tool-selection invariance).
|
|
75
|
+
- Runner (`agentverity.runner`) orchestrating meter, then relations,
|
|
76
|
+
then blindness into a single diagnostics-first report.
|
|
77
|
+
- CLI (`agentverity run --agent module:func --inputs file.txt`).
|
|
78
|
+
- Callable adapter (zero dependencies) and an optional Strands adapter.
|
|
79
|
+
- 64 tests, all passing.
|
|
80
|
+
|
|
81
|
+
### Fixed (found during pre-release review, before this version shipped)
|
|
82
|
+
|
|
83
|
+
- `from_callable` was not exported from the top-level `agentverity`
|
|
84
|
+
package, only from `agentverity.adapters`, so the README's own
|
|
85
|
+
Quickstart (`from agentverity import run, from_callable`) raised
|
|
86
|
+
`ImportError` for anyone who copy-pasted it. Now exported at the top
|
|
87
|
+
level and covered by `tests/test_public_api.py` so the documented
|
|
88
|
+
import surface can't silently drift again.
|
|
89
|
+
- The Quickstart's shown output didn't match what the shown code
|
|
90
|
+
actually produces (stale pair counts, a tri-state call that wasn't
|
|
91
|
+
reachable with that input size). Replaced with output captured from
|
|
92
|
+
an actual run against the fixed code.
|
|
93
|
+
- Lint cleanup across the package: unused imports, a test asserting a
|
|
94
|
+
bare `Exception` narrowed to the specific `FrozenInstanceError` it's
|
|
95
|
+
actually checking for, missing trailing newlines.
|
|
96
|
+
|
|
97
|
+
[Unreleased]: https://github.com/mrwersa/agentverity/compare/v0.2.0...HEAD
|
|
98
|
+
[0.2.0]: https://github.com/mrwersa/agentverity/compare/v0.1.0...v0.2.0
|
|
99
|
+
[0.1.0]: https://github.com/mrwersa/agentverity/releases/tag/v0.1.0
|
|
@@ -0,0 +1,32 @@
|
|
|
1
|
+
# Contributing
|
|
2
|
+
|
|
3
|
+
AgentVerity uses a pull-request-only workflow for `main`.
|
|
4
|
+
|
|
5
|
+
1. Create a branch from the latest `main`:
|
|
6
|
+
|
|
7
|
+
```bash
|
|
8
|
+
git switch main
|
|
9
|
+
git pull --ff-only
|
|
10
|
+
git switch -c feature/short-description
|
|
11
|
+
```
|
|
12
|
+
|
|
13
|
+
2. Keep the change focused and add tests for behavioural changes.
|
|
14
|
+
|
|
15
|
+
3. Run the local quality gate:
|
|
16
|
+
|
|
17
|
+
```bash
|
|
18
|
+
python -m pytest -q
|
|
19
|
+
ruff check .
|
|
20
|
+
```
|
|
21
|
+
|
|
22
|
+
4. Push the branch and open a pull request:
|
|
23
|
+
|
|
24
|
+
```bash
|
|
25
|
+
git push -u origin feature/short-description
|
|
26
|
+
```
|
|
27
|
+
|
|
28
|
+
Direct pushes, force pushes, and deletion of `main` are blocked. Merge only
|
|
29
|
+
after CI passes and all review conversations are resolved.
|
|
30
|
+
|
|
31
|
+
Maintainers should follow [RELEASING.md](RELEASING.md) when publishing a
|
|
32
|
+
version.
|
|
@@ -0,0 +1,121 @@
|
|
|
1
|
+
# agentverity — design
|
|
2
|
+
|
|
3
|
+
A side project, NOT a research paper: an open-source framework for
|
|
4
|
+
**measure-first testing of non-deterministic LLM agents**. Created 2026-07-05.
|
|
5
|
+
Career artifact: "I build reusable tooling for testing agentic AI."
|
|
6
|
+
|
|
7
|
+
## Identity
|
|
8
|
+
|
|
9
|
+
**Name:** `agentverity` — "agent" + "verity" (truth). The tool tells you the
|
|
10
|
+
truth about whether your test suite is meaningful before you trust it.
