Improve agent context: restructure skills, slim AGENTS.md#1292
Open
noisysocks wants to merge 4 commits intomainfrom
Open
Improve agent context: restructure skills, slim AGENTS.md#1292noisysocks wants to merge 4 commits intomainfrom
noisysocks wants to merge 4 commits intomainfrom
Conversation
Consolidated create/fix skills into single cmp-rule skill, added verify-pr skill, trimmed AGENTS.md to focused project context, added YAML frontmatter for agent discovery. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
…ross-references - Expand step 2b with structured CMP investigation techniques (DOM inspection, JS source analysis, cross-site prevalence) - Add iframes and shadow DOM section to reference.md (chained selectors, limitations, cross-origin workaround) - Add cross-reference links between SKILL.md and reference.md sections - Simplify AGENTS.md cookie popup section to point to cmp-rule skill - Improve skill description with explicit trigger conditions Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Addresses recurring agent failures: lint/test:lib in verification checklist, prehide/visible interaction warning, cross-origin iframe guidance, cosmetic rule heuristic, engine modification guard, and runContext empty-string constraint. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Agents weren't reading the separate files. Merged selector strategy, cosmetic breakage fixes, iframe/shadow DOM patterns, diagnosis table, and example rules directly into the workflow steps where they're needed. Removed dangling cross-file references from AGENTS.md. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
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
2 participants
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.
Task/Issue URL: https://app.asana.com/1/137249556945/project/276630244458377/task/1214072784815658
Description:
Improves context engineering for AI agents creating autoconsent CMP rules. Changes were validated through a 7-run benchmark (details in the Asana task).
cmp-ruleskill (275 lines) — full workflow for creating/fixing rules: diagnosis, CMP identification, rule type decision tree, selector strategy with anti-patterns, iframe/shadow DOM, cosmetic breakage fixes, eval snippets, example rules, verification checklist.publicwww-searchskill (57 lines) — source code search via PublicWWW API for identifying whether a popup is from a widely-used CMP or site-specific.verify-prskill (71 lines) — PR verification checklist covering local CI, rule review, Jenkins E2E, browser testing..claude/skills/pointers to.agents/skills/for Claude Code compatibility.Steps to test this PR:
npm run lintpassesnpm run rule-syntax-checkpasses.agents/skills/{cmp-rule,publicwww-search,verify-pr}/SKILL.md.claude/skills/pointers resolve correctlycmp-ruleskill for rule work