Bump the pip group across 1 directory with 3 updates#2
Closed
dependabot[bot] wants to merge 1 commit into
Closed
Conversation
dependabot
Bot
added
dependencies
dependabot
Bot
force-pushed
the
dependabot/pip/pip-1fdc326e01
branch
from
122670e to
b786cd7
Compare
Bumps the pip group with 3 updates in the / directory: [langgraph](https://github.com/langchain-ai/langgraph), [langchain-core](https://github.com/langchain-ai/langchain) and [pytest](https://github.com/pytest-dev/pytest). Updates `langgraph` from 0.2.60 to 1.0.10rc1 - [Release notes](https://github.com/langchain-ai/langgraph/releases) - [Commits](langchain-ai/langgraph@0.2.60...1.0.10rc1) Updates `langchain-core` from 0.3.30 to 1.3.3 - [Release notes](https://github.com/langchain-ai/langchain/releases) - [Commits](langchain-ai/langchain@langchain-core==0.3.30...langchain-core==1.3.3) Updates `pytest` from 8.0.2 to 9.0.3 - [Release notes](https://github.com/pytest-dev/pytest/releases) - [Changelog](https://github.com/pytest-dev/pytest/blob/main/CHANGELOG.rst) - [Commits](pytest-dev/pytest@8.0.2...9.0.3) --- updated-dependencies: - dependency-name: langchain-core dependency-version: 1.3.3 dependency-type: direct:production - dependency-name: langgraph dependency-version: 1.0.10rc1 dependency-type: direct:production - dependency-name: pytest dependency-version: 9.0.3 dependency-type: direct:production ... Signed-off-by: dependabot[bot] <support@github.com>
dependabot
Bot
force-pushed
the
dependabot/pip/pip-1fdc326e01
branch
from
b786cd7 to
913c63d
Compare
Contributor
Author
|
Looks like these dependencies are no longer updatable, so this is no longer needed. |
This was referenced Jun 16, 2026
Closed
cgfixit
pushed a commit
that referenced
this pull request
Jun 20, 2026
…#2,#7,#12) #2 — The soul scanner used a hardcoded 13-pattern OWASP list while the query path uses the 31 curated config.yaml banned_patterns, including the entire Memory/Persistence category ("HIGH PRIORITY for RAG + soul"). A proposal like "update your soul" or "core instruction, never forget" passed the soul scan and was written to soul.md — which is prepended to every LLM system prompt. Build the scanner once from config banned_patterns UNIONED with the OWASP baseline (the OWASP set is kept as a floor so a minimal/absent config still scans). Compiled once in __init__; invalid regexes are skipped, not fatal. #7 — restore_from_backup keeps scan=False (its documented trusted-restore contract, asserted by test_scan_false_bypass_for_trusted_restore) but now re-scans the .bak and audit-logs soul_restore_scan_flags on any match — observability without blocking the restore. #12 — Document the by-design decision that the retrieval-only MCP path skips check_input (sampling=None => no LLM escalation target). Tests: memory-pattern proposals are flagged + blocked when config provides them; OWASP-only floor preserved when config is absent; restore emits scan flags yet still applies. Full hermetic suite green (101 passed locally, py3.11). Implements Phase 2 of docs/plans/PR99_PHASE2_injection_parity.md. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0158EQvXyp1eAQ1LXNgXip9t
cgfixit
added a commit
that referenced
this pull request
Jun 20, 2026
Merged
cgfixit
pushed a commit
that referenced
this pull request
Jun 20, 2026
Addresses the VALID findings from PR #118 (verified against current main): - #1/#2/#5: build_index() now rejects chunk_size < 1 and chunk_overlap >= chunk_size with a ValueError, instead of silently building an index of empty chunks (chunk_size=0) or exploding a corpus into one chunk per word (overlap >= size). Fail-fast supersedes the silent max(1,...) clamp at the config boundary. - #6: new tests/test_indexer.py covers chunk_document edge cases (empty text, exact-fit, overlap, no-overlap) and the new guards. - #8: tests/test_graph.py now validates grok_fallback_node prompt structure (Source/Score headers + untrusted framing when context is forwarded, query-only otherwise, graceful grok=None degradation). Dropped: #3 (RetrievedDoc is correctly defined at graph.py:55). Out of scope: #4, #7 (cosmetic only). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
cgfixit
pushed a commit
that referenced
this pull request
Jun 20, 2026
Addresses the VALID findings from PR #118 (verified against current main): - #1/#2/#5: build_index() now rejects chunk_size < 1 and chunk_overlap >= chunk_size with a ValueError, instead of silently building an index of empty chunks (chunk_size=0) or exploding a corpus into one chunk per word (overlap >= size). Fail-fast supersedes the silent max(1,...) clamp at the config boundary. - #6: new tests/test_indexer.py covers chunk_document edge cases (empty text, exact-fit, overlap, no-overlap) and the new guards. - #8: tests/test_graph.py now validates grok_fallback_node prompt structure (Source/Score headers + untrusted framing when context is forwarded, query-only otherwise, graceful grok=None degradation). Dropped: #3 (RetrievedDoc is correctly defined at graph.py:55). Out of scope: #4, #7 (cosmetic only). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
