llm-sketchkit 0.1.0__tar.gz

llm_sketchkit-0.1.0/LICENSE

@@ -0,0 +1,159 @@
+Apache License
+Version 2.0, January 2004
+http://www.apache.org/licenses/
+
+TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+1. Definitions.
+
+"License" shall mean the terms and conditions for use, reproduction, and
+distribution as defined by Sections 1 through 9 of this document.
+
+"Licensor" shall mean the copyright owner or entity authorized by the copyright
+owner that is granting the License.
+
+"Legal Entity" shall mean the union of the acting entity and all other entities
+that control, are controlled by, or are under common control with that entity.
+For the purposes of this definition, "control" means (i) the power, direct or
+indirect, to cause the direction or management of such entity, whether by
+contract or otherwise, or (ii) ownership of fifty percent (50%) or more of the
+outstanding shares, or (iii) beneficial ownership of such entity.
+
+"You" (or "Your") shall mean an individual or Legal Entity exercising
+permissions granted by this License.
+
+"Source" form shall mean the preferred form for making modifications, including
+but not limited to software source code, documentation source, and configuration
+files.
+
+"Object" form shall mean any form resulting from mechanical transformation or
+translation of a Source form, including but not limited to compiled object code,
+generated documentation, and conversions to other media types.
+
+"Work" shall mean the work of authorship, whether in Source or Object form,
+made available under the License, as indicated by a copyright notice that is
+included in or attached to the work.
+
+"Derivative Works" shall mean any work, whether in Source or Object form, that
+is based on (or derived from) the Work and for which the editorial revisions,
+annotations, elaborations, or other modifications represent, as a whole, an
+original work of authorship. For the purposes of this License, Derivative Works
+shall not include works that remain separable from, or merely link (or bind by
+name) to the interfaces of, the Work and Derivative Works thereof.
+
+"Contribution" shall mean any work of authorship, including the original version
+of the Work and any modifications or additions to that Work or Derivative Works
+thereof, that is intentionally submitted to Licensor for inclusion in the Work
+by the copyright owner or by an individual or Legal Entity authorized to submit
+on behalf of the copyright owner. For the purposes of this definition,
+"submitted" means any form of electronic, verbal, or written communication sent
+to the Licensor or its representatives, including but not limited to
+communication on electronic mailing lists, source code control systems, and
+issue tracking systems that are managed by, or on behalf of, the Licensor for
+the purpose of discussing and improving the Work, but excluding communication
+that is conspicuously marked or otherwise designated in writing by the copyright
+owner as "Not a Contribution."
+
+"Contributor" shall mean Licensor and any individual or Legal Entity on behalf
+of whom a Contribution has been received by Licensor and subsequently
+incorporated within the Work.
+
+2. Grant of Copyright License. Subject to the terms and conditions of this
+License, each Contributor hereby grants to You a perpetual, worldwide,
+non-exclusive, no-charge, royalty-free, irrevocable copyright license to
+reproduce, prepare Derivative Works of, publicly display, publicly perform,
+sublicense, and distribute the Work and such Derivative Works in Source or
+Object form.
+
+3. Grant of Patent License. Subject to the terms and conditions of this License,
+each Contributor hereby grants to You a perpetual, worldwide, non-exclusive,
+no-charge, royalty-free, irrevocable (except as stated in this section) patent
+license to make, have made, use, offer to sell, sell, import, and otherwise
+transfer the Work, where such license applies only to those patent claims
+licensable by such Contributor that are necessarily infringed by their
+Contribution(s) alone or by combination of their Contribution(s) with the Work
+to which such Contribution(s) was submitted. If You institute patent litigation
+against any entity (including a cross-claim or counterclaim in a lawsuit)
+alleging that the Work or a Contribution incorporated within the Work
+constitutes direct or contributory patent infringement, then any patent licenses
+granted to You under this License for that Work shall terminate as of the date
+such litigation is filed.
