Repository navigation
Point agent adapters to AGENTS.md and move multi-line workflow steps into tested scripts - #757
Merged
Merged
Conversation
anth-volk
added this pull request to stack #755
October 12, 2026 19:38
This was referenced Oct 12, 2026
anth-volk
force-pushed
the
rules-followups
branch
from
October 13, 2026 16:56
55ca16d to
6efa6c5
Compare
anth-volk
force-pushed
the
rules-followups
branch
4 times, most recently
from
October 14, 2026 15:56
0e66611 to
64d82cf
Compare
anth-volk
force-pushed
the
rules-followups
branch
2 times, most recently
from
October 14, 2026 17:56
4d764c2 to
1ca1fb8
Compare
Sync scripts/rules with policyengine-api. The rules-registry guard now verifies that agent adapters ([agents]) only point to AGENTS.md, and a new ci-scripts quality guard ([ci_scripts]) fails on run: blocks longer than one logical line, workflow scripts that no test runs, and shellcheck findings, each ratcheted against a baseline that may only shrink. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Both adapters duplicated part of the skill list and had drifted from AGENTS.md (observability and the rules registry were missing). They now only point to AGENTS.md; the safety boundaries move into AGENTS.md. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Document the rule in docs/engineering/skills/ci-workflows.md, install a pinned shellcheck in the quality-guards job, and record today's exceptions as baselines: 16 multi-line run: blocks, 4 untested scripts and 1 script with a shellcheck finding. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Steps 1.3, 1.5 and 1.6: sim.agents.adapters-point-to-agents-md (enforced, with proof), sim.process.pr-tracking-issue (doc-only, split out of sim.process.same-repo-draft-prs), and three ratcheted sim.ci.* rules with proofs. AGENTS.md gains the safety boundaries and a pointer to the CI workflows skill. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Sync scripts/rules with policyengine-api: folded blocks and backslash-continued lines now count line by line, and chaining commands with &&, || or ; counts as more than one line. The multi-line baseline grows from 16 to 20 entries (the backslash-continued commands). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
#743 and #744 landed on main after this stack branched. They add 13 multi-line run: blocks to the deploy and rollback workflows, and #744 replaced the update_routing "Publish active routing state" step. Add the new blocks to the sim.ci.multiline-run-is-script baseline and drop the entry for the removed step, as this PR does for the blocks that existed when the rule was introduced. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
anth-volk
force-pushed
the
rules-followups
branch
from
October 14, 2026 19:46
1ca1fb8 to
ca97a1a
Compare
anth-volk
marked this pull request as ready for review
October 14, 2026 20:04
This was referenced Oct 14, 2026
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.
Fixes #756
Same-Repository Draft PR Check
PolicyEngine/policyengine-sim-api, not a personal fork.Summary
The top of the rules stack: Phase 0 (#754) → rules registry (#752) → this PR, which covers Phase 1 steps 1.3, 1.5 and 1.6. The API counterpart is PolicyEngine/policyengine-api#3890.
1.3: one source of agent instructions.
CLAUDE.mdand.github/copilot-instructions.mdare now short pointers toAGENTS.md. Both had duplicated part of the skill list and drifted from it.main; no production deploys or client publishing unless asked) move intoAGENTS.md.[agents]table. It's registered assim.agents.adapters-point-to-agents-md(enforced, with proof).1.5:
sim.process.pr-tracking-issueis registered as doc-only and split out ofsim.process.same-repo-draft-prs. CI enforcement comes in step 2.8.1.6: workflow logic lives in tested scripts. The new
ci-scriptsquality guard (same tool as the API) checks for:run:blocks with more than one line of code, counting every line, including folded and backslash-continued ones, plus commands chained with&&,||or;;shellcheckfindings.Each check is ratcheted against
rules/baselines/: 20 multi-line blocks, 4 untested scripts (the dependency-update scripts andupdate-policyengine-package.sh) and 1 shellcheck finding (modal-precompute.sh).The rules are
sim.ci.multiline-run-is-script,sim.ci.workflow-scripts-testedandsim.ci.shell-scripts-pass-shellcheck, all ratcheted, with proofs. Thequality-guardsjob installs a pinnedshellcheck-py==0.11.0.1. Seedocs/engineering/skills/ci-workflows.md.scripts/rules/is byte-identical with the API's copy; the tool sha256 inrules.mdmatches.Testing
python -m unittest discover -s scripts/rules/tests -t .: 74 tests pass.python scripts/run_quality_guards.pypassesrules-registryandci-scripts.python -m scripts.rules guard --base-ref origin/mainpasses.projects/policyengine-simulation-entry:uv run pytest tests/passes 159 tests.changelog_entry.yamlparse as YAML.actionlintreports only the pre-existing note on the format loop.🤖 Generated with Claude Code