Skip to content

chore(deps): bump the python-deps group across 1 directory with 6 updates - #453

Open
dependabot[bot] wants to merge 1 commit into
mainfrom
dependabot/pip/python-deps-eee60ea554
Open

chore(deps): bump the python-deps group across 1 directory with 6 updates#453
dependabot[bot] wants to merge 1 commit into
mainfrom
dependabot/pip/python-deps-eee60ea554

Conversation

@dependabot

@dependabot dependabot Bot commented on behalf of github Sep 22, 2026

Copy link
Copy Markdown

Updates the requirements on lancedb, everalgo-user-memory, everalgo-agent-memory, everalgo-knowledge, everalgo-boundary and everalgo-core to permit the latest version.
Updates lancedb to 0.38.0

Release notes

Sourced from lancedb's releases.

LanceDB v0.38.0

🛠 Breaking Changes

🎉 New Features

... (truncated)

Commits
  • 8c68e0c Bump version: 0.38.0-beta.16 → 0.38.0
  • 1a9414c Bump version: 0.38.0-beta.15 → 0.38.0-beta.16
  • c4ee8ae feat: update lance dependency to v11.0.0 (#4093)
  • 57b8d3b Bump version: 0.38.0-beta.14 → 0.38.0-beta.15
  • c6dfe83 feat(python): support large_utf8 function signatures (#4092)
  • d5dac65 feat: support Blob v2 UDF signatures (#4091)
  • 1b0fc2c Bump version: 0.38.0-beta.13 → 0.38.0-beta.14
  • a417e46 feat(functions): support GPU resource requirements (#4085)
  • fcdc3f9 fix: allow multiple function bindings per table (#4090)
  • 0c4e066 Bump version: 0.38.0-beta.12 → 0.38.0-beta.13
  • Additional commits viewable in compare view

Updates everalgo-user-memory from 0.4.0 to 0.7.0

Changelog

Sourced from everalgo-user-memory's changelog.

Changelog

All notable changes to the EverAlgo monorepo are documented here. Format follows Keep a Changelog 1.1.0. Each distribution follows its own Semantic Versioning 2.0 cadence — there is no umbrella version.

Per-distribution changelogs are the source of truth. This file is a navigation index.

Distribution status

The table tracks the current version declared in each packages/everalgo-*/pyproject.toml. All eight distributions are published to PyPI.

Distribution Version Changelog
everalgo-core 0.5.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-core/CHANGELOG.md
everalgo-boundary 0.3.0 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-boundary/CHANGELOG.md
everalgo-clustering 0.2.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-clustering/CHANGELOG.md
everalgo-rank 0.4.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-rank/CHANGELOG.md
everalgo-user-memory 0.8.0rc9 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-user-memory/CHANGELOG.md
everalgo-agent-memory 0.5.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-agent-memory/CHANGELOG.md
everalgo-parser 0.2.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-parser/CHANGELOG.md
everalgo-knowledge 0.2.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-knowledge/CHANGELOG.md

Patch release — 2026-09-16

Three distributions updated together because trajectory truncation and knowledge splitting both consume the shared core tokenizer. Text that quotes an o200k_base control-token literal such as <|endoftext|> is now measured and sliced as ordinary text instead of raising ValueError. The two downstream packages require everalgo-core>=0.5.1, so a clean resolver cannot select an older core that lacks the helper or retains the crash.

Distribution Version Bump
everalgo-core 0.5.1 patch — special-token-safe shared encoding for token counting and splitting
everalgo-agent-memory 0.5.1 patch — special-token-safe trajectory truncation and corrected core floor
everalgo-knowledge 0.2.1 patch — special-token-safe block splitting, corrected core floor, and removal of the redundant direct tiktoken dependency

Detail in https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-core/CHANGELOG.md, https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-agent-memory/CHANGELOG.md, and https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-knowledge/CHANGELOG.md.

[everalgo-knowledge/0.2.0] - 2026-09-15

parse_llm_json no longer throws away a response whose only defect is an unescaped backslash. A model copying LaTeX out of the source document writes $\kappa$ as \k, which JSON does not accept, so a complete and well-formed topic tree was discarded and the document ingested as empty — reproducibly, on every model tried. The parser now retries once with those backslashes doubled, and only after a strict parse has already failed, so valid JSON is never rewritten. LaTeX that collides with a legal escape (\theta / \frac / \bar are tab / form-feed / backspace) stays lossy; recovering those needs the model to emit valid JSON in the first place.

