Language
General / language-agnostic
AI instructions for General / language-agnostic projects — coding rules, agent setup files, and prompts, organized by proficiency.
121 published items · Filter in browse
Security auditor subagent (wshobson)
Claude Code subagent (.claude/agents/security-auditor.md). Claude Code subagent for security audits: threat modelling, OWASP, OAuth2/OIDC, secrets, dependency and container scanning, compliance reporting.
Release manager
Claude Code subagent (.claude/agents/release-manager.md). Subagent that prepares a release - verifies tree and CI, computes the semver bump from merged changes, updates version and changelog files, and tags locally without pushing.
Performance profiler
Claude Code subagent (.claude/agents/performance-profiler.md). Subagent that measures before guessing - establishes a baseline, profiles with the stack's native tools, identifies the top hotspots, and proposes fixes with expected and measured gains.
Docs writer
Claude Code subagent (.claude/agents/docs-writer.md). Subagent that writes or updates README sections, API references, and how-to guides from the code itself, verifying every command and example against the repository.
Dependency auditor
Claude Code subagent (.claude/agents/dependency-auditor.md). Read-only subagent that audits a project's dependency tree for known vulnerabilities, abandoned or risky packages, license conflicts, and lockfile drift, and returns a prioritized remediation list.
Debugger
Claude Code subagent (.claude/agents/debugger.md). Subagent that reproduces a failure, isolates the root cause with evidence, applies a minimal fix, verifies it, and reports cause, fix, and prevention.
Debugger subagent (VoltAgent)
Claude Code subagent (.claude/agents/debugger.md). Claude Code subagent for systematic debugging: symptom gathering, hypotheses, isolation, root-cause analysis and verified fixes, on sonnet.
Debug mode (Copilot agent)
Claude Code subagent (.github/agents/debug.agent.md). Copilot custom agent that debugs in phases: assess, reproduce, isolate, fix, verify and document, running tests and reading terminal output along the way.
Critical thinking mode (Copilot agent)
Claude Code subagent (.github/agents/critical-thinking.agent.md). Copilot custom agent that challenges assumptions and keeps asking why instead of writing code; use before design decisions and big refactors.
Code reviewer
Claude Code subagent (.claude/agents/code-reviewer.md). Read-only subagent that reviews a diff for correctness, edge cases, and maintainability and returns prioritized findings with file:line and fixes.
Code reviewer subagent (wshobson)
Claude Code subagent (.claude/agents/code-reviewer.md). Claude Code subagent for comprehensive code review: security, performance, reliability, maintainability and test coverage with a structured report.
Architect review subagent (wshobson)
Claude Code subagent (.claude/agents/architect-review.md). Claude Code subagent that reviews designs and changes for architectural integrity: clean/hexagonal architecture, service boundaries, DDD, SOLID, cloud-native and security, with High/Medium/Low impact.
Writing Tests
Write unit and integration tests for existing code in the project's own ecosystem. Detects the test runner, identifies untested paths, and writes tests with proper mocking, edge cases, and assertio...
Writing Commit Messages
Write clear, conventional commit messages with proper type prefixes, scopes, and body content. Use when the user asks to commit, write a commit message, or review staged changes for a commit.
Visual Qa Testing
Visually QA a web application in the harness's browser tool: screenshot the changed page, read console errors, and check for failed network requests. Use after making UI changes to verify they look...
Verification Before Completion (superpowers skill)
Jesse Vincent's skill that forbids claiming success without running the verification command and reading its output first; evidence before assertions.
Test-Driven Development (superpowers skill)
Jesse Vincent's TDD skill: the iron law (watch the test fail first), red-green-refactor steps, rationalization red flags and a guide to writing good tests.
TDD red/green/refactor loop
Strict red/green/refactor loop with a behavior list, one failing test at a time, and explicit stop points for user confirmation.
Systematic Debugging (superpowers skill)
Jesse Vincent's debugging skill: four-phase root-cause process, root-cause tracing, defense in depth, condition-based waiting, and a test-polluter finder script.
Suggesting Skills
When the user struggles with a task that a known skill could handle, suggest installing or using a skill from the project's `.agents/skills/` directory.
OWASP-oriented security review of a diff
Boundary-tagged diff review that traces inputs to sinks, applies a 13-section OWASP checklist from references/, and reports severity-rated findings with fixes.
Saving Workspace Context
Persist project knowledge that should outlast this conversation: decisions, gotchas, research, and commands. Writes to AGENTS.md or a docs folder the project already uses. Use when the user asks to...
Responsive Testing
Open the app in the harness's browser tool at multiple viewport sizes, screenshot each, and report layout breakage. Use after UI layout changes or when the user asks to verify responsive behavior.
Release notes from merged PRs since last tag
Collects merged PRs and commits since the last tag, classifies by user impact, rewrites in user vocabulary, and adds upgrade steps.