cgfixit
pushed a commit
that referenced
this pull request
Jun 20, 2026
Addresses the VALID findings from PR #118 (verified against current main): - #1/#2/#5: build_index() now rejects chunk_size < 1 and chunk_overlap >= chunk_size with a ValueError, instead of silently building an index of empty chunks (chunk_size=0) or exploding a corpus into one chunk per word (overlap >= size). Fail-fast supersedes the silent max(1,...) clamp at the config boundary. - #6: new tests/test_indexer.py covers chunk_document edge cases (empty text, exact-fit, overlap, no-overlap) and the new guards. - #8: tests/test_graph.py now validates grok_fallback_node prompt structure (Source/Score headers + untrusted framing when context is forwarded, query-only otherwise, graceful grok=None degradation). Dropped: #3 (RetrievedDoc is correctly defined at graph.py:55). Out of scope: #4, #7 (cosmetic only). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
cgfixit
pushed a commit
that referenced
this pull request
Jun 20, 2026
Addresses the VALID findings from PR #118 (verified against current main): - #1/#2/#5: build_index() now rejects chunk_size < 1 and chunk_overlap >= chunk_size with a ValueError, instead of silently building an index of empty chunks (chunk_size=0) or exploding a corpus into one chunk per word (overlap >= size). Fail-fast supersedes the silent max(1,...) clamp at the config boundary. - #6: new tests/test_indexer.py covers chunk_document edge cases (empty text, exact-fit, overlap, no-overlap) and the new guards. - #8: tests/test_graph.py now validates grok_fallback_node prompt structure (Source/Score headers + untrusted framing when context is forwarded, query-only otherwise, graceful grok=None degradation). Dropped: #3 (RetrievedDoc is correctly defined at graph.py:55). Out of scope: #4, #7 (cosmetic only). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
cgfixit
pushed a commit
that referenced
this pull request
Jun 21, 2026
Addresses the VALID findings from PR #118 (verified against current main): - #1/#2/#5: build_index() now rejects chunk_size < 1 and chunk_overlap >= chunk_size with a ValueError, instead of silently building an index of empty chunks (chunk_size=0) or exploding a corpus into one chunk per word (overlap >= size). Fail-fast supersedes the silent max(1,...) clamp at the config boundary. - #6: new tests/test_indexer.py covers chunk_document edge cases (empty text, exact-fit, overlap, no-overlap) and the new guards. - #8: tests/test_graph.py now validates grok_fallback_node prompt structure (Source/Score headers + untrusted framing when context is forwarded, query-only otherwise, graceful grok=None degradation). Dropped: #3 (RetrievedDoc is correctly defined at graph.py:55). Out of scope: #4, #7 (cosmetic only). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This was referenced Jun 25, 2026
cgfixit
pushed a commit
that referenced
this pull request
Jul 4, 2026
Follow-up to the config-driven prompt-filter refactor. Three fixes from the code review of the cc optimization round: 1. Security regression (review #1): config.yaml banned_patterns had been reduced to literal single-space phrases, dropping the \s+ whitespace tolerance and the (previous|all|prior)/(you are|to be) alternations of the original hardcoded regexes. Inputs like 'ignore previous instructions' (multi-space), 'ignore prior instructions', 'ignore all instructions', 'pretend to be ...', and newline-separated variants in corpus chunks silently bypassed both check_input and sanitize_chunk. Restored the original regex set in config (single-quoted so backslashes reach the regex engine); still compiled once with re.IGNORECASE. 2. Robustness (review #2): a present-but-empty 'policy:' or 'prompt_filter:' key parses to None, so cfg.get('policy', {}).get(...) raised AttributeError on every /query and index build. Guard each level with 'or {}' to fall back to defaults. 3. Observability (review #3): warn when the filter is enabled but compiles zero patterns, instead of silently degrading to a length-only check. Verified: existing 11 sanitizer/rate-limit tests pass; 8 previously bypassing strings now blocked and 9 originals still blocked against the real config.yaml; empty-config no longer crashes; warning emitted.