+
+4. Redistribution. You may reproduce and distribute copies of the Work or
+Derivative Works thereof in any medium, with or without modifications, and in
+Source or Object form, provided that You meet the following conditions:
+
+(a) You must give any other recipients of the Work or Derivative Works a copy of
+this License; and
+
+(b) You must cause any modified files to carry prominent notices stating that
+You changed the files; and
+
+(c) You must retain, in the Source form of any Derivative Works that You
+distribute, all copyright, patent, trademark, and attribution notices from the
+Source form of the Work, excluding those notices that do not pertain to any part
+of the Derivative Works; and
+
+(d) If the Work includes a "NOTICE" text file as part of its distribution, then
+any Derivative Works that You distribute must include a readable copy of the
+attribution notices contained within such NOTICE file, excluding those notices
+that do not pertain to any part of the Derivative Works, in at least one of the
+following places: within a NOTICE text file distributed as part of the
+Derivative Works; within the Source form or documentation, if provided along
+with the Derivative Works; or, within a display generated by the Derivative
+Works, if and wherever such third-party notices normally appear. The contents of
+the NOTICE file are for informational purposes only and do not modify the
+License. You may add Your own attribution notices within Derivative Works that
+You distribute, alongside or as an addendum to the NOTICE text from the Work,
+provided that such additional attribution notices cannot be construed as
+modifying the License.
+
+You may add Your own copyright statement to Your modifications and may provide
+additional or different license terms and conditions for use, reproduction, or
+distribution of Your modifications, or for any such Derivative Works as a whole,
+provided Your use, reproduction, and distribution of the Work otherwise complies
+with the conditions stated in this License.
+
+5. Submission of Contributions. Unless You explicitly state otherwise, any
+Contribution intentionally submitted for inclusion in the Work by You to the
+Licensor shall be under the terms and conditions of this License, without any
+additional terms or conditions. Notwithstanding the above, nothing herein shall
+supersede or modify the terms of any separate license agreement you may have
+executed with Licensor regarding such Contributions.
+
+6. Trademarks. This License does not grant permission to use the trade names,
+trademarks, service marks, or product names of the Licensor, except as required
+for reasonable and customary use in describing the origin of the Work and
+reproducing the content of the NOTICE file.
+
+7. Disclaimer of Warranty. Unless required by applicable law or agreed to in
+writing, Licensor provides the Work (and each Contributor provides its
+Contributions) on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+KIND, either express or implied, including, without limitation, any warranties or
+conditions of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+PARTICULAR PURPOSE. You are solely responsible for determining the
+appropriateness of using or redistributing the Work and assume any risks
+associated with Your exercise of permissions under this License.
+
+8. Limitation of Liability. In no event and under no legal theory, whether in
+tort (including negligence), contract, or otherwise, unless required by
+applicable law (such as deliberate and grossly negligent acts) or agreed to in
+writing, shall any Contributor be liable to You for damages, including any
+direct, indirect, special, incidental, or consequential damages of any character
+arising as a result of this License or out of the use or inability to use the
+Work (including but not limited to damages for loss of goodwill, work stoppage,
+computer failure or malfunction, or any and all other commercial damages or
+losses), even if such Contributor has been advised of the possibility of such
+damages.
+
+9. Accepting Warranty or Additional Liability. While redistributing the Work or
+Derivative Works thereof, You may choose to offer, and charge a fee for,
+acceptance of support, warranty, indemnity, or other liability obligations
+and/or rights consistent with this License. However, in accepting such
+obligations, You may act only on Your own behalf and on Your sole
+responsibility, not on behalf of any other Contributor, and only if You agree to
+indemnify, defend, and hold each Contributor harmless for any liability incurred
+by, or claims asserted against, such Contributor by reason of your accepting any
+such warranty or additional liability.
+
+END OF TERMS AND CONDITIONS

llm_sketchkit-0.1.0/MANIFEST.in

@@ -0,0 +1,4 @@
+# SPDX-License-Identifier: Apache-2.0
+include scripts/__init__.py
+include scripts/release_metadata.py
+recursive-include vectors *.json

llm_sketchkit-0.1.0/PKG-INFO

@@ -0,0 +1,385 @@
+Metadata-Version: 2.4
+Name: llm-sketchkit
+Version: 0.1.0
+Summary: Deterministic Go and Python sketches for high-cardinality LLM data.
