Hi maintainers — sharing a small third-party static architecture audit in case it is useful. Please feel free to close if this is not a useful tracking item for FoundationAgents/MetaGPT.
I ran agchk against the current repository with the personal-development profile:
agchk /path/to/MetaGPT --profile personal
This is not a vulnerability disclosure and not a request to adopt another tool. It is a static architecture/methodology scan intended to surface agent-runtime complexity patterns.
Summary
- Architecture era:
铁器时代, 45/100
- Share line:
这个 Agent 项目处于 铁器时代（45/100）：具备较清晰的工具、记忆和技能分层，开始可维护。
- Overall health signal:
unstable
- Finding counts:
critical=0, high=4, medium=33, low=43
Top static signals
- HIGH: Internal orchestration sprawl detected
- HIGH: Completion closure gap detected
- HIGH: Memory freshness / generation confusion detected
- HIGH: Runtime surface sprawl detected
- MEDIUM: Impression pointers missing
- MEDIUM: Role-play handoff orchestration detected
Caveats
- Because this project intentionally explores multi-agent organization patterns, role/handoff findings should be read as architecture review prompts rather than immediate defects.
- Static regex-style scans can over-count docs, examples, generated fixtures, provider adapters, or intentionally sandboxed execution paths.
- I did not run project tests or dynamic behavior checks; this is only a lightweight static architecture view.
Why I am filing this
The goal is to share an external lens on agent architecture maturity: orchestration drag, memory freshness, completion closure, tool/runtime boundaries, and methodology density.
If any signal is useful, I can follow up with a narrowly scoped docs proposal or convert false positives into agchk regression tests so future scans are less noisy for mature agent projects like this one.
Hi maintainers — sharing a small third-party static architecture audit in case it is useful. Please feel free to close if this is not a useful tracking item for FoundationAgents/MetaGPT.
I ran
agchkagainst the current repository with the personal-development profile:This is not a vulnerability disclosure and not a request to adopt another tool. It is a static architecture/methodology scan intended to surface agent-runtime complexity patterns.
Summary
铁器时代,45/100这个 Agent 项目处于 铁器时代（45/100）：具备较清晰的工具、记忆和技能分层，开始可维护。unstablecritical=0,high=4,medium=33,low=43Top static signals
Caveats
Why I am filing this
The goal is to share an external lens on agent architecture maturity: orchestration drag, memory freshness, completion closure, tool/runtime boundaries, and methodology density.
If any signal is useful, I can follow up with a narrowly scoped docs proposal or convert false positives into agchk regression tests so future scans are less noisy for mature agent projects like this one.