Open
Conversation
Contributor
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Repository UI Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (6)
✅ Files skipped from review due to trivial changes (2)
🚧 Files skipped from review as they are similar to previous changes (4)
WalkthroughAdds a comprehensive set of repository agent rules, command specifications, pointer alias files for Cursor/Claude, post-edit hooks, and Claude settings to standardize coding, testing, deployment, and agent workflows across the repo. Changes
Estimated code review effort🎯 2 (Simple) | ⏱️ ~12 minutes Possibly related PRs
🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
- Removed .cursorrules from .gitignore. - Updated references in add-audit.md and add-new-rule.md to point to the correct .agents directory. - Adjusted transaction analysis mode documentation to reflect the new command path. - Deleted outdated README.md for cursor rules as part of the consolidation effort.
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 10
🧹 Nitpick comments (7)
.agents/hooks/post-edit-validate.sh (1)
17-23: Case statement should match on$RELfor consistency.The case statement matches against
$FILE(which could be absolute), but the inner condition checks$REL(repo-relative). If$FILEis an absolute path like/home/user/repo/src/Foo.sol, the*.solpattern matches, but if$FILEwere passed as a relative path from a different working directory, behavior could differ.For consistency with the
$RELcheck inside, consider matching on$REL:Proposed fix
-case "$FILE" in +case "$REL" in *.sol) # Only lint Solidity in src/ or script/ — skip lib/, test artifacts, generated files if [[ "$REL" == src/* || "$REL" == script/* ]]; then🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In @.agents/hooks/post-edit-validate.sh around lines 17 - 23, The case statement currently matches on FILE but the inner guard uses REL, causing inconsistent behavior for absolute vs repo-relative paths; change the case to operate on REL (i.e., switch case "$REL" in) so the glob pattern (*.sol) is evaluated against the same repo-relative variable used in the if condition, keeping the Solidity lint gating in the block consistent with the REL checks..agents/rules/001-project-structure.mdc (1)
1-7: Consider addingpaths:for Claude compatibility.Other rule files in this PR include both
globs:(for Cursor) andpaths:(for Claude). This file only hasglobs:. SincealwaysApply: trueis set, this may work, but for consistency with the hybrid header pattern mentioned in the PR summary, consider adding:paths: - '**/*'🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In @.agents/rules/001-project-structure.mdc around lines 1 - 7, The rule header only defines globs: and omits paths:, which other rule files include for Claude compatibility; update the rule metadata by adding a paths: key with the same pattern as globs (e.g., '**/*') so both globs: and paths: exist alongside alwaysApply: true to match the hybrid header pattern used by the other rule files (look for the keys globs, paths, and alwaysApply in this rule definition)..agents/rules/300-bash.mdc (1)
75-83: ClarifyuniversalSendRawbehavior for EVM Production.The routing table shows
universalSendRawas "(not used)" for EVM Production, but line 68 mentionssendOrProposeinternally usesuniversalCast "sendRaw"for EVM staging. Consider clarifying whether EVM Production should show "propose-to-safe.ts" or if this row is intentionally documenting that raw calldata bypasses the Safe proposal flow.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In @.agents/rules/300-bash.mdc around lines 75 - 83, Update the routing table and nearby documentation to clarify universalSendRaw behavior for EVM Production: confirm whether sendOrPropose (which internally calls universalCast "sendRaw") should route to propose-to-safe.ts in production or intentionally bypass the Safe flow; if it should route to the Safe, change the EVM Production cell for universalSendRaw from "(not used)" to "propose-to-safe.ts" and add a short note near the sendOrPropose/universalCast documentation indicating that "sendRaw" uses the Safe proposal flow in production; if it intentionally bypasses the Safe, add an explicit note stating "raw calldata bypasses Safe proposal flow" and reference the sendOrPropose and universalCast symbols to make the behavior explicit..claude/rules/002-architecture.md (1)
1-1: Consider adding a trailing newline.POSIX convention recommends that text files end with a newline character. While this doesn't affect functionality in most cases, it's a common best practice for consistency.