+Author: Vijay Erramilli, Codex
+License-Expression: Apache-2.0
+Project-URL: Documentation, https://github.com/llm-measurement/llm-sketchkit#readme
+Project-URL: Issues, https://github.com/llm-measurement/llm-sketchkit/issues
+Project-URL: Source, https://github.com/llm-measurement/llm-sketchkit
+Keywords: sketches,hyperloglog,minhash,bloom-filter,streaming
+Classifier: Development Status :: 4 - Beta
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3 :: Only
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
+Classifier: Typing :: Typed
+Requires-Python: >=3.11
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: protobuf<7,>=6.33.5
+Provides-Extra: dev
+Requires-Dist: mypy>=1.10; extra == "dev"
+Requires-Dist: pytest>=8.0; extra == "dev"
+Requires-Dist: ruff>=0.5; extra == "dev"
+Provides-Extra: oracle
+Requires-Dist: datasketches==5.2.0; extra == "oracle"
+Provides-Extra: release
+Requires-Dist: build==1.5.0; extra == "release"
+Requires-Dist: twine==6.2.0; extra == "release"
+Dynamic: license-file
+
+# llm-sketchkit
+
+[![CI](https://github.com/llm-measurement/llm-sketchkit/actions/workflows/ci.yml/badge.svg)](https://github.com/llm-measurement/llm-sketchkit/actions/workflows/ci.yml)
+
+`llm-sketchkit` provides continuous, bounded answers about high-cardinality agent
+traffic without exporting or indexing every underlying value. It is a small Go and
+Python library with matching semantics for text canonicalization, privacy-preserving
+keyed hashes, mergeable sketches, and a deterministic protobuf wire format.
+
+It provides bounded measurement primitives for AI agent observability pipelines,
+especially when exporting and indexing every identity or event would be expensive,
+slow to query, or inappropriate to retain.
+
+Raw prompts and identifiers do not need to enter sketch state. Producers can
+summarize locally and merge compatible sketches across processes or languages.
+
+## When This Fits
+
+Use `llm-sketchkit` inside telemetry producers and processing components when
+exporting and indexing every key would create uncontrolled cardinality, make
+operational queries slow or unpredictable, or retain values that should not enter
+aggregate state.
+
+Agent fleets and multi-agent systems can produce more identities and events than
+an observability backend should continuously index. The library provides bounded,
+mergeable summaries across workers and windows.
+
+Questions it can help answer include:
+
+- **How many distinct prompts, users, sessions, tools, or documents are active
+  without keeping one counter per value?** HLL++ provides a bounded estimate.
+- **Your token budget is climbing and FinOps wants to know which configured
+  identities account for the reported volume. How certain is the answer?** Weighted
+  frequent-items identifies token-heavy or request-heavy keys with deterministic
+  lower and upper bounds.
+- **Have we already observed this request or document without maintaining an exact
+  set of every value?** Bloom filters provide bounded approximate membership checks.
+- **Did the prompt, tool, or retrieval-document population change materially after
+  a deployment or model change?** MinHash compares large sets using bounded
+  similarity signatures.
+- **Can Go services summarize locally while Python analysis jobs read and merge the
+  same state?** Shared profiles, fixtures, and wire semantics keep the two
+  implementations compatible.
+
+For investigations described as "tokenmaxxing" (also written "token-maxing"),
+reported token counts can be used as weights in the frequent-items sketch to identify
+which keyed values account for the most token volume. The library measures
+concentration; it does not infer task value, enforce budgets, or stop agent loops.
+See [Token-Volume Heavy Hitters](#token-volume-heavy-hitters) for a runnable example.
+
+Inputs can be canonicalized and keyed before entering sketch state. This keeps
+raw values out of the sketch, but the resulting hashes remain pseudonymous and
+linkable while the same secret is in use.
+
+This is a sketch library, not a trace collector, sampling processor, storage
+backend, dashboard, or differential-privacy system. The [FAQ](https://github.com/llm-measurement/llm-sketchkit/blob/main/docs/FAQ.md)
+answers common questions about fit, operational bounds, accuracy, and
+interoperability.