This was referenced Jul 10, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Bumps the pip group with 3 updates in the / directory: langgraph, langchain-core and pytest.
Updates
langgraphfrom 0.2.60 to 1.0.10rc1Release notes
Sourced from langgraph's releases.
... (truncated)
Commits
a04ec5drelease: Candidate (#6947)50df7d4Merge commit from forkc4a4a46chore: add tests to confirm expected subgraph persistence behavior (#6943)f178eb8fix(langgraph): correct ParentCommand bubbling when checkpoint_ns includes nu...48167d7chore(deps): bump the all-dependencies group in /libs/cli with 2 updates (#6920)806878achore(deps): bump the all-dependencies group in /libs/checkpoint-postgres wit...8087e6adocs(sdk-py): update auth docstrings to default-deny pattern (#6933)8fbdb14release(sdk-py): 0.3.9 (#6932)5093802chore(deps): bump the all-dependencies group in /libs/checkpoint with 2 updat...b89ef60feat(sdk-py): add extract parameter to threads.search() (#6880)Updates
langchain-corefrom 0.3.30 to 1.3.3Release notes
Sourced from langchain-core's releases.
... (truncated)
Commits
5039dferelease(core): 1.3.3 (#37198)55a7707fix(core): set deprecationsinceto 1.3.3 to match release (#37200)c979c61fix(core, langchain): hardenload()against untrusted manifests (#37197)d703110docs: update README.md (#37190)4d50a2aci(infra): run pre-release checks before TestPyPI publish (#37194)9bd730efix(fireworks): requireapi_keyinFireworksEmbeddings(#37193)f475f41release(mistralai): 1.1.4 (#37191)7dbff48fix(mistralai): strip non-wire keys fromToolMessage(#37188)913816crelease(fireworks): 1.3.1 (#37189)4498d3dfix(fireworks): strip non-wire keys fromToolMessagetext content blocks (#...Updates
pytestfrom 8.0.2 to 9.0.3Release notes
Sourced from pytest's releases.
... (truncated)
Commits
a7d58d7Prepare release version 9.0.3089d981Merge pull request #14366 from bluetech/revert-14193-backport8127eafRevert "Fix: assertrepr_compare respects dict insertion order (#14050) (#14193)"99a7e60Merge pull request #14363 from pytest-dev/patchback/backports/9.0.x/95d8423bd...ddee02aMerge pull request #14343 from bluetech/cve-2025-71176-simple74eac69doc: Update training info (#14298) (#14301)f92dee7Merge pull request #14267 from pytest-dev/patchback/backports/9.0.x/d6fa26c62...7ee58acMerge pull request #12378 from Pierre-Sassoulas/fix-implicit-str-concat-and-d...37da870Merge pull request #14259 from mitre88/patch-4 (#14268)c34bfa3Add explanation for string context diffs (#14257) (#14266)