bis-codebis-code/claude-toolkit/.claude/rules/common/agents.md

claude-toolkit / common

AGENTS.md8 starsChanged 6 months agoMITView on GitHub

Specialized agents handle tasks that require deep, focused expertise. Use them instead of relying on a generalist model for tasks they are designed for. Use an agent when: - The task requires a specific, repeatable workflow (code review, DB analysis, E2E generation) - The task benefits from a constrained tool set that limits accidental side effects - The task will recur across the project lifecycle Do not use an agent for one-line questions, quick lookups, or tasks that take fewer…

Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,774 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository16 other files it gives its agents.
- agent-creatorSkill
- architect-reviewSkill
- build-fixSkill
- code-reviewSkill
- deep-diveSkill
- docsSkill
- incident-debugSkill
- performance-reviewSkill
- planSkill
- qaSkill
- ralphSkill
- refactor-cleanSkill
- rule-creatorSkill
- security-reviewSkill
- skill-creatorSkill
- tdd-workflowSkill

The file

We split this agents.md into the rules it gives an agent

3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.

≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.

Common
Distinctive
Very specific

Distinctive:
Use an agent when:
line 7
Distinctive:
Do not use an agent for one-line questions, quick lookups, or tasks that take fewer than 3 steps.
line 12
Distinctive:
Never ask a reviewer agent to make changes — use a separate implementation agent or direct Claude for edits.
line 43

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.

Files

agents.md

Agents

Specialized agents handle tasks that require deep, focused expertise. Use them instead of relying on a generalist model for tasks they are designed for.

When to Use an Agent

Use an agent when:

The task requires a specific, repeatable workflow (code review, DB analysis, E2E generation)
The task benefits from a constrained tool set that limits accidental side effects
The task will recur across the project lifecycle

Do not use an agent for one-line questions, quick lookups, or tasks that take fewer than 3 steps.

Agent Decision Tree

Is this a code review or diff analysis?         → code-reviewer
Is this a performance bottleneck investigation? → performance-reviewer
Is this a security audit or threat model?       → security-reviewer
Is this a database query or schema review?      → database-reviewer
Is this E2E test generation or flake diagnosis? → e2e-runner
Is this a TDD cycle for a new feature?          → tdd-guide
Is this architectural planning or ADR?          → architect-reviewer
Is this a production incident?                  → incident-debugger
Is this a refactor or cleanup task?             → refactor-cleaner
Is this documentation or changelog?             → doc-updater
Is this a build or CI failure?                  → build-error-resolver

How to Invoke

Reference the agent by name in your task description. The agent's system prompt and tool permissions load automatically.

@database-reviewer review the migration in db/migrations/20240310_add_orders.sql
@code-reviewer check the diff on branch issue/45-payment-modal
@e2e-runner generate tests for the checkout flow

Agent Scope Rules

Agents are READ-ONLY by default unless they explicitly include Write or Edit tools.
Never ask a reviewer agent to make changes — use a separate implementation agent or direct Claude for edits.
Agents inherit project rules (coding style, git workflow, testing) unless their system prompt overrides them.

Composing Agents

For complex tasks, chain agents sequentially:

architect-reviewer — validate the approach before building
tdd-guide — drive implementation with tests
code-reviewer — review the output before committing
security-reviewer — audit any auth or data-access changes

Each agent operates independently. Pass its output as context to the next.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

It worked
Partly
It didn't

What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free