feat: add test-driven assessment workflow for policy generation#124
feat: add test-driven assessment workflow for policy generation#124jpower432 wants to merge 3 commits into
Conversation
jpower432
force-pushed
the
feat/test-driven-assessment
branch
from
f05222d to
de5f077
Compare
Add ADR 018, test-driven-assessment process skill, and using-complypack router skill to enforce test-first policy generation from MCP data. Assisted-by: Claude (Anthropic, Claude Opus 4.6) Signed-off-by: Jennifer Power <barnabei.jennifer@gmail.com>
Single mode (default) walks through each requirement one at a time with interactive pause points using AskUserQuestion. Batch mode processes all requirements at once in two phases. Assisted-by: Claude (Anthropic, Claude Opus 4.6) Signed-off-by: Jennifer Power <barnabei.jennifer@gmail.com>
jpower432
force-pushed
the
feat/test-driven-assessment
branch
from
de5f077 to
6df7fec
Compare
- Add invocable: false to test-driven-assessment and pack-assessment - Update router, mcp-setup, and OpenCode symlinks/commands to reference comply:build-assessment as the entry point - Add volume mount guidance for file:// sources in mcp-setup Assisted-by: Claude (Anthropic, Claude Opus 4.6) Signed-off-by: Jennifer Power <barnabei.jennifer@gmail.com>
jpower432
force-pushed
the
feat/test-driven-assessment
branch
from
6df7fec to
5d9d387
Compare
jpower432
requested review from
em-redhat and
gxmiranda
and removed request for
a team
gxmiranda
left a comment
gxmiranda
left a comment
There was a problem hiding this comment.
PR Review: #124 — feat: add test-driven assessment workflow for policy generation
The test-driven workflow design is sound — test-first generation catches semantic errors that structural validation misses, and the ADR provides clear rationale. However, two missing symlinks leave the skill discovery and delegation chain incomplete.
1 HIGH, 1 MEDIUM finding. See inline comments.
This review was generated by /review-pr (AI-assisted).
| @@ -0,0 +1 @@ | |||
| ../../skills/build-assessment No newline at end of file | |||
There was a problem hiding this comment.
[HIGH] Missing symlinks for using-complypack and test-driven-assessment
This PR adds skills/using-complypack/SKILL.md and skills/test-driven-assessment/SKILL.md but does not create corresponding symlinks under .opencode/skills/. The build-assessment skill gets a symlink here, but these two do not.
OpenCode discovers skills through .opencode/skills/ symlinks. Without them:
test-driven-assessmentcannot be loaded whenbuild-assessmentdelegates to itusing-complypackcannot function as a router skill
Please add:
ln -sf ../../skills/test-driven-assessment .opencode/skills/test-driven-assessment
ln -sf ../../skills/using-complypack .opencode/skills/using-complypack| @@ -2,5 +2,5 @@ | |||
| description: Generate Rego policies from the child policy | |||
There was a problem hiding this comment.
[MEDIUM] Command description is stale
The frontmatter description: Generate Rego policies from the child policy no longer reflects the new test-driven workflow. Now that this command loads build-assessment, the description should mention the test-first approach or single/batch modes to set correct user expectations.
2 participants
Summary
Introduce
comply:build-assessmentas the single user-facing entry point for Rego policy generation, replacing direct invocation ofpack-assessment:comply:build-assessmentskill with two modes: single (default, one requirement at a time with interactive pause points) and batch (all at once)AskUserQuestionat every pause point for structured option selectiontest-driven-assessmentandpack-assessmentmarkedinvocable: false(internal-only)using-complypack) and post-setup guidance (mcp-setup) updated to point users tobuild-assessmentSingle Mode Flow
AskUserQuestionfor structured inputBatch Mode Flow