📝 Proposed fix to add trailing newline
-../../.agents/rules/002-architecture.mdc +../../.agents/rules/002-architecture.mdc +🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In @.claude/rules/002-architecture.md at line 1, The file .claude/rules/002-architecture.md should end with a POSIX newline; open the file and add a single trailing newline character at the end (ensure the final byte is a '\n') so it matches the referenced ../../.agents/rules/002-architecture.mdc convention and avoids missing-newline warnings..claude/skills/add-network/SKILL.md (1)
1-1: Consider adding a trailing newline.For consistency with POSIX conventions, add a trailing newline.
📝 Proposed fix
-../../../.agents/commands/add-network.md +../../../.agents/commands/add-network.md +🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In @.claude/skills/add-network/SKILL.md at line 1, The file .claude/skills/add-network/SKILL.md is missing a POSIX-style trailing newline; open the SKILL.md file and ensure the file ends with a single newline character (i.e., insert a newline at EOF) so the file ends with a trailing newline as expected..claude/rules/003-context-monitor.md (1)
1-1: Consider adding a trailing newline.For consistency with POSIX conventions, text files should end with a newline character.
📝 Proposed fix
-../../.agents/rules/003-context-monitor.mdc +../../.agents/rules/003-context-monitor.mdc +🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In @.claude/rules/003-context-monitor.md at line 1, The file 003-context-monitor.md is missing a trailing newline; open that markdown file and add a single newline character at the end of the file (ensure the file ends with "\n") so it conforms to POSIX conventions and editor/CI checks that expect a final newline..agents/rules/099-finish.mdc (1)
12-13: Make Bash checks explicit with concrete commands.Current wording is a bit ambiguous for Bash validation. Add explicit command guidance (
bash -n, optionallyshellcheck) to reduce interpretation drift.Suggested wording update
-- **Testing**: After Solidity changes → `forge test` (or note suites remaining); after TS → lint/tests with Bun; after Bash → check execution flags/sourcing. State explicitly if anything not run. +- **Testing**: After Solidity changes → `forge test` (or note suites remaining); after TS → lint/tests with Bun; after Bash → run `bash -n <edited .sh files>` (and `shellcheck` if available), plus check execution flags/sourcing. State explicitly if anything not run.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In @.agents/rules/099-finish.mdc around lines 12 - 13, The "Bash" validation guidance in the "Testing" / "Linting" section is ambiguous; update the text to include explicit commands for shell script checks by adding concrete examples such as recommending "bash -n <file>" for syntax checking and suggesting running "shellcheck" (optionally with suggested flags) on any Bash scripts you created or edited, and instruct contributors to state if they didn't run these checks; reference the existing "Testing" and "Linting" wording to replace the vague Bash line with these concrete commands and the requirement to report results.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In @.agents/commands/add-audit.md:
- Line 146: The unlabeled fenced code block (the triple-backtick example block
containing "Extracted Information: ...") needs a language tag to satisfy
markdownlint MD040; update that opening fence from ``` to ```text (or another
appropriate language) so the block becomes labeled (e.g., ```text) while leaving
the block contents and closing ``` unchanged.
- Around line 125-134: Update the "Locate and save PDF file" step to prefer the
pasted/attached PDF bytes as the canonical source: instruct the agent to write
the attachment content directly to "audit/reports/<generated-filename>.pdf"
(instead of running the broad find/cp flow), then verify existence and non-zero
size (ls -lh or equivalent); only if attachment bytes are unavailable, fall back
to the current search heuristic (find ... -name "*.pdf") and prompt the user to
confirm the chosen file before copying. Reference the "Locate and save PDF file"
heading and the existing copy/verify commands so the author replaces the cp
"<source-path>" line with writing the attachment and keeps the verification
step.
In @.agents/context.md:
- Around line 8-13: The fenced code block in .agents/context.md is missing a
language tag (MD040); update the opening fence for the block that contains the
.agents/ listing so it reads as a language-tagged fenced block (for example add
"text" after the triple backticks) to silence the markdownlint warning and
preserve formatting for the block.
