* fix: harden NL2SQLTool — read-only by default, parameterized queries, query validation
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: address CI lint failures and remove unused import
- Remove unused `sessionmaker` import from test_nl2sql_security.py
- Use `Self` return type on `_apply_env_override` (fixes UP037/F821)
- Fix ruff errors auto-fixed in lib/crewai (UP007, etc.)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: expand _WRITE_COMMANDS and block multi-statement semicolon injection
- Add missing write commands: UPSERT, LOAD, COPY, VACUUM, ANALYZE,
ANALYSE, REINDEX, CLUSTER, REFRESH, COMMENT, SET, RESET
- _validate_query() now splits on ';' and validates each statement
independently; multi-statement queries are rejected outright in
read-only mode to prevent 'SELECT 1; DROP TABLE users' bypass
- Extract single-statement logic into _validate_statement() helper
- Add TestSemicolonInjection and TestExtendedWriteCommands test classes
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* ci: retrigger
* fix: use typing_extensions.Self for Python 3.10 compat
* chore: update tool specifications
* docs: document NL2SQLTool read-only default and DML configuration
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: close three NL2SQLTool security gaps (writable CTEs, EXPLAIN ANALYZE, multi-stmt commit)
- Remove WITH from _READ_ONLY_COMMANDS; scan CTE body for write keywords so
writable CTEs like `WITH d AS (DELETE …) SELECT …` are blocked in read-only mode.
- EXPLAIN ANALYZE/ANALYSE now resolves the underlying command; EXPLAIN ANALYZE DELETE
is treated as a write and blocked in read-only mode.
- execute_sql commit decision now checks ALL semicolon-separated statements so
a SELECT-first batch like `SELECT 1; DROP TABLE t` still triggers a commit
when allow_dml=True.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: handle parenthesized EXPLAIN options syntax; remove unused _seed_db
_validate_statement now strips parenthesized options from EXPLAIN (e.g.
EXPLAIN (ANALYZE) DELETE, EXPLAIN (ANALYZE, VERBOSE) DELETE) before
checking whether ANALYZE/ANALYSE is present — closing the bypass where
the options-list form was silently allowed in read-only mode.
Adds three new tests:
- EXPLAIN (ANALYZE) DELETE → blocked
- EXPLAIN (ANALYZE, VERBOSE) DELETE → blocked
- EXPLAIN (VERBOSE) SELECT → allowed
Also removes the unused _seed_db helper from test_nl2sql_security.py.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* chore: update tool specifications
* fix: smarter CTE write detection, fix commit logic for writable CTEs
- Replace naive token-set matching with positional AS() body inspection
to avoid false positives on column names like 'comment', 'set', 'reset'
- Fix execute_sql commit logic to detect writable CTEs (WITH + DELETE/INSERT)
not just top-level write commands
- Add tests for false positive cases and writable CTE commit behavior
- Format nl2sql_tool.py to pass ruff format check
* fix: catch write commands in CTE main query + handle whitespace in AS()
- WITH cte AS (SELECT 1) DELETE FROM users now correctly blocked
- AS followed by newline/tab/multi-space before ( now detected
- execute_sql commit logic updated for both cases
- 4 new tests
* fix: EXPLAIN ANALYZE VERBOSE handling, string literal paren bypass, commit logic for EXPLAIN ANALYZE
- EXPLAIN handler now consumes all known options (ANALYZE, ANALYSE, VERBOSE) before
extracting the real command, fixing 'EXPLAIN ANALYZE VERBOSE SELECT' being blocked
- Paren walker in _extract_main_query_after_cte now skips string literals, preventing
'WITH cte AS (SELECT '\''('\'' FROM t) DELETE FROM users' from bypassing detection
- _is_write_stmt in execute_sql now resolves EXPLAIN ANALYZE to underlying command
via _resolve_explain_command, ensuring session.commit() fires for write operations
- 10 new tests covering all three fixes
* fix: deduplicate EXPLAIN parsing, fix AS( regex in strings, block unknown CTE commands, bump langchain-core
- Refactor _validate_statement to use _resolve_explain_command (single source of truth)
- _iter_as_paren_matches skips string literals so 'AS (' in data doesn't confuse CTE detection
- Unknown commands after CTE definitions now blocked in read-only mode
- Bump langchain-core override to >=1.2.28 (GHSA-926x-3r5x-gfhw)
* fix: add return type annotation to _iter_as_paren_matches
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* fix: add path and URL validation to RAG tools
Add validation utilities to prevent unauthorized file reads and SSRF
when RAG tools accept LLM-controlled paths/URLs at runtime.
Changes:
- New crewai_tools.utilities.safe_path module with validate_file_path(),
validate_directory_path(), and validate_url()
- File paths validated against base directory (defaults to cwd).
Resolves symlinks and ../ traversal. Rejects escape attempts.
- URLs validated: file:// blocked entirely. HTTP/HTTPS resolves DNS
and blocks private/reserved IPs (10.x, 172.16-31.x, 192.168.x,
127.x, 169.254.x, 0.0.0.0, ::1, fc00::/7).
- Validation applied in RagTool.add() — catches all RAG search tools
(JSON, CSV, PDF, TXT, DOCX, MDX, Directory, etc.)