Detail in https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-knowledge/CHANGELOG.md.

[everalgo-knowledge/0.2.0rc1] - 2026-09-14

Knowledge extraction now caps a single atom at 20,000 tokens before assembling 80,000-token LLM batches. Oversized Markdown tables and lists split on line boundaries, table headers repeat across pieces, and block ids remain dense so downstream topic references preserve their positional contract. This release candidate fixes the production failure where a table-heavy document collapsed into one 270,000-token request; ordinary documents below the cap retain byte-identical atomization.

Detail in https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-knowledge/CHANGELOG.md.

[everalgo-user-memory/0.8.0rc9] - 2026-09-14

... (truncated)

Commits
  • 20c8622 Merge branch 'release/user-memory-v0.7.0' into 'main'
  • 522b313 🔖 release(user-memory): 0.7.0 — profile prompt rewrite around one definition ...
  • 8675b16 Merge branch 'fix/profile-explicit-info-event-pollution' into 'main'
  • 2fb9ddf 🐛 fix(user-memory): rewrite the profile prompts around one definition of a po...
  • 6756474 Merge branch 'release/boundary-user-agent-memory' into 'main'
  • 27f400c 🔖 release(release): boundary 0.3.0 + user-memory 0.6.0 + agent-memory 0.5.0 —...
  • 305ed10 Merge branch 'fix/boundary-should-wait' into 'main'
  • 0da1697 🐛 fix(boundary)!: surface the LLM's should_wait verdict instead of discarding it
  • 0f0ae7d Merge branch 'release/core-user-memory-v0.5.0' into 'main'
  • b68b0fe 🔖 release(release): core 0.5.0 + user-memory 0.5.0 — model-written Episode.su...
  • Additional commits viewable in compare view

Updates everalgo-agent-memory from 0.4.0 to 0.5.1

Changelog

Sourced from everalgo-agent-memory's changelog.

Changelog

All notable changes to the EverAlgo monorepo are documented here. Format follows Keep a Changelog 1.1.0. Each distribution follows its own Semantic Versioning 2.0 cadence — there is no umbrella version.

Per-distribution changelogs are the source of truth. This file is a navigation index.

Distribution status

The table tracks the current version declared in each packages/everalgo-*/pyproject.toml. All eight distributions are published to PyPI.

Distribution Version Changelog
everalgo-core 0.5.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-core/CHANGELOG.md
everalgo-boundary 0.3.0 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-boundary/CHANGELOG.md
everalgo-clustering 0.2.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-clustering/CHANGELOG.md
everalgo-rank 0.4.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-rank/CHANGELOG.md
everalgo-user-memory 0.8.0rc9 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-user-memory/CHANGELOG.md
everalgo-agent-memory 0.5.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-agent-memory/CHANGELOG.md
everalgo-parser 0.2.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-parser/CHANGELOG.md
everalgo-knowledge 0.2.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-knowledge/CHANGELOG.md

Patch release — 2026-09-16

Three distributions updated together because trajectory truncation and knowledge splitting both consume the shared core tokenizer. Text that quotes an o200k_base control-token literal such as <|endoftext|> is now measured and sliced as ordinary text instead of raising ValueError. The two downstream packages require everalgo-core>=0.5.1, so a clean resolver cannot select an older core that lacks the helper or retains the crash.

Distribution Version Bump
everalgo-core 0.5.1 patch — special-token-safe shared encoding for token counting and splitting
everalgo-agent-memory 0.5.1 patch — special-token-safe trajectory truncation and corrected core floor
everalgo-knowledge 0.2.1 patch — special-token-safe block splitting, corrected core floor, and removal of the redundant direct tiktoken dependency

Detail in https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-core/CHANGELOG.md, https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-agent-memory/CHANGELOG.md, and https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-knowledge/CHANGELOG.md.

[everalgo-knowledge/0.2.0] - 2026-09-15

parse_llm_json no longer throws away a response whose only defect is an unescaped backslash. A model copying LaTeX out of the source document writes $\kappa$ as \k, which JSON does not accept, so a complete and well-formed topic tree was discarded and the document ingested as empty — reproducibly, on every model tried. The parser now retries once with those backslashes doubled, and only after a strict parse has already failed, so valid JSON is never rewritten. LaTeX that collides with a legal escape (\theta / \frac / \bar are tab / form-feed / backspace) stays lossy; recovering those needs the model to emit valid JSON in the first place.