|
|
11
|
+
|
|
12
|
+
**Wedge:** not "run metamorphic relations" but "diagnose whether your agent
|
|
13
|
+
test suite is lying to you" — suite-quality diagnostics for non-deterministic
|
|
14
|
+
agents. Compete on non-determinism honesty, NOT on breadth (lose to
|
|
15
|
+
DeepEval/promptfoo) or on the MR taxonomy (CheckList).
|
|
16
|
+
|
|
17
|
+
## Prior-art landscape (researched 2026-07-06)
|
|
18
|
+
|
|
19
|
+
| Tool | Stars | Non-determinism | Metamorphic | Suite-quality diagnostic | License |
|
|
20
|
+
|---|---|---|---|---|---|
|
|
21
|
+
| DeepEval | 16.7k | Repeated runs | No | No | Apache-2.0 |
|
|
22
|
+
| promptfoo | 22.9k | Repeated runs | No | Red-team coverage only | MIT |
|
|
23
|
+
| agentevals (langchain) | 636 | No | No | No | MIT |
|
|
24
|
+
| CheckList | 2k | Assumes deterministic | INV/DIR (owns it) | No | MIT |
|
|
25
|
+
| AgentAssay | paper | Yes (Wilson CIs) | Yes | Mutation testing | AGPL |
|
|
26
|
+
| agentrial | 17 | Yes (Wilson CIs) | No | No | MIT |
|
|
27
|
+
| **agentverity** | — | **Yes + measure-first** | **Yes (inherited)** | **Meter + blindness** | **Apache-2.0** |
|
|
28
|
+
|
|
29
|
+
**The three differentiators that survive the research:**
|
|
30
|
+
|
|
31
|
+
1. **Verdict-layer oracle selection.** Existing tools repeat tests, report
|
|
32
|
+
uncertainty, and in AgentAssay's case calibrate trial budgets. AgentVerity
|
|
33
|
+
asks whether the chosen categorical decision layer is stable enough that a
|
|
34
|
+
frozen baseline dominates MRs. The novelty claim is this oracle-selection
|
|
35
|
+
use, not awareness that agents are non-deterministic.
|
|
36
|
+
|
|
37
|
+
2. **Constant-gate-blindness detector.** A gate that returns `"allow"` on 96%
|
|
38
|
+
of a probe set can satisfy many invariance checks without exercising a
|
|
39
|
+
boundary. The detector flags the pass as potentially vacuous. This is a
|
|
40
|
+
suite-power warning, not a correctness judgement.
|
|
41
|
+
|
|
42
|
+
3. **Suite-quality diagnostic framing.** Not "run tests and report pass/fail"
|
|
43
|
+
but "tell you if your tests are meaningful before you trust them." The
|
|
44
|
+
report leads with the diagnostics, then per-relation results.
|
|
45
|
+
|
|
46
|
+
**Borrowed, established (cite in README):**
|
|
47
|
+
- **Metamorphic relations** (Chen et al. 1998): assert a law between two runs
|
|
48
|
+
(transform input, check the outputs relate correctly) — no golden output
|
|
49
|
+
needed. The escape from the oracle problem for non-deterministic systems.
|
|
50
|
+
- **Semantic-invariance transforms** (CheckList / LLMORPH): normalisation,
|
|
51
|
+
casing, whitespace.
|
|
52
|
+
- **Wilson CIs** (AgentAssay, agentrial use them for pass-rate CIs; we use
|
|
53
|
+
them for verdict-stability certification — same primitive, different use).
|
|
54
|
+
|
|
55
|
+
**Apache-2.0** is an adoption edge over AgentAssay's AGPL.
|
|
56
|
+
|
|
57
|
+
---
|
|
58
|
+
|
|
59
|
+
## 1. What it is (one line)
|
|
60
|
+
|
|
61
|
+
`agentverity` runs two diagnostics before any test relation: does the agent's
|
|
62
|
+
verdict flip across identical reruns (meter), and does the agent return a
|
|
63
|
+
near-constant verdict across a diverse input set (blindness). The first guides
|
|
64
|
+
oracle selection. The second warns when green relation results may be vacuous.
