oscerd commented on PR #6:
URL: https://github.com/apache/camel-kit/pull/6#issuecomment-6576349931

   Thanks @luigidemasi — both corrections are applied in a32966fe, on top of a 
rebase onto `main` (f27b4702) that resolved the `CHANGELOG.md` conflict with 
the Antigravity, Qwen and Codex native Ship entries (both sides kept).
   
   **Configuration versus breach (P10, §4.11a, §4.13, V3, YAML).** The known 
Claude wildcard pre-approval is now a documented deviation, not a breach of 
P10: a report that only restates that the generated `settings.json` carries 
`mcp__<server>__*` is tracked under V3 as `VALID-HARDENING`, and duplicates of 
*that* report are closed against V3. A report that demonstrates what the grant 
lets through — a credential read or a destructive action reaching the 
developer's account without a prompt, or any other in-scope P10 boundary breach 
— is evaluated on the breach and its impact under P10's severity tiers; it is 
neither downgraded to V3 nor closed as its duplicate. P10's violation text, the 
§4.11a entry, the §4.13 `VALID-HARDENING` row and note, V3's interim triage and 
the YAML (`P10.violation_symptom`, the non-finding entry, `triage_notes`, 
`verification_items.V3`, a new `maintainer_decisions` entry) now say the same 
thing.
   
   **Process inventory (§4.5, §4.4a, P13, YAML).** The Pi row separates 
*intended* writes (the staging workspace and its session directory under the 
state root) from *reach*: `PiWorker.runRpc` sets only the working directory and 
the environment of the `setsid`/`node`/`pi` process and imposes no filesystem 
confinement, so the worker — and whatever Pi or its provider run — can write 
wherever the controller's OS account can; what Pi writes on its own account is 
external behaviour (§4.3), consistent with Q8 and §4.9. The Knowledge row and 
the §4.4a host-effects paragraph now document the classpath fallback: when 
index resolution fails and no `knowledge.index.version` is pinned, 
`LuceneSearchService.resolveIndex()` extracts a classpath-bundled index into a 
`knowledge-index*` temporary directory (JDK default temporary-directory 
permissions; deleted on orderly shutdown via `@PreDestroy`); a pinned client 
gets the resolution error instead; the monorepo runner bundles no index, so th
 e fallback serves legacy uber-jars that do. The permission and bundling 
statements are tagged *(inferred — V6)*. P13 and its YAML entry carry the same 
clause, and the acquisition bullet in §4.4a names the fallback.
   
   **V1–V6 links.** Each item now has a tracking issue — #153 (V1), #154 (V2), 
#155 (V3), #156 (V4), #157 (V5), #158 (V6) — linked from §4.14.2, the YAML 
`verification_items` (`tracking`) and the PR table; V6 also lists the related 
Knowledge hardening (#137 via #143, #139 via #142, #140). V4 and V6 remain 
open: the inventory and Knowledge statements stay *(inferred)* until you verify 
them.
   
   **Also folded in.** The Knowledge transport statements now follow 
apache/camel-kit#143 (`https`, or `http` to a loopback host only; every 
redirect target validated before the hop), so the model is not stale when both 
PRs land: §4.4a, §4.5, §4.5a, §4.7, P13, §4.9, §4.10, the §4.11 misuse pattern 
(now a loopback proxy with a cleartext upstream), the §4.11a 
`followRedirects(NORMAL)` entry (init downloaders only), §4.12 and the YAML. 
`docs/architecture.md` records the round-3 decisions and the rejected 
alternatives.
   
   Draft confidence after this round: 200 documented / 33 maintainer / 21 
inferred tags (tags of the form `*(documented — …)*`, `*(maintainer, …)*`, 
`*(inferred — …)*`). Verification: YAML parses with duplicate-key rejection; 
every `(inferred — Vn)` tag names V1–V6; relative links resolve; `git diff 
--check` and `./mvnw -B -Psourcecheck validate` pass.
   
   _AI-generated by Claude Code (Claude Fable 5.1) on behalf of @oscerd._
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to