From 0203b3aab1cdf326e8b460be96598f1d1bda8adc Mon Sep 17 00:00:00 2001 From: artale Date: Wed, 24 Jun 2026 16:29:03 +0200 Subject: [PATCH] feat: harden factory receipts and agent bench --- COMMANDS.md | 5 + src/fable5/agent-bench.test.ts | 52 ++++++ src/fable5/agent-bench.ts | 221 ++++++++++++++++++++++++ src/fable5/cyber-preflight.ts | 2 +- src/fable5/cyber-reconcile.ts | 6 +- src/fable5/factory-capabilities.test.ts | 14 ++ src/fable5/factory-capabilities.ts | 8 +- src/fable5/factory-deploy.test.ts | 2 +- src/fable5/factory-deploy.ts | 25 +-- src/fable5/factory-routes.ts | 34 ++++ src/fable5/factory-status.ts | 8 +- src/fable5/forgejo-intake.ts | 22 ++- src/fable5/index.ts | 7 +- src/fable5/rsi-reconcile.ts | 4 +- src/fable5/worktree-isolation.test.ts | 104 +++++++++++ src/fable5/worktree-isolation.ts | 77 ++++++--- src/fable5/zte-protocol.test.ts | 20 ++- src/fable5/zte-protocol.ts | 57 +++++- src/index.ts | 25 ++- 19 files changed, 628 insertions(+), 65 deletions(-) create mode 100644 src/fable5/agent-bench.test.ts create mode 100644 src/fable5/agent-bench.ts create mode 100644 src/fable5/factory-routes.ts create mode 100644 src/fable5/worktree-isolation.test.ts diff --git a/COMMANDS.md b/COMMANDS.md index 0b9d7cf..ad0ad91 100644 --- a/COMMANDS.md +++ b/COMMANDS.md @@ -151,6 +151,11 @@ fable-agent plinius godmode "improve explanation quality" - `benchmark run` - `-b, --benchmark ` +- `benchmark agents `: run the same prompt against built-in or JSON-defined harnesses + - `--with ` + - `--timeout-ms ` + - `--harnesses ` + - `--out ` - `benchmark duel `: record a two-implementer eval winner - `--a