|
|
65
|
+
|
|
66
|
+
## 2. Architecture (three layers)
|
|
67
|
+
|
|
68
|
+
```
|
|
69
|
+
your agent (Strands / LangGraph / any callable)
|
|
70
|
+
| adapter: normalise to Observation
|
|
71
|
+
v
|
|
72
|
+
agentverity.core
|
|
73
|
+
|
|
|
74
|
+
├── meter.py — verdict-stochasticity meter (headline #1)
|
|
75
|
+
├── blindness.py — constant-gate-blindness detector (headline #2)
|
|
76
|
+
├── relations.py — typed metamorphic relations (the vehicle)
|
|
77
|
+
├── runner.py — orchestrates meter -> blindness -> relations
|
|
78
|
+
└── cli.py — `agentverity run` entry point
|
|
79
|
+
```
|
|
80
|
+
|
|
81
|
+
### 3a. Adapter layer (`agentverity/adapters/`)
|
|
82
|
+
Turns a real agent into a uniform `run(input) -> Observation`. `Observation`
|
|
83
|
+
carries what relations can assert over:
|
|
84
|
+
- `text` — final response string
|
|
85
|
+
- `verdict`— an optional extracted categorical decision
|
|
86
|
+
- `tools` — the ordered list of tool names the agent called (the trajectory)
|
|
87
|
+
- `raw` — the underlying result object, for custom relations
|
|
88
|
+
|
|
89
|
+
Adapters:
|
|
90
|
+
- **Strands:** `Agent(prompt) -> AgentResult`. Adapter calls the agent, reads
|
|
91
|
+
the final message text and the tool-use blocks for `tools`.
|
|
92
|
+
- **LangGraph:** compiled graph `.invoke(state)` (planned).
|
|
93
|
+
- **callable:** any `fn(input) -> str | dict | Observation` for non-library agents.
|
|
94
|
+
Adapters are OPTIONAL imports — the core installs without any agent library.
|
|
95
|
+
|
|
96
|
+
### 3b. Core (`agentverity/`)
|
|
97
|
+
- `relations.py` — Relation = (name, type, transform, check). Built-in
|
|
98
|
+
catalogue: normalisation, casing, whitespace invariance (text-level);
|
|
99
|
+
tool-selection-invariance (agent-specific, our emphasis).
|
|
100
|
+
- `meter.py` — verdict-stochasticity meter. Tri-state call with Wilson CI.
|
|
101
|
+
- `blindness.py` — constant-gate-blindness detector. Skew scan + warning.
|
|
102
|
+
- `runner.py` — orchestrates meter -> blindness -> relations. Returns RunResult.
|
|
103
|
+
- `cli.py` — `agentverity run --agent module:func --inputs file.txt`.
|
|
104
|
+
|
|
105
|
+
## 3. Scope discipline (what it is NOT)
|
|
106
|
+
|
|
107
|
+
- Not a benchmark, not a leaderboard, not an LLM-judge scorer.
|
|
108
|
+
- Not correctness testing — it tests *relations*, and explicitly warns it cannot
|
|
109
|
+
tell a correct agent from an indifferent one (blindness detector).
|
|
110
|
+
- Not tied to any provider or the research programme's own gate.
|
|
111
|
+
- Zero dependency on the `mnem`/sibling-paper code. Fully standalone.
|
|
112
|
+
|
|
113
|
+
## 4. Status (2026-07-06)
|
|
114
|
+
|
|
115
|
+
- M1 core: DONE — observation, meter, blindness, relations, runner, CLI.
|
|
116
|
+
- M2 Strands adapter: DONE — adapter written, tested, worked example runs.
|
|
117
|
+
- M2 LangGraph adapter: PLANNED.
|
|
118
|
+
- M3 agent-specific relations: PLANNED (tool-selection-invariance is built-in;
|
|
119
|
+
more user-extensible relations to follow).
|
|
120
|
+
- M4 packaging: pyproject.toml written, README written, LICENSE added.
|
|
121
|
+
PyPI name `agentverity` verified free. Not yet published.
|
|
@@ -0,0 +1,202 @@
|
|
|
1
|
+
|
|
2
|
+
Apache License
|
|
3
|
+
Version 2.0, January 2004
|
|
4
|
+
http://www.apache.org/licenses/
|
|
5
|
+
|
|
6
|
+
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
|
|
7
|
+
|
|
8
|
+
1. Definitions.
|
|
9
|
+
|
|
10
|
+
"License" shall mean the terms and conditions for use, reproduction,
|
|
11
|
+
and distribution as defined by Sections 1 through 9 of this document.