Detail in https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-knowledge/CHANGELOG.md.

[everalgo-knowledge/0.2.0rc1] - 2026-09-14

Knowledge extraction now caps a single atom at 20,000 tokens before assembling 80,000-token LLM batches. Oversized Markdown tables and lists split on line boundaries, table headers repeat across pieces, and block ids remain dense so downstream topic references preserve their positional contract. This release candidate fixes the production failure where a table-heavy document collapsed into one 270,000-token request; ordinary documents below the cap retain byte-identical atomization.

Detail in https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-knowledge/CHANGELOG.md.

[everalgo-user-memory/0.8.0rc9] - 2026-09-14

... (truncated)

Commits
  • 7746a87 Merge branch 'release/tokenizer-special-token-patch' into 'main'
  • 49daece 🔖 release(release): core 0.5.1 + agent-memory 0.5.1 + knowledge 0.2.1 — speci...
  • d3a3117 Merge branch 'fix/tokenizer-special-tokens' into 'main'
  • 4bb3e31 🐛 fix(core): stop tiktoken rejecting text that quotes a special token
  • d830fec Merge branch 'release/knowledge-v0.2.0-changelog' into 'main'
  • 31881a3 📝 docs(knowledge): restore Unreleased changelog section
  • dc89c78 Merge branch 'fix/knowledge-json-invalid-escape' into 'main'
  • 91b11ec 🐛 fix(knowledge): recover LLM JSON whose only defect is an invalid backslash ...
  • b1fa05a Merge branch 'release/user-memory-v0.8.0rc9' into 'main'
  • 2030d42 🔖 release(user-memory): 0.8.0rc9 — the historical pass stops re-adding supers...
  • Additional commits viewable in compare view

Updates everalgo-knowledge from 0.1.1 to 0.2.1

Changelog

Sourced from everalgo-knowledge's changelog.

Changelog

All notable changes to the EverAlgo monorepo are documented here. Format follows Keep a Changelog 1.1.0. Each distribution follows its own Semantic Versioning 2.0 cadence — there is no umbrella version.

Per-distribution changelogs are the source of truth. This file is a navigation index.

Distribution status

The table tracks the current version declared in each packages/everalgo-*/pyproject.toml. All eight distributions are published to PyPI.

Distribution Version Changelog
everalgo-core 0.5.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-core/CHANGELOG.md
everalgo-boundary 0.3.0 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-boundary/CHANGELOG.md
everalgo-clustering 0.2.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-clustering/CHANGELOG.md
everalgo-rank 0.4.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-rank/CHANGELOG.md
everalgo-user-memory 0.8.0rc9 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-user-memory/CHANGELOG.md
everalgo-agent-memory 0.5.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-agent-memory/CHANGELOG.md
everalgo-parser 0.2.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-parser/CHANGELOG.md
everalgo-knowledge 0.2.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-knowledge/CHANGELOG.md

Patch release — 2026-09-16

Three distributions updated together because trajectory truncation and knowledge splitting both consume the shared core tokenizer. Text that quotes an o200k_base control-token literal such as <|endoftext|> is now measured and sliced as ordinary text instead of raising ValueError. The two downstream packages require everalgo-core>=0.5.1, so a clean resolver cannot select an older core that lacks the helper or retains the crash.

Distribution Version Bump
everalgo-core 0.5.1 patch — special-token-safe shared encoding for token counting and splitting
everalgo-agent-memory 0.5.1 patch — special-token-safe trajectory truncation and corrected core floor
everalgo-knowledge 0.2.1 patch — special-token-safe block splitting, corrected core floor, and removal of the redundant direct tiktoken dependency

Detail in https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-core/CHANGELOG.md, https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-agent-memory/CHANGELOG.md, and https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-knowledge/CHANGELOG.md.

[everalgo-knowledge/0.2.0] - 2026-09-15

parse_llm_json no longer throws away a response whose only defect is an unescaped backslash. A model copying LaTeX out of the source document writes $\kappa$ as \k, which JSON does not accept, so a complete and well-formed topic tree was discarded and the document ingested as empty — reproducibly, on every model tried. The parser now retries once with those backslashes doubled, and only after a strict parse has already failed, so valid JSON is never rewritten. LaTeX that collides with a legal escape (\theta / \frac / \bar are tab / form-feed / backspace) stays lossy; recovering those needs the model to emit valid JSON in the first place.

