Repository navigation
feat(security): pre-action content firewall over MCP/web/RAG/file/stdout untrusted inputs (#10552) - #10758
Merged
Conversation
…out untrusted inputs (#10552) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Contributor
✅ SSOT Configuration Compliance: Passing🎉 No hardcoded values detected that have SSOT config equivalents! |
Applied Black, isort, and autoflake to match code-quality checks. Triggered by workflow auto-fix-formatting.yml. Co-Authored-By: Paperclip <noreply@paperclip.ing>
4 tasks
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Thinking Path
Agent-framework epic #10542: wire the existing PromptInjectionDetector over all untrusted inputs (security, highest-risk gap).
What Changed
New
security/content_firewall.py— one shared firewall run at every untrusted-input choke point: parallel executor (stdout/file), MCP client (call_tool + read_resource), web_fetch extract_url, RAG context assembly. Risk policy (env-tunable QUARANTINE/BLOCK/ESCALATE thresholds); untrusted spans delimited as DATA with a system note; high-risk recorded in the trajectory.Verification
12 tests pass (seeded-injection quarantine/block, escalate, benign no-false-positive, singleton reuse, provenance). Closes #10552.
Model Used
Claude Opus 4.8 (subagent).