- Removed file:// scheme support from DataTypes.from_content()
- CREWAI_TOOLS_ALLOW_UNSAFE_PATHS=true env var for backward compat
- 27 tests covering traversal, symlinks, private IPs, cloud metadata,
IPv6, escape hatch, and valid paths/URLs
* fix: validate path/URL keyword args in RagTool.add()
The original patch validated positional *args but left all keyword
arguments (path=, file_path=, directory_path=, url=, website=,
github_url=, youtube_url=) unvalidated, providing a trivial bypass
for both path-traversal and SSRF checks.
Applies validate_file_path() to path/file_path/directory_path kwargs
and validate_url() to url/website/github_url/youtube_url kwargs before
they reach the adapter. Adds a regression-test file covering all eight
kwarg vectors plus the two existing positional-arg checks.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: address CodeQL and review comments on RAG path/URL validation
- Replace insecure tempfile.mktemp() with inline symlink target in test
- Remove unused 'target' variable and unused tempfile import
- Narrow broad except Exception: pass to only catch urlparse errors;
validate_url ValueError now propagates instead of being silently swallowed
- Fix ruff B904 (raise-without-from-inside-except) in safe_path.py
- Fix ruff B007 (unused loop variable 'family') in safe_path.py
- Use validate_directory_path in DirectorySearchTool.add() so the
public utility is exercised in production code
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* style: fix ruff format + remaining lint issues
* fix: resolve mypy type errors in RAG path/URL validation
- Cast sockaddr[0] to str() to satisfy mypy (socket.getaddrinfo returns
sockaddr where [0] is str but typed as str | int)
- Remove now-unnecessary `type: ignore[assignment]` and
`type: ignore[literal-required]` comments in rag_tool.py
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: unroll dynamic TypedDict key loops to satisfy mypy literal-required
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* test: allow tmp paths in RAG data-type tests via CREWAI_TOOLS_ALLOW_UNSAFE_PATHS
TemporaryDirectory creates files under /tmp/ which is outside CWD and is
correctly blocked by the new path validation. These tests exercise
data-type handling, not security, so add an autouse fixture that sets
CREWAI_TOOLS_ALLOW_UNSAFE_PATHS=true for the whole file. Path/URL
security is covered by test_rag_tool_path_validation.py.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* test: allow tmp paths in search-tool and rag_tool tests via CREWAI_TOOLS_ALLOW_UNSAFE_PATHS
test_search_tools.py has tests for TXTSearchTool, CSVSearchTool,
MDXSearchTool, JSONSearchTool, and DirectorySearchTool that create
files under /tmp/ via tempfile, which is outside CWD and correctly
blocked by the new path validation. rag_tool_test.py has one test
that calls tool.add() with a TemporaryDirectory path.
Add the same autouse allow_tmp_paths fixture used in
test_rag_tool_add_data_type.py. Security is covered separately by
test_rag_tool_path_validation.py.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* chore: update tool specifications
* docs: document CodeInterpreterTool removal and RAG path/URL validation
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: address three review comments on path/URL validation
- safe_path._is_private_or_reserved: after unwrapping IPv4-mapped IPv6
to IPv4, only check against IPv4 networks to avoid TypeError when
comparing an IPv4Address against IPv6Network objects.
- safe_path.validate_file_path: handle filesystem-root base_dir ('/')
by not appending os.sep when the base already ends with a separator,
preventing the '//'-prefix bug.
- rag_tool.add: path-detection heuristic now checks for both '/' and
os.sep so forward-slash paths are caught on Windows as well as Unix.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: remove unused _BLOCKED_NETWORKS variable after IPv4/IPv6 split
* chore: update tool specifications
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
* docs: update quickstart and installation guides for improved clarity
- Revised the quickstart guide to emphasize creating a Flow and running a single-agent crew that generates a report.
- Updated the installation documentation to reflect changes in the quickstart process and enhance user understanding.
* translations
* docs: add AMP Training Tab guide for enterprise deployments
* docs: add training guide translations for ar, ko, pt-BR
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
---------
Co-authored-by: Alex <alex@crewai.com>
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
* docs: Add coding agent skills demo video to getting started pages
Add Loom demo video embed showing how to build CrewAI agents and flows
using coding agent skills. Added to introduction, quickstart, and
installation pages across all languages (en, ko, pt-BR, ar).
* docs: update coding skills description with install instructions
Replace demo description text with actionable install copy across
all languages (en, ko, pt-BR, ar) in introduction, quickstart, and
installation pages.
- Apps run locally (with CREWAI_PLATFORM_INTEGRATION_TOKEN env var), not remotely
- Apps auth is an integration token, not OAuth
- Updated comparison tables and card descriptions in en, pt-BR, ko, ar
* docs: add Agent Capabilities overview page and improve Skills docs
- New 'Agent Capabilities' page explaining all 5 extension types (Tools, MCPs, Apps, Skills, Knowledge) with comparison table and decision guide
- Rewrite Skills page with practical examples showing Skills + Tools patterns, common FAQ, and Skills vs Knowledge comparison
- Add cross-reference callout on Tools page linking to the capabilities overview
- Add agent-capabilities to Core Concepts navigation (after agents)
* docs: add pt-BR and ko translations for agent-capabilities and updated skills/tools
* docs: add Arabic (ar) translations for agent-capabilities and updated skills/tools
