agent-harness GitHub
Menu

Overview

agent-harness Agent Harness Blueprint Blueprint plugin product brief The Harness Book Deny-floor known limitations Licensing scope and decision Replay tool product brief Agent Harness roadmap Harness Specs

Docs

Advisory agent-operations measurements Mapped authority-document drift probe Mapped authority drift: pinned local trial Benchmarks and measured evidence Corpus path boundary Advisory entropy measurements Learning evaluator evidence boundaries Outcome stream validation System comparisons and canary evidence Advisory logical-job accounting Replay lab Python check inventory Replay CLI input identity and timeout limits Replay input decoding failures Replay report output boundary Replay process policy-root aliases Replay supervisor marker scope Review evidence and acceptance oracles Offline reviewer-packet diagnostics System state Process-fixture startup is not containment evidence

Agentic

Architecture additives — perfect assistant → estate owners Perfect autonomous agentic assistant — estate synthesis

Evals

Harness extension notes Evaluation merge-boundary contract: v0 Merge-gate model Outcome ledger Review under flood Agentic development evaluation taxonomy: v0

Evals / Examples

Worked evidence packet: a private product PR

Extraction

Replay v0 extraction inventory

Freeze

Legacy deny-floor v1 freeze record

Maintenance

Managed Codex agent publication Sync input and Doctor session roots

Observability

Coding-agent loop observability v0 Offline source-record identity inspector

Observability / Adapters

Claude Code to agent-loop v0: mapping note

Review evidence

Review/oracle wave: source ledger

UX evaluation

Frozen observation verifier Evidence-bound issue seeds Advisory product UX evaluation Offline scenario binder Scenario authoring contract v0 Taskdeck-first adapter sketch Workspace handoff acceptance profile Workspace receipt lab

Replay input decoding failures

Follow-up to #145. Source-output failures remain distinct from invalid inputs: #334 preserves reports when a running policy emits invalid decision JSON; this change rejects invalid corpus or recorded-sidecar JSON before policy execution.

Boundary

Translate JSON decoder ValueError at the existing corpus-manifest, corpus-JSONL and recorded-sidecar boundaries. Python integer-conversion limits are not raised or disabled. Existing duplicate-key and schema errors keep their more specific diagnostics instead of being replaced by a generic decoding message.

The CLI returns input-invalid exit 2, writes a bounded diagnostic, and neither launches a policy nor creates a report directory for these input failures. Both replay and validate use the corpus boundary. Direct RecordedDecisionSource use returns recording-manifest-invalid with all decisions indeterminate when its sidecar cannot be decoded. The direct source API has no CLI exit status.

Controls

python -m unittest tests.test_replay_input_decoding -v exercises actual CLI subprocesses for an oversized integer in the corpus manifest, recorded sidecar, events and cases. JSONL digests are refreshed so tests reach decoding rather than fail for unrelated stale bytes. A valid fixture proves that the candidate can launch and publish all three reports. Additional controls preserve JSON syntax, UTF-8, duplicate-key and recorded-schema diagnostics, and exercise the direct recorded-source API. Only test-child integer limits are tightened.

No schema, production interpreter limit, process-launch protocol, timeout, policy effect, floor, permission or merge gate changes. This is not a universal exception handler or a memory/resource sandbox. Other #145 cases retain their own acceptance evidence; this document does not close them by implication.