Detail in https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-knowledge/CHANGELOG.md.

[everalgo-knowledge/0.2.0rc1] - 2026-09-14

Knowledge extraction now caps a single atom at 20,000 tokens before assembling 80,000-token LLM batches. Oversized Markdown tables and lists split on line boundaries, table headers repeat across pieces, and block ids remain dense so downstream topic references preserve their positional contract. This release candidate fixes the production failure where a table-heavy document collapsed into one 270,000-token request; ordinary documents below the cap retain byte-identical atomization.

Detail in https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-knowledge/CHANGELOG.md.

[everalgo-user-memory/0.8.0rc9] - 2026-09-14

... (truncated)

Commits
  • 7746a87 Merge branch 'release/tokenizer-special-token-patch' into 'main'
  • 49daece 🔖 release(release): core 0.5.1 + agent-memory 0.5.1 + knowledge 0.2.1 — speci...
  • d3a3117 Merge branch 'fix/tokenizer-special-tokens' into 'main'
  • 4bb3e31 🐛 fix(core): stop tiktoken rejecting text that quotes a special token
  • d830fec Merge branch 'release/knowledge-v0.2.0-changelog' into 'main'
  • 31881a3 📝 docs(knowledge): restore Unreleased changelog section
  • dc89c78 Merge branch 'fix/knowledge-json-invalid-escape' into 'main'
  • 91b11ec 🐛 fix(knowledge): recover LLM JSON whose only defect is an invalid backslash ...
  • b1fa05a Merge branch 'release/user-memory-v0.8.0rc9' into 'main'
  • 2030d42 🔖 release(user-memory): 0.8.0rc9 — the historical pass stops re-adding supers...
  • Additional commits viewable in compare view

Updates everalgo-boundary from 0.2.1 to 0.3.0

Changelog

Sourced from everalgo-boundary's changelog.

[everalgo-rank/0.3.0] - 2026-05-28

Per-distribution minor bump of everalgo-rank only. Other distributions are unchanged.

Changed (breaking)

  • everalgo-rank: skill ranker's standalone post-rerank verify stage is removed; relevance grading folds into a single LLM pass with a new min_rerank_score hard-threshold gate. Public API averify / verify and the enable_verify / verify_threshold / verify_prompt parameters are gone. See packages/everalgo-rank/CHANGELOG.md for migration.

Changed

  • License relicensed from MIT to Apache-2.0.

[0.2.0] - 2026-05-27

Coordinated release-readiness baseline across the seven published distributions. Per-distribution detail lives in each package's CHANGELOG; highlights:

Changed (breaking)

  • everalgo-clustering: cluster_by_geometry is now synchronous (pure compute, no I/O).
  • everalgo-boundary / everalgo-user-memory: the unimplemented WorkspaceMemCellExtractor stub is removed from the public API (reachable only via explicit submodule import).
  • everalgo-core: the empty everalgo.protocols placeholder is removed.

Changed

  • Packaging metadata enriched across all distributions (keywords, full classifiers, author email).
  • Coverage gate centralised in pyproject.toml; CI now tests Python 3.12 / 3.13 / 3.14; tag-triggered PyPI publish via Trusted Publishing added.
  • Community health files added (SECURITY.md, CODE_OF_CONDUCT.md, CONTRIBUTING.md, CITATION.cff).
  • everalgo-knowledge marked not-for-publish (Private :: Do Not Upload); namespace reserved only.

