feat(skills): add ai-wafr — kernel-backed, evidence-verified WA review - #142
Closed
ibrahimcesar wants to merge 1 commit into
Closed
feat(skills): add ai-wafr — kernel-backed, evidence-verified WA review#142ibrahimcesar wants to merge 1 commit into
ibrahimcesar wants to merge 1 commit into
Conversation
Adds the AI-WAFR skill: a kernel-backed Well-Architected review that drives the offline ai-wafr binary rather than performing the review in the prompt. The skill's contract is thin — ensure the binary (build-from-source today; verified download once signed releases publish), run `ai-wafr bootstrap --workspace .`, and follow the pinned protocol verbatim. It carries no risk logic of its own; every rule comes from the kernel's bootstrap output, so the review is identical in every host. The kernel records an evidence ledger with exact locators, keeps agent proposals structurally separate from human decisions (independent reviewer evaluation before agent feedback is revealed), verifies findings against a pinned framework contract and a hash-chained event log, and produces a re-verifiable bundle plus a loopback-only explorable report. It is the evidence-verified counterpart to the prompt-layer aws-well-architected-framework-review skill — complementary, not a replacement. - skills/ai-wafr/SKILL.md + metadata.json (auto-discovered by install.sh across all supported agents; no wiring needed). - skills.sh.json: new "Verified Reviews (kernel-backed)" grouping. Honest limits stated in the skill: the bundled pack is reference-derived and unverified (no HRI/MRI); binary install is build-from-source until signed releases exist; the install step must verify integrity (no unverified binary).
Member
Author
|
Closing — not opening this yet. Keeping the ai-wafr kernel-backed skill local for now. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Adds a new skill
ai-wafr: a kernel-backed Well-Architected review that drives the offlineai-wafrbinary (the AI-WAFR reference protocol kernel) instead of performing the review in the prompt. It is the evidence-verified counterpart to the prompt-layeraws-well-architected-framework-review— complementary, not a replacement.Why
The prompt-layer skill produces the model's findings directly and fast. The kernel-backed skill produces a verifiable bundle: an evidence ledger with exact locators, agent proposals kept structurally separate from human decisions (independent reviewer evaluation recorded before agent feedback is revealed), findings verified against a pinned framework contract and a hash-chained event log, and a loopback-only explorable report. Use it when a review must be auditable or governance-grade.
Contract (thin by design)
SHA256SUMSonce signed releases exist. Never run an unverified binary.ai-wafr bootstrap --workspace .and follow the pinned protocol verbatim.The skill carries no risk logic of its own — every rule comes from the kernel's bootstrap output, so the review is identical in every host.
Changes
skills/ai-wafr/SKILL.md+metadata.json— auto-discovered byinstall.shacross all supported agents; no wiring needed.skills.sh.json— new "Verified Reviews (kernel-backed)" grouping.Honest limits (stated in the skill)
unverified— no HRI/MRI classifications.No relative Markdown links (link-check safe); no
references/tree or Power mirror (parity checks unaffected).