Repository navigation
v0.2 foundation: roadmap, rule-ID fixes, and corpus protocol - #1
Conversation
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: d8833c32cb
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
|
|
||
| A source is eligible when: | ||
|
|
||
| - it is public, licensed for the intended analysis, or otherwise available for lawful internal evaluation; |
There was a problem hiding this comment.
Restrict corpus inputs to approved sources
When a contributor has lawful internal access to a private document but no explicit approval to use it in public repository metrics, this eligibility rule permits that source and the protocol later requires reporting derived results. Require sources contributing to repository-facing validation artifacts to be public or explicitly approved, while carving out any strictly internal-only evaluation separately.
AGENTS.md reference: AGENTS.md:L19-L19
Useful? React with 👍 / 👎.
| if ";" in line and not line.lstrip().startswith("#"): | ||
| findings.append(Finding( | ||
| "error", "STG-DE-8.1", "Semikolon verwenden Sie nicht. Teilen Sie den Satz.", | ||
| "error", "STG-8.1", "Semikolon verwenden Sie nicht. Teilen Sie den Satz.", |
There was a problem hiding this comment.
Synchronize the packaged linter after changing IDs
When users run the packaged Agent Skill, skills/simplified-technical-german/scripts/stg_lint.py remains the old generated copy and still emits nonexistent STG-DE-* IDs, while these changes and tests affect only the canonical linter. Run python tools/sync_skill.py so the bundled execution path receives the corrected IDs as well.
AGENTS.md reference: AGENTS.md:L12-L13
Useful? React with 👍 / 👎.
| @@ -0,0 +1,179 @@ | |||
| # STG-DE Roadmap | |||
There was a problem hiding this comment.
Rebuild the manifest for this file set
Adding these documents and modifying README, tests, and the canonical linter without regenerating manifest.yaml leaves the new files unlisted and retains stale hashes for the modified files. After the separate skill-sync failure is corrected, the manifest-integrity loop in tools/validate_release.py rejects this tree at the README hash, preventing release validation from passing.
AGENTS.md reference: AGENTS.md:L15-L15
Useful? React with 👍 / 👎.
|
|
||
| - **75 rules and recommendations** for German controlled writing. | ||
| - A stable rule index for all **75** STG-DE rules and recommendations. | ||
| - A traceable mapping of all **53 ASD-STE100 Issue 9 writing rules**. |
There was a problem hiding this comment.
Add the claimed ASD-STE100 mapping
The reviewed commit tree contains no mapping/ path, and a repo-wide search for ASD-STE100 traceability finds only this README claim; rules/rule-index.yaml provides internal STG rule indexing but no mapping of the 53 source rules. Add the canonical mapping data or remove the advertised feature and directory until it exists.
AGENTS.md reference: AGENTS.md:L11-L11
Useful? React with 👍 / 👎.
Summary
Start the v0.2 Evidence and Reliability phase.
Included
ROADMAP.mdcovering v0.2 through v1.0 and the parallel Agent Skill trackSTG-DE-*IDs instead of the published STG rule IDsrules/rule-index.yamlquality/RULE-COVERAGE.mdcorpus/PROTOCOL.mdImportant validator corrections
STG-DE-8.1->STG-8.1for semicolon findingsSTG-DE-1.1->STG-1.1for controlled vocabulary findingsSTG-DE-9.4->STG-9.4for project terminology findingsSTG-DE-5.1->STG-5.1for procedure sentence limitsSTG-DE-6.3->STG-6.3for description sentence limitsThis PR intentionally does not add speculative NLP enforcement. v0.2 starts by making the deterministic surface trustworthy and measurable.