[0.1.0] - 2026-05-17

Added

  • Monorepo scaffold — 8-distribution uv virtual workspace with a shared lockfile and PEP 420 native namespace packages under everalgo.*.
  • Release infrastructure — cliff.toml for per-distribution CHANGELOG generation, py.typed markers, pre-commit hook (ruff check --fix + ruff format + standard sanitisers).
  • Logging conventions (ADR-013) — NullHandler on every subpackage logger; SensitiveHeadersFilter on the everalgo.llm logger; ruff rules G + LOG + TRY enforce library-safe logging patterns.
  • 6 runnable quickstart examples under examples/ covering boundary detection, geometry clustering, Episode extraction, AgentCase extraction, rerank, and the full user-memory pipeline. All use FakeLLMClient; no API key required.
  • Cross-package integration tests under tests/integration/ for the full user-memory pipeline and full agent-memory pipeline.
  • Cluster.members: list[str] — caller-supplied entity ids; algorithm appends on merge, never inspects semantics. Caller uses this to track which business entities belong to each cluster.
  • EpisodeExtractor generic mode — pass sender_id=None to extract one whole-memcell generic episode using EPISODE_GENERATION_PROMPT (cheaper than per-user fan-out). Pass a user id to use the USER-focused USER_EPISODE_GENERATION_PROMPT.
  • AtomicFactExtractor generic mode — pass sender_id=None to extract facts not bound to any user.
  • ProfileExtractor UPDATE mode — pass old_profile=Profile to use the incremental ops-based update path. When the merged item count exceeds an internal compact threshold, a second LLM pass runs automatically (caller-transparent). old_profile=None continues to trigger full INIT extraction.
  • AgentSkillExtractor expanded kwargs — all policy thresholds exposed as per-call keyword arguments: skip_quality_threshold, skip_maturity_scoring, maturity_threshold, retire_confidence, failure_quality_threshold, max_case_history, max_description_tokens, max_content_tokens, maturity_trivial_change_ratio, maturity_reeval_change_ratio. SkillConfig public class removed; _SkillCfg is now an internal dataclass.
  • skip_quality_threshold=0.2 short-circuit in AgentSkillExtractor — cases with quality_score < threshold return [] without calling the LLM.
  • Time-format utilities down to core — format_message_timestamp / format_natural_language_time live in everalgo.llm.format and are shared across all subpackages.
  • Robust LLM JSON parser down to core — everalgo.llm.parse.parse_llm_json_object (fence → direct loads → outermost-braces fallback) shared across all subpackages.
  • Boundary prompt ISO 8601 UTC format — all boundary detection prompts (EN and ZH) now use YYYY-MM-DDTHH:MM:SSZ for timestamps.
  • Instance-only LLM injection — llm= is passed at constructor time only; the 4-layer global resolution (global default / scoped context / per-call override) has been removed.
  • AgentBoundaryDetector — filter → detect → remap pipeline for mixed ConversationItem trajectories; tool-call items are preserved in output MemCell.items.
  • BoundaryDetector renamed — ChatBoundaryDetector → UserBoundaryDetector → BoundaryDetector (final, matching the no-prefix convention used by all other facades).

... (truncated)

Commits
  • 27f400c 🔖 release(release): boundary 0.3.0 + user-memory 0.6.0 + agent-memory 0.5.0 —...
  • 305ed10 Merge branch 'fix/boundary-should-wait' into 'main'
  • 0da1697 🐛 fix(boundary)!: surface the LLM's should_wait verdict instead of discarding it
  • 0f0ae7d Merge branch 'release/core-user-memory-v0.5.0' into 'main'
  • b68b0fe 🔖 release(release): core 0.5.0 + user-memory 0.5.0 — model-written Episode.su...
  • 86368f3 Merge branch 'feat/episode-summary' into 'main'
  • dc6c8c7 ✨ feat(user-memory)!: make Episode.summary a model-written preview instead of...
  • a2ad5f8 Merge branch 'feat/reflect-output-language' into 'main'
  • eb669cd ✨ feat(user-memory): let the caller choose EpisodeReflector's output language...
  • 9eb1a24 Merge branch 'feat/output-language-argument' into 'main'
  • Additional commits viewable in compare view

Updates everalgo-core from 0.3.0 to 0.5.1

Changelog

Sourced from everalgo-core's changelog.

Changelog

All notable changes to the EverAlgo monorepo are documented here. Format follows Keep a Changelog 1.1.0. Each distribution follows its own Semantic Versioning 2.0 cadence — there is no umbrella version.

Per-distribution changelogs are the source of truth. This file is a navigation index.

Distribution status

The table tracks the current version declared in each packages/everalgo-*/pyproject.toml. All eight distributions are published to PyPI.

