feat(fleet): multi-runtime harness + erp#63 parity/bench evidence #69

Merged
arcodange merged 2 commits from arcodange/harness-portability into main 2026-07-18 19:59:04 +02:00
Owner

Closes #63.

The harness layer (builder sessions, cold verifiers, evidence flow) gets a committable, multi-runtime home — per the PRD model-fleet › harness portability (operator direction 2026-07-15: Claude is an escalation tier, not a hard dependency).

Ships

  • fleet/harness/verifier/ — the two canonical verifier tests (locate-test, cold-reader backlog audit): verbatim prompts, pinned inputs, ground truth, pass rules.
  • fleet/harness/bin/run-verifier.sh — runs a test against any OpenAI-style local endpoint (Ornith/MLX) or vibe -p (Mistral); sha256-pinned JSON transcripts.
  • fleet/harness/bin/vibe-builder.sh — the bounded shell for scoped builders and recurring tasks: linked-worktree guard (refuses the trunk, same structural pattern as dol-write.sh), hard --max-turns/--max-price caps, full JSON journal.
  • fleet/harness/runs/2026-07-18/ — the erp#63 evidence: 8 verifier transcripts, blind-judging verdicts (16/16 unanimous), the erp#56 builder-bench journal.
  • Same-change doc updates: fleet/README.md layout + AGENTS.md Fleet section.

Evidence highlights (full table in the #63 evidence comment): Mistral (mistral-medium-3.5 via vibe -p) and Ornith 35B (hermes MLX) reach verdict parity with the Claude baseline on both verifier tests → admitted to verifier duty; Qwen2.5-7B-4bit fails both (honest floor). Builder bench: erp#56 completed end-to-end by the Mistral runtime (0 code corrections, 261 s) → PR #68.

Paired factory PR codifies the cross-family verification rule in the PRD qa-strategy.

🤖 Generated with Claude Code

https://claude.ai/code/session_01VRShc4QhLLU73FLHx9vskh

Closes #63. The harness layer (builder sessions, cold verifiers, evidence flow) gets a committable, multi-runtime home — per the PRD [model-fleet › harness portability](https://gitea.arcodange.lab/arcodange-org/factory/src/branch/main/vibe/PRD/ai-back-office/model-fleet.md#harness-portability) (operator direction 2026-07-15: Claude is an escalation tier, not a hard dependency). **Ships** - `fleet/harness/verifier/` — the two canonical verifier tests (locate-test, cold-reader backlog audit): verbatim prompts, pinned inputs, ground truth, pass rules. - `fleet/harness/bin/run-verifier.sh` — runs a test against any OpenAI-style local endpoint (Ornith/MLX) or `vibe -p` (Mistral); sha256-pinned JSON transcripts. - `fleet/harness/bin/vibe-builder.sh` — the bounded shell for scoped builders and recurring tasks: linked-worktree guard (refuses the trunk, same structural pattern as `dol-write.sh`), hard `--max-turns`/`--max-price` caps, full JSON journal. - `fleet/harness/runs/2026-07-18/` — the erp#63 evidence: 8 verifier transcripts, blind-judging verdicts (16/16 unanimous), the erp#56 builder-bench journal. - Same-change doc updates: `fleet/README.md` layout + `AGENTS.md` Fleet section. **Evidence highlights** (full table in the [#63 evidence comment](https://gitea.arcodange.lab/arcodange-org/erp/issues/63#issuecomment-2939)): Mistral (`mistral-medium-3.5` via `vibe -p`) and Ornith 35B (hermes MLX) reach verdict parity with the Claude baseline on both verifier tests → admitted to verifier duty; Qwen2.5-7B-4bit fails both (honest floor). Builder bench: erp#56 completed end-to-end by the Mistral runtime (0 code corrections, 261 s) → PR #68. Paired factory PR codifies the cross-family verification rule in the PRD qa-strategy. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01VRShc4QhLLU73FLHx9vskh
arcodange added 2 commits 2026-07-18 19:57:38 +02:00
The harness layer (builder sessions, cold verifiers, evidence flow) gets a
committable home, per the PRD model-fleet § harness portability and erp#63:

- fleet/harness/verifier/: the two canonical verifier tests (locate-test,
  cold-reader backlog audit) with pinned inputs, verbatim prompts, ground
  truth and pass rules — judged context-free, never self-graded.
- fleet/harness/bin/run-verifier.sh: runs a test against any OpenAI-style
  local endpoint (Ornith/MLX) or vibe -p (Mistral); emits sha256-pinned
  JSON transcripts.
- fleet/harness/bin/vibe-builder.sh: the bounded shell for scoped builders
  and recurring tasks — refuses the trunk (linked-worktree guard), hard
  --max-turns/--max-price caps, full JSON journal per run.
- fleet/README.md layout + AGENTS.md Fleet section updated in the same
  change (same-change freshness rule).

Part of erp#63 (harness portability spike, D2).

Co-Authored-By: Claude Fable 5 <[email protected]>
Claude-Session: https://claude.ai/code/session_01VRShc4QhLLU73FLHx9vskh
- runs/2026-07-18/: the 8 sha256-pinned verifier transcripts (4 runtimes ×
  2 tests), blind-judging verdicts (2 independent judges per cell, unanimous),
  the erp#56 builder-bench journal + prompt + caps, and the evidence README
  with the parity table.
- run-verifier.sh: mistral runtime drops the tool-filter flag (--enabled-tools
  with a no-match pattern hangs vibe 2.21.0); plain -p with --max-turns 1.

Verdicts: Mistral (vibe -p, mistral-medium-3.5) and Ornith 35B (hermes MLX)
reach verdict parity with the Claude baseline on both tests → admitted to
verifier duty. Qwen2.5-7B-4bit fails both → the honest small-model floor.
Builder bench: erp#56 completed by the Mistral runtime, 0 code corrections,
261 s, acceptance run clean (0 bank-UNKNOWN) → merged as PR #68.

Closes #63 (with the paired factory qa-strategy PR).

Co-Authored-By: Claude Fable 5 <[email protected]>
Claude-Session: https://claude.ai/code/session_01VRShc4QhLLU73FLHx9vskh
arcodange merged commit 66b277abfa into main 2026-07-18 19:59:04 +02:00
arcodange deleted branch arcodange/harness-portability 2026-07-18 19:59:04 +02:00
Sign in to join this conversation.
No Reviewers
No labels
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: arcodange-org/erp#69