docs(readme): "What it detects" capabilities banner (TrustFall + agents/features)#166
Merged
Merged
Conversation
Surfaces the TrustFall-style AI-agent-attack detection and the covered agents / built-in capabilities as a scannable checklist high in the README, for discoverability. All claims map to shipped features: - TrustFall (project-scope MCP auto-execution) — #153 - prompt-injection directives in instruction files — #155 - sandbox-disabled / auto-approve / broad-permission / remote-MCP scoring - 7 agent parsers (Claude Code, Codex, Cursor, Gemini CLI, Antigravity, Continue.dev, Claude Desktop) - SIEM integration, read-only MCP (host + fleet), fleet monitoring, signed policy distribution No tool-poisoning claim (that's #148, still Future). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Adds a scannable What it detects section near the top of the README, per discoverability advice (lead with the TrustFall keyword + a ✅ checklist of covered agents and built-in capabilities).
Why
"TrustFall-style AI agent attacks" is currently the strongest search/keyword hook, and the README buried its agent coverage and capabilities far below the fold. This puts both up top.
Accuracy
Every claim maps to a shipped feature — nothing aspirational:
Deliberately omits tool-poisoning detection — that's #148 (still Future) and would be overclaiming.
🤖 Generated with Claude Code