Distribution Version Changelog
everalgo-core 0.5.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-core/CHANGELOG.md
everalgo-boundary 0.3.0 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-boundary/CHANGELOG.md
everalgo-clustering 0.2.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-clustering/CHANGELOG.md
everalgo-rank 0.4.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-rank/CHANGELOG.md
everalgo-user-memory 0.8.0rc9 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-user-memory/CHANGELOG.md
everalgo-agent-memory 0.5.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-agent-memory/CHANGELOG.md
everalgo-parser 0.2.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-parser/CHANGELOG.md
everalgo-knowledge 0.2.1 https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-knowledge/CHANGELOG.md

Patch release — 2026-09-16

Three distributions updated together because trajectory truncation and knowledge splitting both consume the shared core tokenizer. Text that quotes an o200k_base control-token literal such as <|endoftext|> is now measured and sliced as ordinary text instead of raising ValueError. The two downstream packages require everalgo-core>=0.5.1, so a clean resolver cannot select an older core that lacks the helper or retains the crash.

Distribution Version Bump
everalgo-core 0.5.1 patch — special-token-safe shared encoding for token counting and splitting
everalgo-agent-memory 0.5.1 patch — special-token-safe trajectory truncation and corrected core floor
everalgo-knowledge 0.2.1 patch — special-token-safe block splitting, corrected core floor, and removal of the redundant direct tiktoken dependency

Detail in https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-core/CHANGELOG.md, https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-agent-memory/CHANGELOG.md, and https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-knowledge/CHANGELOG.md.

[everalgo-knowledge/0.2.0] - 2026-09-15

parse_llm_json no longer throws away a response whose only defect is an unescaped backslash. A model copying LaTeX out of the source document writes $\kappa$ as \k, which JSON does not accept, so a complete and well-formed topic tree was discarded and the document ingested as empty — reproducibly, on every model tried. The parser now retries once with those backslashes doubled, and only after a strict parse has already failed, so valid JSON is never rewritten. LaTeX that collides with a legal escape (\theta / \frac / \bar are tab / form-feed / backspace) stays lossy; recovering those needs the model to emit valid JSON in the first place.

Detail in https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-knowledge/CHANGELOG.md.

[everalgo-knowledge/0.2.0rc1] - 2026-09-14

Knowledge extraction now caps a single atom at 20,000 tokens before assembling 80,000-token LLM batches. Oversized Markdown tables and lists split on line boundaries, table headers repeat across pieces, and block ids remain dense so downstream topic references preserve their positional contract. This release candidate fixes the production failure where a table-heavy document collapsed into one 270,000-token request; ordinary documents below the cap retain byte-identical atomization.

Detail in https://github.com/EverMind-AI/EverAlgo/blob/main/packages/everalgo-knowledge/CHANGELOG.md.

[everalgo-user-memory/0.8.0rc9] - 2026-09-14

... (truncated)

Commits
  • 7746a87 Merge branch 'release/tokenizer-special-token-patch' into 'main'
  • 49daece 🔖 release(release): core 0.5.1 + agent-memory 0.5.1 + knowledge 0.2.1 — speci...
  • d3a3117 Merge branch 'fix/tokenizer-special-tokens' into 'main'
  • 4bb3e31 🐛 fix(core): stop tiktoken rejecting text that quotes a special token
  • d830fec Merge branch 'release/knowledge-v0.2.0-changelog' into 'main'
  • 31881a3 📝 docs(knowledge): restore Unreleased changelog section
  • dc89c78 Merge branch 'fix/knowledge-json-invalid-escape' into 'main'
  • 91b11ec 🐛 fix(knowledge): recover LLM JSON whose only defect is an invalid backslash ...
  • b1fa05a Merge branch 'release/user-memory-v0.8.0rc9' into 'main'
  • 2030d42 🔖 release(user-memory): 0.8.0rc9 — the historical pass stops re-adding supers...
  • Additional commits viewable in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

  • @dependabot rebase will rebase this PR
  • @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
  • @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
  • @dependabot ignore <dependency name> major version will close this group update PR and stop Dependabot creating any more for the specific dependency's major version (unless you unignore this specific dependency's major version or upgrade to it yourself)
  • @dependabot ignore <dependency name> minor version will close this group update PR and stop Dependabot creating any more for the specific dependency's minor version (unless you unignore this specific dependency's minor version or upgrade to it yourself)
  • @dependabot ignore <dependency name> will close this group update PR and stop Dependabot creating any more for the specific dependency (unless you unignore this specific dependency or upgrade to it yourself)
  • @dependabot unignore <dependency name> will remove all of the ignore conditions of the specified dependency
  • @dependabot unignore <dependency name> <ignore condition> will remove the ignore condition of the specified dependency and ignore conditions