+
+## Choosing An Integration
+
+| Your pipeline | Use | Why |
+|---|---|---|
+| GenAI spans already flow through an OpenTelemetry Collector | [OpenTelemetry Collector connector](https://github.com/llm-measurement/otelcol-genai-sketches) | It applies keyed hashing, bounded windows, cardinality controls, and trace-to-metrics conversion at the collector boundary. |
+| A custom Go or Python streaming service processes events | `llm-sketchkit` directly | Update sketches inside each bounded window, then serialize or merge compatible summaries. |
+| A batch or warehouse job reads stored events | `llm-sketchkit` directly | Build bounded summaries per partition or window and merge them before publishing results. |
+| You only need to store or visualize finished metrics | Your existing backend integration | The library is not an exporter; ClickHouse, Datadog, Prometheus, and similar systems normally receive the aggregated results. |
+
+Use the collector path when the source spans already flow through OpenTelemetry. Use
+the library when you own the event-processing code or need matching Go and Python
+summaries outside an OpenTelemetry pipeline. In either case, compatible producers
+must agree on profile, hash domain, hash algorithm, secret, and window boundaries.
+
+## Included Sketches
+
+| Component | Use it for | Important property |
+|---|---|---|
+| HLL++ | Approximate distinct counts | Bounded, mergeable state |
+| Weighted frequent-items | Heavy hitters and top items | Deterministic lower and upper bounds |
+| Bloom filter | Set membership | No false negatives; configurable false-positive rate |
+| MinHash | Approximate Jaccard similarity | Bounded, mergeable signatures |
+
+The Go and Python implementations share the same profiles, hash domains, test
+vectors, and serialized representation.
+
+## Evidence At A Glance
+
+Measurements use deterministic workloads and report the least favorable of five
+Linux benchmark runs where applicable.
+
+- HMAC-SHA256-64 sustained at least **1.65 million 64-byte inputs/s/core** and
+  **850,340 1 KiB inputs/s/core** on an Intel Xeon Platinum 8573C with Go 1.26.5.
+- HLL++ and weighted frequent-items updates took at most **10.58 ns/op** and
+  **145.6 ns/op**, respectively, with **0 allocations/op** in the measured paths.
+- The HLL++ `small` profile's maximum observed relative error was **2.3301%**
+  across the characterization grid, within its **2.4375%** enforced bound.
+- Bloom profile false-positive rates were at or below their configured targets
+  in the measured trials, with zero false negatives among inserted hashes.
+- MinHash mean absolute error fell from **0.02845** at `k=128` to **0.02009** at
+  `k=256`, closely following the expected inverse-square-root relationship.
+- Both weighted frequent-items oracle workloads retained **100% true top-20
+  recall** and valid no-false-positive query results in both implementations.
+
+See the [visual scorecard](https://github.com/llm-measurement/llm-sketchkit/blob/main/reports/scorecard.md),
+[raw measurement records](https://github.com/llm-measurement/llm-sketchkit/blob/main/reports/README.md),
+and [general-purpose library comparison](https://github.com/llm-measurement/llm-sketchkit/blob/main/docs/DATASKETCHES.md)
+for methods, limitations, and reproduction commands.
