Repository navigation
Regression plan includes entries from releases the matrix skips - #219
Conversation
… notes Semantic review on #218 warned that the 'both constructors changed' claims in MAF11-SKILL-INLINE-001 and MAF11-SKILL-PROVIDER-001 were not grounded in the notes; the diff shows both overloads changing, now quoted. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The plan's steps follow the compatibility matrix, which has no rows for 1.6.0, 1.6.2, 1.7.0, 1.8.0 or 1.9.0. Each step listed only entries with VersionIntroduced == step, so a 1.7.0 -> 1.10.0 plan (one step) dropped the 1.8.0/1.9.0 entries backfilled in #217 (session isolation default, declarative code-gen removal, Harness). A step now lists entries introduced in (source, step], still evaluated against the source codebase. New test: Plan_StepOverSkippedReleases_IncludesTheirEntries. 1318+1 tests. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
🟢 MAF health grade: Ano errors, no warnings, no silent-starvation risks, all agent call sites bounded Repo:
✅ No fixes recommended — repo is clean against all configured rules. Want more
Audited commit: |
|
✅ AI-fill verification passed (registry + cross-file + scaffold-preservation checks). Audited commit: |
|
Reviewed head: ✅ AI Semantic Review — verdict: PASSReviewed 2 changed registry entries: All filled entries are semantically consistent with the This is rung-2 of the AI trust ladder (comment-only). The human still merges. To turn this into a hard gate, flip ✅
|
🔍 Cisco mcp-scanner resultScan covers prompt-injection, tool-poisoning, credential-harvesting, Audited commit: |
There was a problem hiding this comment.
Copilot review overview
🟡 Changes recommended
The new test must follow the repository’s mandatory naming and AAA structure.
Review effort: Balanced
Findings: 1
Open (1)
What changed in this PR
Updates regression-plan generation so skipped compatibility-matrix releases still contribute registry entries.
Changes:
- Selects entries from the full
(source, target]version range. - Adds regression coverage for skipped releases.
- Clarifies registry evidence and records the fix.
| File | Description |
|---|---|
RegressionPlanTool.cs |
Adds range-based registry selection. |
RegressionPlanToolTests.cs |
Tests skipped-release handling. |
CHANGELOG.md |
Documents the regression fix. |
registry.yaml |
Expands 1.1.0 overload evidence. |
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
| public void Plan_StepOverSkippedReleases_IncludesTheirEntries() | ||
| { | ||
| // The compatibility matrix has no 1.8.0/1.9.0 rows, so 1.7.0 -> 1.10.0 is | ||
| // one step. Entries introduced in the skipped releases must still appear. | ||
| var result = _tool.MafGenerateRegressionPlan("1.7.0", "1.10.0"); | ||
| Assert.Contains("`MAF180-HOSTING-STORE-001`", result); | ||
| Assert.Contains("`MAF190-HARNESS-AGENT-001`", result); | ||
| Assert.Contains("`MAF110-HARNESS-AGENT-001`", result); | ||
| // Already-passed releases (1.6.x and earlier) must not appear. | ||
| Assert.DoesNotContain("`MAF162-FOUNDRY-EXTENSIONS-001`", result); | ||
| } |
Bug:
maf_generate_regression_plan's steps follow the compatibility matrix, which has no rows for 1.6.0, 1.6.2, 1.7.0, 1.8.0 or 1.9.0. Each step listed only entries introduced exactly at that step's version. So a 1.7.0 → 1.10.0 plan, which is one step, dropped the 1.8.0/1.9.0 entries backfilled in #217: the session-isolation default, the declarative code-gen removal, and the Harness changes.Fix: a step now lists the entries introduced in (source, step], still evaluated against the source codebase.
Tests: new
Plan_StepOverSkippedReleases_IncludesTheirEntries; full suite passes (1319 tests).Also in this PR: the notes fix for the two 1.1.0 entries that the semantic review warned about on #218.
🤖 Generated with Claude Code