In @.agents/rules/004-config-structure.mdc:
- Around line 4-11: The YAML frontmatter in this rule is malformed because the
last entry in the "globs" sequence is concatenated with the "alwaysApply: false"
key; fix it by inserting a newline (and proper indentation) so "alwaysApply:
false" is its own top-level key. Locate the "globs:" block (the quoted pattern
ending with deployRequirements.json) and separate the sequence item and the
"alwaysApply: false" declaration so the "globs" list, the "paths" list, and
"alwaysApply: false" are each valid YAML entries.
In @.agents/rules/105-security.mdc:
- Around line 3-11: The rule description mentions tests but the glob/path
patterns in .agents/rules/105-security.mdc currently omit test files; update the
globs and paths entries to include test/**/*.sol and test/**/*.ts so the rule
covers {src,script,test}/**/*.{sol,ts} (i.e., add 'test/**/*.sol' and
'test/**/*.ts' to both the globs and paths arrays) to ensure test contracts and
scripts are validated by the security baseline.
In @.agents/rules/500-github-actions.mdc:
- Line 8: The YAML has a syntax error where the glob string '.github/**/*.yaml'
is concatenated with the key name alwaysApply; separate them by placing a
newline (or proper YAML separator) so the mapping key alwaysApply is on its own
line and the glob entry is a valid sequence item; update the entry containing
'.github/**/*.yaml' and the alwaysApply key so they are distinct YAML tokens.
In @.agents/rules/501-audits.mdc:
- Line 8: The YAML frontmatter in .agents/rules/501-audits.mdc has the field
`alwaysApply: false` concatenated to the previous glob string causing a syntax
error; edit the frontmatter so the glob pattern line
('.github/workflows/**/*version*.yml') ends with a newline and `alwaysApply:
false` is on its own line under the same frontmatter block, mirroring the fix
used in 502-whitelist-branching.mdc and ensuring proper YAML key/value
separation.
In @.agents/rules/502-whitelist-branching.mdc:
- Line 6: The YAML frontmatter has 'alwaysApply: false' appended to the previous
pattern string; fix the frontmatter in .agents/rules/502-whitelist-branching.mdc
by inserting a newline (line break) between the final glob entry
'config/composerWhitelist.json' and the scalar key alwaysApply so that
alwaysApply is a distinct YAML field (ensure 'config/composerWhitelist.json'
remains quoted and alwaysApply: false is on its own line).
In @.agents/rules/600-transaction-analysis.mdc:
- Around line 1-8: Add a required name field to the YAML frontmatter of this
rule: insert a top-level line like name: "Transaction analysis activation gate"
(matching the description) in the existing frontmatter block so the frontmatter
contains schema, scope, name, description, globs, and alwaysApply; this will
restore conformity with the other .agents/rules/*.mdc files and allow
tooling/IDE integrations to recognize the rule.
In @.claude/settings.json:
- Line 17: Fix the unclosed single quote in the permission pattern string
"Bash(bash -lc 'source script/helperFunctions.sh*)" by adding the missing
closing single quote before the final parenthesis and ensure the parenthesis
balance is correct; update the string so it becomes a properly quoted shell
command invocation (e.g., close the quote after .sh) and verify any wildcard or
glob character usage is intentional in the value.
---
Nitpick comments:
In @.agents/hooks/post-edit-validate.sh:
- Around line 17-23: The case statement currently matches on FILE but the inner
guard uses REL, causing inconsistent behavior for absolute vs repo-relative
paths; change the case to operate on REL (i.e., switch case "$REL" in) so the
glob pattern (*.sol) is evaluated against the same repo-relative variable used
in the if condition, keeping the Solidity lint gating in the block consistent
with the REL checks.
In @.agents/rules/001-project-structure.mdc:
- Around line 1-7: The rule header only defines globs: and omits paths:, which
other rule files include for Claude compatibility; update the rule metadata by
adding a paths: key with the same pattern as globs (e.g., '**/*') so both globs:
and paths: exist alongside alwaysApply: true to match the hybrid header pattern
used by the other rule files (look for the keys globs, paths, and alwaysApply in
this rule definition).