+
+## Requirements
+
+- Go 1.25 or newer, with the latest security patch for that release line
+- Python 3.11 or newer
+
+## Install
+
+### Python
+
+```sh
+python -m pip install llm-sketchkit
+```
+
+Generate a process secret:
+
+```sh
+export LLM_SKETCHKIT_SECRET="$(python -c 'import secrets; print(secrets.token_hex(32))')"
+```
+
+Run a distinct-count example:
+
+```sh
+python - <<'PY'
+from llm_sketchkit import PROMPT_V1, canonicalize_text_v1, hash64, hllpp
+from llm_sketchkit import secret_from_env
+
+secret = secret_from_env("LLM_SKETCHKIT_SECRET")
+sketch = hllpp.new("small", PROMPT_V1)
+
+canonical = canonicalize_text_v1("  cafe\u0301\r\n")
+sketch.add_hash(hash64(secret, PROMPT_V1, canonical))
+
+print(f"estimated distinct prompts: {sketch.estimate():.0f}")
+PY
+```
+
+Expected output:
+
+```text
+estimated distinct prompts: 1
+```
+
+### Go
+
+From an existing Go module, add the package used by the equivalent example:
+
+```sh
+go get github.com/llm-measurement/llm-sketchkit/go/sketchkit/hllpp@latest
+```
+
+Use the same `LLM_SKETCHKIT_SECRET` and run:
+
+```go
+package main
+
+import (
+	"fmt"
+	"log"
+
+	"github.com/llm-measurement/llm-sketchkit/go/sketchkit/canon"
+	sketchhash "github.com/llm-measurement/llm-sketchkit/go/sketchkit/hash"
+	"github.com/llm-measurement/llm-sketchkit/go/sketchkit/hllpp"
+)
+
+func main() {
+	secret, err := sketchhash.SecretFromEnv("LLM_SKETCHKIT_SECRET")
+	if err != nil {
+		log.Fatal(err)
+	}
+
+	sketch, err := hllpp.New(
+		hllpp.ProfileSmall,
+		sketchhash.PromptV1,
+		sketchhash.HMACSHA25664,
+	)
+	if err != nil {
+		log.Fatal(err)
+	}
+
+	canonical, err := canon.CanonicalizeString(canon.TextV1, "  cafe\u0301\r\n")
+	if err != nil {
+		log.Fatal(err)
+	}
+	digest, err := sketchhash.Hash64(secret, sketchhash.PromptV1, canonical)
+	if err != nil {
+		log.Fatal(err)
+	}
+
+	sketch.AddHash(digest)
+	fmt.Printf("estimated distinct prompts: %.0f\n", sketch.Estimate())
+}
+```
+
+## Go To Python Notebook
+
+The [runnable notebook](https://github.com/llm-measurement/llm-sketchkit/blob/main/examples/go-to-python/go-to-python.ipynb) produces
+per-service HLL++ and weighted frequent-items summaries in Go, then loads,
+validates, merges, and plots them in Python. It checks that serialization round trips
+return the same bytes,
+explicit merge rejection for incompatible profiles, distinct-count estimates,
+and deterministic bounds around token-heavy pseudonymous keys.
+
+The tutorial uses an exact synthetic side channel only to check its estimates.
+Raw synthetic identifiers do not enter the emitted files. See the
+[example guide](https://github.com/llm-measurement/llm-sketchkit/blob/main/examples/go-to-python/README.md)
+for setup and security details.
+
+## Token-Volume Heavy Hitters
+
+If by "tokenmaxxing" (also written "token-maxing") you mean unexpected or runaway
+token consumption, use a bounded frequent-items sketch to find where reported volume
+is concentrated. Hash the value being investigated, such as a prompt template, tool,
+or user, with the registered domain for that entity class, and use the reported token
+count as its weight. This example measures prompt templates with `prompt:v1`:
+
+```python
+from llm_sketchkit import PROMPT_V1, canonicalize_text_v1, frequentitems
+from llm_sketchkit import hash64, secret_from_env
+
+secret = secret_from_env("LLM_SKETCHKIT_SECRET")
+sketch = frequentitems.Sketch("small", PROMPT_V1)
+
+events = [
+    ("support/refund", 1_240),
+    ("research/synthesis", 8_900),
+    ("support/refund", 980),
+    ("coding/review", 3_600),
+]
+
+for prompt_template, reported_tokens in events:
+    canonical = canonicalize_text_v1(prompt_template)
+    digest = hash64(secret, PROMPT_V1, canonical)
+    sketch.add_hash(digest, reported_tokens)
+
+for item in sketch.frequent_items(frequentitems.NO_FALSE_NEGATIVES)[:10]:
+    print(
+        f"{item.hash:016x} estimate={item.estimate} "
+        f"bounds=[{item.lower_bound}, {item.upper_bound}]"
+    )
+```
+
+The sketch retains at most the selected profile's bounded map size. Returned hashes
+are pseudonymous and remain linkable while the same secret and domain are in use.
+The estimate is not a billing total; use the lower and upper bounds when deciding
+whether an item is meaningfully heavy.
