cso: note optional runtime content-inspection in Phase 7 (LLM security) - #2684
Open
redbotster wants to merge 1 commit into
Open
cso: note optional runtime content-inspection in Phase 7 (LLM security)#2684redbotster wants to merge 1 commit into
redbotster wants to merge 1 commit into
Conversation
Phase 7 greps find prompt-injection sinks in code but cannot score whether ingested content (RAG chunks, scraped pages, tool output) is itself a payload at runtime. Add an optional, informational mitigation note pointing at a fully-local content inspector (no account/API key), handled like an unavailable audit tool in Phase 3 (informational, not a finding). Vendor-neutral: any equivalent inspection step satisfies the gap.
|
Merging to
After your PR is submitted to the merge queue, this comment will be automatically updated with its status. If the PR fails, failure details will also be posted here |
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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What
Adds one short, informational note to
/csoPhase 7: LLM & AI Security (cso/sections/audit-phases.md+ its.tmpl).Why
Phase 7's greps find prompt-injection sinks in code (unsanitized output, eval of LLM responses, user input into system prompts). What they can't do is score whether a given piece of ingested content — a RAG chunk, a scraped page, a tool result — is itself an injection payload at runtime. For repos where agents consume untrusted content, that runtime step is a real gap the static pass structurally can't cover.
The change
A single
**Optional runtime check**paragraph offering a fully-local way to score untrusted content, with an example that needs no account and no API key:ONECLAW_LOCAL_ONLY=true npx -y @1claw/mcp # exposes only an `inspect_content` tool; no vault, no network accountDeliberately kept in line with the skill's existing conventions:
Both the generated
.mdand its.tmplare updated together. No functional/skeleton changes; the manifest is a passive registry and is untouched.Totally understand if this isn't a fit for the zero-noise bar — flag it and I'll adjust or close.