Retire deepgrade-guard; playbook document templates, runbook, release checklist, postmortems - #20
Merged
Merged
Conversation
… release checklist, postmortems Borrowed from anthropics/knowledge-work-plugins after a side-by-side comparison against the six-stage playbook. Deepgrade's workflow, debugging, and testing guidance were stronger and are untouched; what it lacked was fixed document shapes. - documentation: adr, brd, prd, readme templates gain a fill-in document body. The PRD template pointed at a "standard template" that did not exist. New runbook-template.md wired into the dispatcher (subcommand, menu, audit table, document-chain check, command reference). Stage 4 already gated on a reviewed runbook without defining one. - plan/templates/spec.md: requirements carry P0/P1/P2, trace to intent.md, and have Given/When/Then criteria with a negative case; new Success metrics table. stage-2-design.md requirements guidance updated to match. - plan/templates/review.md: pre-deploy / deploy / post-deploy / rollback checklist with numeric rollback triggers over a window; stage-5-deploy.md points at the runbook and forbids "looks wrong" as a trigger. - troubleshoot: SEV1/SEV2 status-update template and cadence in the pre-flow; blameless postmortem template in step 5 whose action items feed the proposed intent. Log template gains Status Updates and Postmortem sections. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01MpBfdua2J8gduiE6YcS8Ut
…(BREAKING) The always-on safety plugin (force-push and hard-reset guard, migration guard, DB deploy guard, change and test trackers, session summary) is removed from the tree and the marketplace. Claude Code permission rules reach parity for every blocking behavior with no runtime dependency and no hook-error failure mode, and a project carrying its own `ask` rules for the same paths ended up with a second layer silently overriding its deliberate choice. Three plugins remain in lockstep. Removed with it: tests/run-hook-corpus.js, tests/fixtures/hook-corpus.json, the guard rows of tests/layer2-ledger-rows.js, and the $TMPDIR/dg-* marker bus. Suite repointed: per-plugin profile and loops, plugin-count assertions (SPLIT-2, SPLIT-4, CI validate job) now three, release-preflight lockstep mutation targets the audit manifest, layer 7 checklist drops the guard items, layer 2 fails if the corpus runner reappears undispatched. METHODOLOGY §6 Layer 1 becomes "Permission Rules" with a recommended settings.json deny/ask baseline; the hook description is kept under a historical heading and §11's handler table marks the five retired handlers. Root README, CONTRIBUTING, interop, and every plugin README/GUIDE say three plugins. plugins/deepgrade GUIDE.md and README.md are rewritten to describe the current shape: the six-stage artifact chain, the design gate (canary classes with criterion IDs, evidence-validator flags, the pass expression, known limits), the two gate tools as distinct from the three hooks, seven document templates, and every output location. CHANGELOG entry sits under "## Unreleased (next: 9.0.0)" so the version cross-check keeps reading 8.0.0 until release.sh cuts 9.0.0. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01MpBfdua2J8gduiE6YcS8Ut
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.
Summary
Two commits, reviewable separately.
1. Templates and stage artifacts (
fcb5d54). Five items borrowed from anthropics/knowledge-work-plugins after a side-by-side comparison; deepgrade's own workflow, debugging, and testing guidance were stronger and are untouched.runbookdocumentation subcommand and template. Stage 4 already gated on a reviewed runbook without defining one.2. Retire
deepgrade-guard(BREAKING) (56ba1fd). The always-on safety plugin is removed from the tree and the marketplace; three plugins remain in lockstep. Claude Code permission rules reach parity for every blocking behavior with no runtime dependency, and a project carrying its ownaskrules ended up with a second layer silently overriding its choice. METHODOLOGY §6 now documents the recommendedsettings.jsondeny/ask baseline and keeps the hook description under a historical heading. Suite, CI plugin-count assertion, release preflight, and all cross-references updated.plugins/deepgradeGUIDE and README rewritten to describe the design gate, six-stage chain, gate tools vs hooks, and output locations.CHANGELOG entry is under
## Unreleased (next: 9.0.0); rename to## 9.0.0when cutting the release.Test plan
bash tests/run-all.shgreen, 8/8 layers, on Windows Git Bash🤖 Generated with Claude Code
https://claude.ai/code/session_01MpBfdua2J8gduiE6YcS8Ut