80 lines
3.9 KiB
JSON
80 lines
3.9 KiB
JSON
{
|
|
"schema": "fable.external_project.review.v1",
|
|
"createdAt": "2026-07-05T14:05:00.000Z",
|
|
"taskLabel": "T3MP3ST-pxpipe-GRC_Claw-review",
|
|
"task": "Review T3MP3ST (autonomous red-teaming), pxpipe (token compression), and GRC_Claw (GRC automation) for fable-agent applicability",
|
|
"source": {
|
|
"classification": "external-project-review",
|
|
"items": [
|
|
"https://github.com/elder-plinius/T3MP3ST",
|
|
"https://github.com/teamchong/pxpipe",
|
|
"https://github.com/AAH20/GRC_Claw"
|
|
],
|
|
"contentVerified": true,
|
|
"liveVerification": "GitHub API metadata + raw README fetched at review time"
|
|
},
|
|
"externalProjectReviews": [
|
|
{
|
|
"repo": "elder-plinius/T3MP3ST",
|
|
"stars": 847,
|
|
"license": "AGPL-3.0",
|
|
"description": "Multi-agent offensive-security meta-harness for autonomous red-teaming",
|
|
"keyFindings": [
|
|
"verify-claims pattern (24/24 green) \u00e2\u20ac\u201d identical to fable-agent's receipt-first philosophy",
|
|
"Keyless \u00e2\u20ac\u201d uses local agent as backbone, same as fable-agent's model-agnostic pipeline",
|
|
"90.1% pass@1 on XBEN security benchmark",
|
|
"AGPL-3.0 \u00e2\u20ac\u201d copyleft, cannot merge code"
|
|
],
|
|
"safeToVendor": false,
|
|
"decision": "learn-only \u00e2\u20ac\u201d verify-claims validates fable-agent's approach"
|
|
},
|
|
{
|
|
"repo": "teamchong/pxpipe",
|
|
"stars": 2383,
|
|
"license": "MIT",
|
|
"description": "Cut Fable 5 token usage by rendering text context as images (~59-70% lower bill)",
|
|
"keyFindings": [
|
|
"~3.1 chars per image-token vs ~1 char per text-token on dense content",
|
|
"10/10 SWE-bench Lite both arms at -65% request size",
|
|
"Lossy: 13/15 hex strings on Fable 5, 0/15 on Opus \u00e2\u20ac\u201d silent confabulations",
|
|
"Escape hatch for byte-exact content (keeps lossy items as text)"
|
|
],
|
|
"safeToVendor": true,
|
|
"decision": "advisory \u00e2\u20ac\u201d token compression relevant but lossy nature requires careful routing"
|
|
},
|
|
{
|
|
"repo": "AAH20/GRC_Claw",
|
|
"stars": 2,
|
|
"license": "MIT",
|
|
"description": "ISO 42001-compliant agentic AI governance engine \u00e2\u20ac\u201d swarm harness, anti-swarm WAF, MAVLink/UAS compliance, a2zsoc.com bridge",
|
|
"keyFindings": [
|
|
"Massive scope \u00e2\u20ac\u201d 90 packages, 375+ cross-framework mappings, Terraform provider, VS Code extension, MCP server",
|
|
"Primarily a GRC (Governance, Risk, Compliance) platform, not an agent harness",
|
|
"Relevant package: Agent Policy Firewall + agent attestation ledger for MCP/AI tool calls",
|
|
"Commercial A2Z SOC hosted platform \u00e2\u20ac\u201d the OSS chassis is the inspectable engine",
|
|
"Very early (2 stars) but well-structured with CLI, SDK, MCP server"
|
|
],
|
|
"relevantPatterns": [
|
|
"Agent Policy Firewall \u00e2\u20ac\u201d governs tool calls, model routing, memory (same problem fable-agent solves)",
|
|
"Agent attestation ledger \u00e2\u20ac\u201d logs model/tool/memory decisions without exposing sensitive data",
|
|
"Evidence graph with verifier network \u00e2\u20ac\u201d similar to fable-agent's receipt verification"
|
|
],
|
|
"safeToVendor": true,
|
|
"decision": "advisory \u00e2\u20ac\u201d GRC_Claw's agent policy firewall and attestation ledger are directly relevant to fable-agent's governance pattern"
|
|
}
|
|
],
|
|
"fableAgentAlignment": [
|
|
"All three repos validate fable-agent's approach from different angles:",
|
|
"T3MP3ST: verify-claims pattern = receipt-first philosophy",
|
|
"pxpipe: token compression for pipeline context (repo maps, tool docs)",
|
|
"GRC_Claw: Agent Policy Firewall + attestation ledger = same governance problem"
|
|
],
|
|
"skipped": [
|
|
"actual integration or vendoring",
|
|
"benchmark comparisons",
|
|
"deploy actions"
|
|
],
|
|
"deployAttempted": false,
|
|
"autoPatchProven": false,
|
|
"updatedAt": "2026-07-05T14:25:00.000Z"
|
|
} |