Index of every page on this site: llms.txt
All skills
acedergren avatar

/semgrep-coderabbit

@641a79b
Run it yourself

Use when reviewing code changes before commit or PR merge. Covers tool sequencing, finding severity priorities, fix ordering strategy, and conflict resolution between tools. Keywords: code review, semgrep, coderabbit, security scan, pre-commit review, PR review, hardcoded secrets, SQL injection, XSS. Triggers on "review with Semgrep" or "run CodeRabbit review".

Use this Skill: https://skilld.dev/gh/acedergren/agentic-tools/semgrep-coderabbit

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

SKILL.md

≈96 tokens for metadata: the name and description. ≈968 when used: this file.

Description uses 4.8% of example budget

Before choosing a Skill, your Agent reads its name and description. All available Skills share that space.

  • A shorter description leaves more room for other Skills. This entry exceeds our 1% size suggestion.

In our Claude Code example, all Skill names and descriptions share 8,000 characters. This Skill uses ≈384 characters, or 4.8%.

The 1% threshold is a size suggestion. Longer descriptions can still fit.

Your model, settings, and other Skills decide how much text your Agent can read.

Example settings and source

The example uses a 200k-token context and default Claude Code settings. The count includes the name, description, separators, and when_to_use when present. Codex also counts local file paths.

Skit's source and limits: Codex 0.160.1, Claude Code 2.1.292.

Semgrep + CodeRabbit Review

Two-stage code review: fast deterministic pattern detection first, then semantic AI analysis. Order is non-negotiable.

NEVER

  • Never run CodeRabbit before Semgrep passes — Semgrep failures are blockers, not suggestions.
  • Never skip re-running both tools after fixes — fixes can introduce new issues.
  • Never treat LOW findings as blocking — they are optional polish.
  • Never do more than 3 review cycles on the same PR — if still failing after 3, break the PR into smaller pieces.
  • Never context-switch between issue types mid-fix — batch similar issues together.

Execution Order

Stage 1: Semgrep (10-20 seconds)
  └─ FAIL → fix ALL violations → re-run until PASS
  └─ PASS → proceed to Stage 2

Stage 2: CodeRabbit (5-30 minutes)
  └─ Evaluate findings by priority
  └─ Fix CRITICAL + HIGH before merge
  └─ Consider MEDIUM if reasonable
  └─ LOW is optional

Stage 3: Verify
  └─ Re-run both tools after fixes
  └─ Confirm no new issues introduced

Commands

# Stage 1 — run on each changed file (loop required for Semgrep 1.146.0+)
for f in $(git diff --name-only HEAD); do
  semgrep scan --config auto --json "$f" 2>/dev/null
done

# Stage 2
coderabbit review --diff uncommitted

# Verify after fixes (repeat both)
semgrep scan --config auto <changed-files>
coderabbit review --diff uncommitted

If a project has a .semgrep.yaml, use --config .semgrep.yaml instead of --config auto.

Priority Reference

Priority Level Examples Deadline
CRITICAL Blocking Hardcoded secrets, auth bypass, SQL injection, XSS, data leakage Must fix before merge
HIGH Architectural Missing auth guards, schema mismatch, breaking changes, race conditions Fix before merge
MEDIUM Quality Weak crypto, poor error handling, type safety violations, duplication Fix if reasonable
LOW Polish Suggestions, optimization opportunities, style Consider optional

What Each Tool Catches (non-obvious)

Semgrep: Hardcoded secrets/API keys, missing auth guards on routes, weak crypto (MD5/SHA1), debug statements in prod, unsafe any types, SQL injection patterns.

CodeRabbit: N+1 query patterns, multi-tenant isolation gaps, API contract drift between services, test coverage gaps on behavioral changes — things that require semantic understanding of the codebase.

Fix Strategy

Batch by type, not by file. Fix all secrets issues, then all auth issues, then all injection issues. Run unit tests after each batch.

CRITICAL fixes first — security and auth bypass issues must be resolved before anything else. They are non-negotiable blockers.

3-pass rule — if the review-fix-reverify cycle exceeds 3 iterations, the PR scope is too large. Split it.

Conflict Resolution

When Semgrep and CodeRabbit give conflicting feedback on the same line: trust Semgrep. Its rules are deterministic and explicit. Validate CodeRabbit's concern with context before acting on it.

When to Use Each Mode

Situation Use
Pre-commit quick check --semgrep-only (10-20s)
Before creating PR (empty) full two-stage
Review specific recent commits --since HEAD~3
Only staged files (partial work) --staged
CodeRabbit not installed/available --semgrep-only + manual checklist

Arguments

$ARGUMENTS: Optional user-provided target, path, environment, symptom, or constraint. When empty, infer the narrowest safe scope from the current repository context and ask only if multiple high-impact choices remain.

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No alerts7mo4 checks · Risk SAFE
  • Gen Agent Trust Hub8mo

    This skill defines a structured code review process using Semgrep and CodeRabbit. It uses standard command-line tools to scan code changes for vulnerabilities and architectural issues. No malicious patterns, obfuscation, or data exfiltration were found.

  • Socket8mo

    No alerts

  • Snyk8mo

    Risk: LOW · No issues

  • ZeroLeaks7mo

    Score: 93/100 · 2 sections analyzed

Signed by skilld at 641a79b. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub last month.

Steadyupdated 6 months ago

README badge

README badge for acedergren/agentic-tools/semgrep-coderabbit