|
|
12
|
+
|
|
13
|
+
"Licensor" shall mean the copyright owner or entity authorized by
|
|
14
|
+
the copyright owner that is granting the License.
|
|
15
|
+
|
|
16
|
+
"Legal Entity" shall mean the union of the acting entity and all
|
|
17
|
+
other entities that control, are controlled by, or are under common
|
|
18
|
+
control with that entity. For the purposes of this definition,
|
|
19
|
+
"control" means (i) the power, direct or indirect, to cause the
|
|
20
|
+
direction or management of such entity, whether by contract or
|
|
21
|
+
otherwise, or (ii) ownership of fifty percent (50%) or more of the
|
|
22
|
+
outstanding shares, or (iii) beneficial ownership of such entity.
|
|
23
|
+
|
|
24
|
+
"You" (or "Your") shall mean an individual or Legal Entity
|
|
25
|
+
exercising permissions granted by this License.
|
|
26
|
+
|
|
27
|
+
"Source" form shall mean the preferred form for making modifications,
|
|
28
|
+
including but not limited to software source code, documentation
|
|
29
|
+
source, and configuration files.
|
|
30
|
+
|
|
31
|
+
"Object" form shall mean any form resulting from mechanical
|
|
32
|
+
transformation or translation of a Source form, including but
|
|
33
|
+
not limited to compiled object code, generated documentation,
|
|
34
|
+
and conversions to other media types.
|
|
35
|
+
|
|
36
|
+
"Work" shall mean the work of authorship, whether in Source or
|
|
37
|
+
Object form, made available under the License, as indicated by a
|
|
38
|
+
copyright notice that is included in or attached to the work
|
|
39
|
+
(an example is provided in the Appendix below).
|
|
40
|
+
|
|
41
|
+
"Derivative Works" shall mean any work, whether in Source or Object
|
|
42
|
+
form, that is based on (or derived from) the Work and for which the
|
|
43
|
+
editorial revisions, annotations, elaborations, or other modifications
|
|
44
|
+
represent, as a whole, an original work of authorship. For the purposes
|
|
45
|
+
of this License, Derivative Works shall not include works that remain
|
|
46
|
+
separable from, or merely link (or bind by name) to the interfaces of,
|
|
47
|
+
the Work and Derivative Works thereof.
|
|
48
|
+
|
|
49
|
+
"Contribution" shall mean any work of authorship, including
|
|
50
|
+
the original version of the Work and any modifications or additions
|
|
51
|
+
to that Work or Derivative Works thereof, that is intentionally
|
|
52
|
+
submitted to Licensor for inclusion in the Work by the copyright owner
|
|
53
|
+
or by an individual or Legal Entity authorized to submit on behalf of
|
|
54
|
+
the copyright owner. For the purposes of this definition, "submitted"
|
|
55
|
+
means any form of electronic, verbal, or written communication sent
|
|
56
|
+
to the Licensor or its representatives, including but not limited to
|
|
57
|
+
communication on electronic mailing lists, source code control systems,
|
|
58
|
+
and issue tracking systems that are managed by, or on behalf of, the
|
|
59
|
+
Licensor for the purpose of discussing and improving the Work, but
|
|
60
|
+
excluding communication that is conspicuously marked or otherwise
|
|
61
|
+
designated in writing by the copyright owner as "Not a Contribution."
|
|
62
|
+
|
|
63
|
+
"Contributor" shall mean Licensor and any individual or Legal Entity
|
|
64
|
+
on behalf of whom a Contribution has been received by Licensor and
|
|
65
|
+
subsequently incorporated within the Work.
|
|
66
|
+
|
|
67
|
+
2. Grant of Copyright License. Subject to the terms and conditions of
|
|
68
|
+
this License, each Contributor hereby grants to You a perpetual,
|
|
69
|
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
|
70
|
+
copyright license to reproduce, prepare Derivative Works of,
|
|
71
|
+
publicly display, publicly perform, sublicense, and distribute the
|
|
72
|
+
Work and such Derivative Works in Source or Object form.