In @.agents/rules/099-finish.mdc:
- Around line 12-13: The "Bash" validation guidance in the "Testing" / "Linting"
section is ambiguous; update the text to include explicit commands for shell
script checks by adding concrete examples such as recommending "bash -n <file>"
for syntax checking and suggesting running "shellcheck" (optionally with
suggested flags) on any Bash scripts you created or edited, and instruct
contributors to state if they didn't run these checks; reference the existing
"Testing" and "Linting" wording to replace the vague Bash line with these
concrete commands and the requirement to report results.
In @.agents/rules/300-bash.mdc:
- Around line 75-83: Update the routing table and nearby documentation to
clarify universalSendRaw behavior for EVM Production: confirm whether
sendOrPropose (which internally calls universalCast "sendRaw") should route to
propose-to-safe.ts in production or intentionally bypass the Safe flow; if it
should route to the Safe, change the EVM Production cell for universalSendRaw
from "(not used)" to "propose-to-safe.ts" and add a short note near the
sendOrPropose/universalCast documentation indicating that "sendRaw" uses the
Safe proposal flow in production; if it intentionally bypasses the Safe, add an
explicit note stating "raw calldata bypasses Safe proposal flow" and reference
the sendOrPropose and universalCast symbols to make the behavior explicit.
In @.claude/rules/002-architecture.md:
- Line 1: The file .claude/rules/002-architecture.md should end with a POSIX
newline; open the file and add a single trailing newline character at the end
(ensure the final byte is a '\n') so it matches the referenced
../../.agents/rules/002-architecture.mdc convention and avoids missing-newline
warnings.
In @.claude/rules/003-context-monitor.md:
- Line 1: The file 003-context-monitor.md is missing a trailing newline; open
that markdown file and add a single newline character at the end of the file
(ensure the file ends with "\n") so it conforms to POSIX conventions and
editor/CI checks that expect a final newline.
In @.claude/skills/add-network/SKILL.md:
- Line 1: The file .claude/skills/add-network/SKILL.md is missing a POSIX-style
trailing newline; open the SKILL.md file and ensure the file ends with a single
newline character (i.e., insert a newline at EOF) so the file ends with a
trailing newline as expected.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Repository UI
Review profile: CHILL
Plan: Pro
Run ID: 693a43bf-e3fb-4097-8e6e-cf842a8d0a89
📒 Files selected for processing (130)
.agents/commands/add-audit.md.agents/commands/add-network.md.agents/commands/add-new-rule.md.agents/commands/analyze-tx.md.agents/commands/deprecate-network.md.agents/commands/review-bounty-report.md.agents/context.md.agents/hooks/post-edit-format.sh.agents/hooks/post-edit-validate.sh.agents/rules/000-global-standards.mdc.agents/rules/001-project-structure.mdc.agents/rules/002-architecture.mdc.agents/rules/003-context-monitor.mdc.agents/rules/004-config-structure.mdc.agents/rules/099-finish.mdc.agents/rules/100-solidity-basics.mdc.agents/rules/101-solidity-contracts.mdc.agents/rules/102-facets.mdc.agents/rules/103-solidity-interfaces.mdc.agents/rules/104-receiver-contracts.mdc.agents/rules/105-security.mdc.agents/rules/106-gas.mdc.agents/rules/107-solidity-scripts.mdc.agents/rules/200-typescript.mdc.agents/rules/201-safe-decode-scripts.mdc.agents/rules/300-bash.mdc.agents/rules/400-solidity-tests.mdc.agents/rules/401-testing-patterns.mdc.agents/rules/402-typescript-tests.mdc.agents/rules/500-github-actions.mdc.agents/rules/501-audits.mdc.agents/rules/502-whitelist-branching.mdc.agents/rules/600-transaction-analysis.mdc.agents/rules/README.md.claude/rules/000-global-standards.md.claude/rules/001-project-structure.md.claude/rules/002-architecture.md.claude/rules/003-context-monitor.md.claude/rules/004-config-structure.md.claude/rules/099-finish.md.claude/rules/100-solidity-basics.md.claude/rules/101-solidity-contracts.md.claude/rules/102-facets.md.claude/rules/103-solidity-interfaces.md.claude/rules/104-receiver-contracts.md.claude/rules/105-security.md.claude/rules/106-gas.md.claude/rules/107-solidity-scripts.md.claude/rules/200-typescript.md.claude/rules/201-safe-decode-scripts.md.claude/rules/300-bash.md.claude/rules/400-solidity-tests.md.claude/rules/401-testing-patterns.md.claude/rules/402-typescript-tests.md.claude/rules/500-github-actions.md.claude/rules/501-audits.md.claude/rules/502-whitelist-branching.md.claude/rules/600-transaction-analysis.md.claude/settings.json.claude/skills/add-audit/SKILL.md.claude/skills/add-network/SKILL.md.claude/skills/add-new-rule/SKILL.md.claude/skills/analyze-tx/SKILL.md.claude/skills/deprecate-network/SKILL.md.claude/skills/review-bounty-report/SKILL.md.cursor/commands/add-audit.md.cursor/commands/add-audit.md.cursor/commands/add-network.md.cursor/commands/add-network.md.cursor/commands/add-new-rule.md.cursor/commands/add-new-rule.md.cursor/commands/analyze-tx.md.cursor/commands/analyze-tx.md.cursor/commands/deprecate-network.md.cursor/commands/deprecate-network.md.cursor/commands/review-bounty-report.md.cursor/commands/review-bounty-report.md.cursor/rules/000-global-standards.mdc.cursor/rules/000-global-standards.mdc.cursor/rules/001-project-structure.mdc.cursor/rules/001-project-structure.mdc.cursor/rules/002-architecture.mdc.cursor/rules/002-architecture.mdc.cursor/rules/003-context-monitor.mdc.cursor/rules/003-context-monitor.mdc.cursor/rules/004-config-structure.mdc.cursor/rules/004-config-structure.mdc.cursor/rules/099-finish.mdc.cursor/rules/099-finish.mdc.cursor/rules/100-solidity-basics.mdc.cursor/rules/100-solidity-basics.mdc.cursor/rules/101-solidity-contracts.mdc.cursor/rules/101-solidity-contracts.mdc.cursor/rules/102-facets.mdc.cursor/rules/102-facets.mdc.cursor/rules/103-solidity-interfaces.mdc.cursor/rules/103-solidity-interfaces.mdc.cursor/rules/104-receiver-contracts.mdc.cursor/rules/104-receiver-contracts.mdc.cursor/rules/105-security.mdc.cursor/rules/105-security.mdc.cursor/rules/106-gas.mdc.cursor/rules/106-gas.mdc.cursor/rules/107-solidity-scripts.mdc.cursor/rules/107-solidity-scripts.mdc.cursor/rules/200-typescript.mdc.cursor/rules/200-typescript.mdc.cursor/rules/201-safe-decode-scripts.mdc.cursor/rules/201-safe-decode-scripts.mdc.cursor/rules/300-bash.mdc.cursor/rules/300-bash.mdc.cursor/rules/400-solidity-tests.mdc.cursor/rules/400-solidity-tests.mdc.cursor/rules/401-testing-patterns.mdc.cursor/rules/401-testing-patterns.mdc.cursor/rules/402-typescript-tests.mdc.cursor/rules/402-typescript-tests.mdc.cursor/rules/500-github-actions.mdc.cursor/rules/500-github-actions.mdc.cursor/rules/501-audits.mdc.cursor/rules/501-audits.mdc.cursor/rules/502-whitelist-branching.mdc.cursor/rules/502-whitelist-branching.mdc.cursor/rules/600-transaction-analysis.mdc.cursor/rules/600-transaction-analysis.mdc.cursor/rules/README.md.cursor/rules/README.md.cursorrules.gitignoreCLAUDE.md
💤 Files with no reviewable changes (1)
- .gitignore
Contributor
|
I'm not sure that Claude will load rules from
Some options for rules content is to be included as:
|
0xDEnYO
approved these changes
Apr 3, 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
4 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.