@dependabot
chore(deps): bump the python-deps group across 1 directory with 6 upd…
14237c6 
…ates

Updates the requirements on [lancedb](https://github.com/lancedb/lancedb), [everalgo-user-memory](https://github.com/EverMind-AI/EverAlgo), [everalgo-agent-memory](https://github.com/EverMind-AI/EverAlgo), [everalgo-knowledge](https://github.com/EverMind-AI/EverAlgo), [everalgo-boundary](https://github.com/EverMind-AI/EverAlgo) and [everalgo-core](https://github.com/EverMind-AI/EverAlgo) to permit the latest version.

Updates `lancedb` to 0.38.0
- [Release notes](https://github.com/lancedb/lancedb/releases)
- [Changelog](https://github.com/lancedb/lancedb/blob/main/release_process.md)
- [Commits](lancedb/lancedb@python-v0.34.0...v0.38.0)

Updates `everalgo-user-memory` from 0.4.0 to 0.7.0
- [Release notes](https://github.com/EverMind-AI/EverAlgo/releases)
- [Changelog](https://github.com/EverMind-AI/EverAlgo/blob/main/CHANGELOG.md)
- [Commits](EverMind-AI/EverAlgo@everalgo-user-memory/v0.4.0...everalgo-user-memory/v0.7.0)

Updates `everalgo-agent-memory` from 0.4.0 to 0.5.1
- [Release notes](https://github.com/EverMind-AI/EverAlgo/releases)
- [Changelog](https://github.com/EverMind-AI/EverAlgo/blob/main/CHANGELOG.md)
- [Commits](EverMind-AI/EverAlgo@everalgo-agent-memory/v0.4.0...everalgo-agent-memory/v0.5.1)

Updates `everalgo-knowledge` from 0.1.1 to 0.2.1
- [Release notes](https://github.com/EverMind-AI/EverAlgo/releases)
- [Changelog](https://github.com/EverMind-AI/EverAlgo/blob/main/CHANGELOG.md)
- [Commits](EverMind-AI/EverAlgo@everalgo-knowledge/v0.1.1...everalgo-knowledge/v0.2.1)

Updates `everalgo-boundary` from 0.2.1 to 0.3.0
- [Release notes](https://github.com/EverMind-AI/EverAlgo/releases)
- [Changelog](https://github.com/EverMind-AI/EverAlgo/blob/main/CHANGELOG.md)
- [Commits](EverMind-AI/EverAlgo@everalgo-boundary/v0.2.1...everalgo-boundary/v0.3.0)

Updates `everalgo-core` from 0.3.0 to 0.5.1
- [Release notes](https://github.com/EverMind-AI/EverAlgo/releases)
- [Changelog](https://github.com/EverMind-AI/EverAlgo/blob/main/CHANGELOG.md)
- [Commits](EverMind-AI/EverAlgo@everalgo-core/v0.3.0...everalgo-core/v0.5.1)

---
updated-dependencies:
- dependency-name: lancedb
  dependency-version: 0.38.0
  dependency-type: direct:production
  dependency-group: python-deps
- dependency-name: everalgo-user-memory
  dependency-version: 0.7.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
  dependency-group: python-deps
- dependency-name: everalgo-agent-memory
  dependency-version: 0.5.1
  dependency-type: direct:production
  update-type: version-update:semver-minor
  dependency-group: python-deps
- dependency-name: everalgo-knowledge
  dependency-version: 0.2.1
  dependency-type: direct:production
  update-type: version-update:semver-minor
  dependency-group: python-deps
- dependency-name: everalgo-boundary
  dependency-version: 0.3.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
  dependency-group: python-deps
- dependency-name: everalgo-core
  dependency-version: 0.5.1
  dependency-type: direct:production
  update-type: version-update:semver-minor
  dependency-group: python-deps
...

Signed-off-by: dependabot[bot] <support@github.com>
@dependabot dependabot Bot added dependencies Pull requests that update a dependency file python Pull requests that update python code labels Sep 22, 2026

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

dependencies Pull requests that update a dependency file python Pull requests that update python code

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

0 participants