|
|
73
|
+
|
|
74
|
+
3. Grant of Patent License. Subject to the terms and conditions of
|
|
75
|
+
this License, each Contributor hereby grants to You a perpetual,
|
|
76
|
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
|
77
|
+
(except as stated in this section) patent license to make, have made,
|
|
78
|
+
use, offer to sell, sell, import, and otherwise transfer the Work,
|
|
79
|
+
where such license applies only to those patent claims licensable
|
|
80
|
+
by such Contributor that are necessarily infringed by their
|
|
81
|
+
Contribution(s) alone or by combination of their Contribution(s)
|
|
82
|
+
with the Work to which such Contribution(s) was submitted. If You
|
|
83
|
+
institute patent litigation against any entity (including a
|
|
84
|
+
cross-claim or counterclaim in a lawsuit) alleging that the Work
|
|
85
|
+
or a Contribution incorporated within the Work constitutes direct
|
|
86
|
+
or contributory patent infringement, then any patent licenses
|
|
87
|
+
granted to You under this License for that Work shall terminate
|
|
88
|
+
as of the date such litigation is filed.
|
|
89
|
+
|
|
90
|
+
4. Redistribution. You may reproduce and distribute copies of the
|
|
91
|
+
Work or Derivative Works thereof in any medium, with or without
|
|
92
|
+
modifications, and in Source or Object form, provided that You
|
|
93
|
+
meet the following conditions:
|
|
94
|
+
|
|
95
|
+
(a) You must give any other recipients of the Work or
|
|
96
|
+
Derivative Works a copy of this License; and
|
|
97
|
+
|
|
98
|
+
(b) You must cause any modified files to carry prominent notices
|
|
99
|
+
stating that You changed the files; and
|
|
100
|
+
|
|
101
|
+
(c) You must retain, in the Source form of any Derivative Works
|
|
102
|
+
that You distribute, all copyright, patent, trademark, and
|
|
103
|
+
attribution notices from the Source form of the Work,
|
|
104
|
+
excluding those notices that do not pertain to any part of
|
|
105
|
+
the Derivative Works; and
|
|
106
|
+
|
|
107
|
+
(d) If the Work includes a "NOTICE" text file as part of its
|
|
108
|
+
distribution, then any Derivative Works that You distribute must
|
|
109
|
+
include a readable copy of the attribution notices contained
|
|
110
|
+
within such NOTICE file, excluding those notices that do not
|
|
111
|
+
pertain to any part of the Derivative Works, in at least one
|
|
112
|
+
of the following places: within a NOTICE text file distributed
|
|
113
|
+
as part of the Derivative Works; within the Source form or
|
|
114
|
+
documentation, if provided along with the Derivative Works; or,
|
|
115
|
+
within a display generated by the Derivative Works, if and
|
|
116
|
+
wherever such third-party notices normally appear. The contents
|
|
117
|
+
of the NOTICE file are for informational purposes only and
|
|
118
|
+
do not modify the License. You may add Your own attribution
|
|
119
|
+
notices within Derivative Works that You distribute, alongside
|
|
120
|
+
or as an addendum to the NOTICE text from the Work, provided
|
|
121
|
+
that such additional attribution notices cannot be construed
|
|
122
|
+
as modifying the License.
|
|
123
|
+
|
|
124
|
+
You may add Your own copyright statement to Your modifications and
|
|
125
|
+
may provide additional or different license terms and conditions
|
|
126
|
+
for use, reproduction, or distribution of Your modifications, or
|
|
127
|
+
for any such Derivative Works as a whole, provided Your use,
|
|
128
|
+
reproduction, and distribution of the Work otherwise complies with
|
|
129
|
+
the conditions stated in this License.
|
|
130
|
+
|
|
131
|
+
5. Submission of Contributions. Unless You explicitly state otherwise,
|
|
132
|
+
any Contribution intentionally submitted for inclusion in the Work
|
|
133
|
+
by You to the Licensor shall be under the terms and conditions of
|
|
134
|
+
this License, without any additional terms or conditions.
|
|
135
|
+
Notwithstanding the above, nothing herein shall supersede or modify
|
|
136
|
+
the terms of any separate license agreement you may have executed
|
|
137
|
+
with Licensor regarding such Contributions.
|
|
138
|
+
|
|
139
|
+
6. Trademarks. This License does not grant permission to use the trade
|
|
140
|
+
names, trademarks, service marks, or product names of the Licensor,
|
|
141
|
+
except as required for reasonable and customary use in describing the
|
|
142
|
+
origin of the Work and reproducing the content of the NOTICE file.
|
|
143
|
+
|
|
144
|
+
7. Disclaimer of Warranty. Unless required by applicable law or
|
|
145
|
+
agreed to in writing, Licensor provides the Work (and each
|
|
146
|
+
Contributor provides its Contributions) on an "AS IS" BASIS,
|
|
147
|
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
|
|
148
|
+
implied, including, without limitation, any warranties or conditions
|
|
149
|
+
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
|
|
150
|
+
PARTICULAR PURPOSE. You are solely responsible for determining the
|
|
151
|
+
appropriateness of using or redistributing the Work and assume any
|
|
152
|
+
risks associated with Your exercise of permissions under this License.
|
|
153
|
+
|
|
154
|
+
8. Limitation of Liability. In no event and under no legal theory,
|
|
155
|
+
whether in tort (including negligence), contract, or otherwise,
|
|
156
|
+
unless required by applicable law (such as deliberate and grossly
|
|
157
|
+
negligent acts) or agreed to in writing, shall any Contributor be
|
|
158
|
+
liable to You for damages, including any direct, indirect, special,
|
|
159
|
+
incidental, or consequential damages of any character arising as a
|
|
160
|
+
result of this License or out of the use or inability to use the
|
|
161
|
+
Work (including but not limited to damages for loss of goodwill,
|
|
162
|
+
work stoppage, computer failure or malfunction, or any and all
|
|
163
|
+
other commercial damages or losses), even if such Contributor
|
|
164
|
+
has been advised of the possibility of such damages.
|
|
165
|
+
|
|
166
|
+
9. Accepting Warranty or Additional Liability. While redistributing
|
|
167
|
+
the Work or Derivative Works thereof, You may choose to offer,
|
|
168
|
+
and charge a fee for, acceptance of support, warranty, indemnity,
|
|
169
|
+
or other liability obligations and/or rights consistent with this
|
|
170
|
+
License. However, in accepting such obligations, You may act only
|
|
171
|
+
on Your own behalf and on Your sole responsibility, not on behalf
|
|
172
|
+
of any other Contributor, and only if You agree to indemnify,
|
|
173
|
+
defend, and hold each Contributor harmless for any liability
|
|
174
|
+
incurred by, or claims asserted against, such Contributor by reason
|
|
175
|
+
of your accepting any such warranty or additional liability.
|
|
176
|
+
|
|
177
|
+
END OF TERMS AND CONDITIONS
|
|
178
|
+
|
|
179
|
+
APPENDIX: How to apply the Apache License to your work.
|
|
180
|
+
|
|
181
|
+
To apply the Apache License to your work, attach the following
|
|
182
|
+
boilerplate notice, with the fields enclosed by brackets "[]"
|
|
183
|
+
replaced with your own identifying information. (Don't include
|
|
184
|
+
the brackets!) The text should be enclosed in the appropriate
|
|
185
|
+
comment syntax for the file format. We also recommend that a
|
|
186
|
+
file or class name and description of purpose be included on the
|
|
187
|
+
same "printed page" as the copyright notice for easier
|
|
188
|
+
identification within third-party archives.
|
|
189
|
+
|
|
190
|
+
Copyright [yyyy] [name of copyright owner]
|
|
191
|
+
|
|
192
|
+
Licensed under the Apache License, Version 2.0 (the "License");
|
|
193
|
+
you may not use this file except in compliance with the License.
|
|
194
|
+
You may obtain a copy of the License at
|
|
195
|
+
|
|
196
|
+
http://www.apache.org/licenses/LICENSE-2.0
|
|
197
|
+
|
|
198
|
+
Unless required by applicable law or agreed to in writing, software
|
|
199
|
+
distributed under the License is distributed on an "AS IS" BASIS,
|
|
200
|
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
201
|
+
See the License for the specific language governing permissions and
|
|
202
|
+
limitations under the License.
|