+
+Do not substitute guessed weights when token usage is missing. Count missing usage
+separately so operators know how complete the reported totals are. For an OTLP
+pipeline with ready-made metrics, bounded slices, missing-usage accounting, and
+token-weighted top-k snapshots, use
+[`otelcol-genai-sketches`](https://github.com/llm-measurement/otelcol-genai-sketches).
+
+## Merge Sketches
+
+Sketches merge only when their kind, profile, hash domain, hash algorithm, and
+shape metadata match. A mismatch is an error rather than an implicit conversion.
+
+```python
+left = hllpp.new("small", PROMPT_V1)
+right = hllpp.new("small", PROMPT_V1)
+
+left.add_hash(hash64(secret, PROMPT_V1, canonicalize_text_v1("alpha")))
+right.add_hash(hash64(secret, PROMPT_V1, canonicalize_text_v1("beta")))
+
+left.merge(right)
+print(f"merged distinct prompts: {left.estimate():.0f}")
+```
+
+## Security And Privacy
+
+- Hash inputs with a registered domain and a high-entropy secret before adding
+  them to a sketch. The built-in secret loaders require at least 16 bytes and
+  reject known placeholder values.
+- Keyed hashes are pseudonymous, not anonymous. Anyone with the secret can test
+  candidate values, and repeated hashes remain linkable while the same secret
+  and domain are in use.
+- Never log, serialize, or commit the hash secret. Rotate it when the trust
+  boundary changes; rotation intentionally breaks comparison with older state.
+- Bound raw input size before canonicalization. Canonicalization operates on
+  in-memory text and intentionally leaves application-specific limits to callers.
+- Sketches reveal bounded aggregate information and may reveal membership or
+  recurrence. They do not provide differential privacy.
+- Treat serialized sketches as untrusted input at process boundaries. The parse
+  APIs cap input size and reject invalid profiles, domains, shapes, counters, and
+  register values.
+
+See [SECURITY.md](https://github.com/llm-measurement/llm-sketchkit/blob/main/SECURITY.md)
+for private vulnerability reporting.
+
+## Development
+
+```sh
+git clone https://github.com/llm-measurement/llm-sketchkit.git
+cd llm-sketchkit
+python -m venv .venv
+source .venv/bin/activate
+python -m pip install --upgrade pip==26.2.1 setuptools==83.0.0
+python -m pip install -e '.[dev]'
+```
+
+## Wire Compatibility
+
+Deterministic protobuf encoding is part of the compatibility surface. Go and
+Python are checked against the same canonicalization, hashing, sketch, and
+cross-language merge fixtures in
+[`vectors/`](https://github.com/llm-measurement/llm-sketchkit/tree/main/vectors/).
+
+Run all local checks:
+
+```sh
+go test ./... -race
+python -m pytest -q
+ruff check .
+mypy --strict
+```
+
+The optional Apache DataSketches comparison checks weighted frequent-items
+query behavior against an independent implementation:
+
+```sh
+python -m pip install -e '.[oracle]'
+python scripts/datasketches_oracle.py --check
+```
+
+## Reference
+
+- [`spec/`](https://github.com/llm-measurement/llm-sketchkit/tree/main/spec/) defines canonicalization, hashing, profiles, and wire encoding.
+- [`vectors/`](https://github.com/llm-measurement/llm-sketchkit/tree/main/vectors/) contains executable conformance fixtures.
+- [`reports/`](https://github.com/llm-measurement/llm-sketchkit/tree/main/reports/) contains benchmark, accuracy, and oracle results.
+- [`bench/`](https://github.com/llm-measurement/llm-sketchkit/tree/main/bench/) contains the Go benchmark harnesses.
+- [`docs/FAQ.md`](https://github.com/llm-measurement/llm-sketchkit/blob/main/docs/FAQ.md) answers common adoption questions.
+- [`CHANGELOG.md`](https://github.com/llm-measurement/llm-sketchkit/blob/main/CHANGELOG.md) records release-level changes.
+
+## Status
+
+`llm-sketchkit 0.1.x` is a supported release line. Patch releases preserve the
+documented wire formats, named hash domains, and Go and Python APIs exercised by
+the checked-in conformance vectors. Additive or incompatible changes to that
+surface receive a new minor version and are called out in the changelog before
+`1.0`.
+
+## License
+
+Apache-2.0.