Which Linear task belongs to this PR?
Why did I implement it this way?
This PR makes the repository fully usable by both Cursor and Claude Code (Claude CLI) without any content duplication. All AI configuration — rules, commands/skills, hooks, and project context — lives in a single neutral location and is symlinked into each tool's expected directory structure.
Problem
Before this PR, the repo had:
.cursor/rules/*.mdc— Cursor-only, not visible to Claude Code.cursorruleswith 109 lines that duplicated rule content, referenced non-existentconventions.mdfiles, and listed deprecated ethers helpers that conflicted with existing rule 200.claude/rules/directory and noCLAUDE.mdSolution:
.agents/as the Single Source of TruthA new
.agents/directory holds all AI configuration. Tool-specific directories become pure symlink bridges — zero real content lives there.How Symlinks Work Here
A symlink is a filesystem pointer — a file that contains only a path to another file. When any program reads a symlink, the OS transparently redirects it to the target. There is zero content duplication: one physical file, multiple visible names.
Both entries point to the same inode. Editing the file from either side edits the source.
gittracks symlinks as 1-byte pointer objects — the real content is committed once under.agents/.Extension Mapping (the key gotcha)
Cursor only recognises rule files with the
.mdcextension. Claude Code only recognises.md. A directory-level symlink can't solve this — the files would have the wrong extension for one of the tools.The solution: individual file symlinks with extension remapping.
The
.mdsymlink resolves to a.mdcfile. Both tools read the same bytes.Hybrid Frontmatter
Each rule file uses a dual-header understood by both tools. Unknown keys are silently ignored by each parser.
Negation patterns (
!src/**/*.s.sol) are supported inglobs:but not inpaths:— positive globs only in thepaths:block.Commands → Skills Bridge
Cursor commands (
.mdfiles that become/command-namesuggestions) and Claude Code skills (directory +SKILL.md) are different structures for the same content.Claude Code requires each skill to live in its own named subdirectory as
SKILL.md. The symlink provides the expected path without duplicating content.Post-Edit Hooks (Claude Code)
Two hook scripts in
.agents/hooks/wire the existing linting/formatting toolchain into the Claude Code edit loop:post-edit-format.shbunx prettier --writeon.sol/.ts/.js— silent, always succeedspost-edit-validate.shbunx solhint(.solinsrc//script/),bunx tsc-files --noEmit(.ts), orbash -n(.sh) — prints errors to stdout so Claude sees them and self-corrects in the same turnConfigured in
.claude/settings.jsonunder thehookskey. No new permissions were needed —bunx *,bun *, andbash -n *were already in the allow list.Shared Project Context
.cursorruleswas replaced with a symlink to.agents/context.md. The original 109-line.cursorruleswas mostly redundant with existingalwaysApply: truerules and contained two broken references (conventions.md,docs/conventions_digest/— neither exists) and one conflict (listed deprecated ethers helpers that rule 200 explicitly forbids)..agents/context.md(28 lines) contains only what the rules don't cover: project identity, the.agents/structure explanation, and the common commands table.CLAUDE.md(repo root) is the Claude Code entry point. It@-importscontext.mdand documents the hooks. It does not re-import any rules — Claude Code loads global rules automatically (files with nopaths:key are always active).Both developers get identical behavioral rules. The only difference is the delivery mechanism.
Files Changed
.agents/rules/*.mdc.cursor/rules/).agents/commands/*.md.cursor/commands/).agents/hooks/*.sh.agents/context.md.agents/rules/README.md.cursor/rules/README.md.cursor/rules/*.mdc.cursor/commands/*.md.claude/rules/*.md.mdc).claude/skills/*/SKILL.md.claude/settings.jsonhookssection +Bash(ls *)permissionCLAUDE.md.cursorrules.agents/context.mdglobs:+paths:headersadd-new-rule.mdcommand.agents/workflow.agents/files (.cursor/rules/→.agents/rules/etc.)Checklist before requesting a review
Checklist for reviewer (DO NOT DEPLOY and contracts BEFORE CHECKING THIS!!!)