Revise README to focus on deterministic operational replay and add Comptextv7 logo#80
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
✅ Deploy Preview for comptext-v7 canceled.
|
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request significantly refactors the README.md to focus on deterministic operational replay validation for AI agents, replacing the previous research-oriented narrative with a more concise, benchmark-driven structure. It introduces a new project logo and updates the repository map and reproducibility instructions. Feedback focuses on improving readability by rounding high-precision benchmark values and restoring a more detailed repository map to aid developer navigation.
| | Paper avg compression | 1.347063 | | ||
| | Agent avg compression | 1.773954 | | ||
| | Paper replay consistency | 0.791667 | | ||
| | Agent replay consistency | 1.000000 | | ||
| | Agent operational drift | 0.000000 | |
There was a problem hiding this comment.
The benchmark values in the 'Proof at a glance' table are provided with 6 decimal places of precision. For a high-level summary in a README, this level of detail is difficult to read and increases the maintenance burden for future updates. Consider rounding these values to 3 decimal places to match the formatting used in the 'Complementary adversarial replay stress suite' table later in the document.
| | Paper avg compression | 1.347063 | | |
| | Agent avg compression | 1.773954 | | |
| | Paper replay consistency | 0.791667 | | |
| | Agent replay consistency | 1.000000 | | |
| | Agent operational drift | 0.000000 | | |
| | Paper avg compression | 1.347 | | |
| | Agent avg compression | 1.774 | | |
| | Paper replay consistency | 0.792 | | |
| | Agent replay consistency | 1.000 | | |
| | Agent operational drift | 0.000 | |
| The committed 250-iteration report records Comptextv7 mean final continuity at `0.571783`, rounded to `0.572` here. | ||
| Detail fidelity still degrades: hidden truth survival is `0.570173`, and evaluator agreement divergence is `0.421743`. |
There was a problem hiding this comment.
The values in this section are provided with high precision (6 decimal places), which is inconsistent with the rounding mentioned in the text ('rounded to 0.572 here'). For better readability and consistency, consider using the rounded values throughout the narrative.
Example:
The committed 250-iteration report records Comptextv7 mean final continuity at `0.572`.
Detail fidelity still degrades: hidden truth survival is `0.570`, and evaluator agreement divergence is `0.422`.|
|
||
| Positioning statement: | ||
| ## Repository map | ||
| Core review surfaces live in [`artifacts/`](artifacts/), [`docs/benchmarks/`](docs/benchmarks/), [`reports/replay_continuity/`](reports/replay_continuity/), and [`tests/`](tests/). |
There was a problem hiding this comment.
The previous comprehensive repository map (directory tree) has been replaced by a single sentence focusing only on 'review surfaces'. This significantly reduces the discoverability of the project's internal structure for new developers. It is recommended to restore a simplified version of the directory tree to show where core components like src/, scripts/, and dashboard/ are located.
Example:
Comptextv7/
├── artifacts/ # deterministic benchmark results
├── benchmarks/ # compression and replay runners
├── dashboard/ # operations console
├── docs/ # documentation and benchmarks
├── reports/replay_continuity/ # adversarial metrics and charts
├── scripts/ # validation and artifact tooling
├── src/ # core KVTC engine and agents
├── tests/ # regression and validation tests
└── README.md
1 participant
Motivation
Description
README.mdto refocus project positioning toward "Deterministic operational replay" and reorganized sections (why it exists, architecture, benchmarks, reproducibility, limitations, next steps, repository map, safety boundaries).artifacts/,docs/benchmarks/, andreports/replay_continuity/, updated command examples and test/run guidance to the current scripts and pytest targets.docs/assets/comptextv7-logo.svgand updated the README to reference it.python tests/utils/paper_replay_runner.py,python tests/utils/agent_trace_replay_runner.py,benchmarks/run_replay_continuity.py, andpytest tests/test_paper_replay_bench.py tests/test_agent_trace_replay.py tests/test_replay_continuity.py.Testing
python -m pytest, includingtests/test_paper_replay_bench.py,tests/test_agent_trace_replay.py, andtests/test_replay_continuity.py. All tests completed successfully.docs/assets/comptextv7-logo.svgrenders as a static asset in the docs tree (visual inspection in the dev environment).Codex Task