claude-sentient / commands
CLAUDE.md2 starsChanged 7 months agoMITView on GitHub
Context for working on /cs-* command files. All commands use YAML frontmatter + XML-structured instructions: Claude Sentient automates documentation based on task context: During /cs-loop INIT, rules are loaded by task keywords. Full mapping: .claude/rules/_index.md Commands can invoke each other via the Skill tool:
In this repository6 other files it gives its agents.
- .claude/hooks/CLAUDE.mdCLAUDE.md
- CLAUDE.mdCLAUDE.md
- .cursor/rules/claude-sentient.mdcCursor rule
- profile-detectionSkill
- quality-gatesSkill
- team-orchestrationSkill
The file
We split this CLAUDE.md into the rules it gives an agent
7 rules. About 0 are common, 2 are distinctive, and 5 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 7
All commands use YAML frontmatter + XML-structured instructions:
- Very specific: line 95
Create .claude/commands/cs-{name}.md with frontmatter
- Very specific: line 96
Use XML tags (<role>, <task>, <steps>, <avoid>, etc.)
- Very specific: line 97
Add <avoid> section with command-specific anti-patterns
- Very specific: line 98
Update cs-validate.md required commands list
- Distinctive: line 99
Update root CLAUDE.md commands table
- Very specific: line 100
Update README.md, CHANGELOG.md, install.sh, install.ps1
≈ “Common” is an estimate: at least 1 in 5 of the 28,029 CLAUDE.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.
Files
Commands — Claude Sentient
Context for working on
/cs-*command files.
Command File Structure
All commands use YAML frontmatter + XML-structured instructions:
---
description: What the command does
argument-hint: <args>
allowed-tools: Tool1, Tool2, ...
---
# /cs-command-name
<role>Expert persona for this command</role>
<task>Clear objective statement</task>
<context>Background info, tables, nested data</context>
<steps>Ordered procedure with <thinking> blocks</steps>
<criteria>Success metrics</criteria>
<output_format>Response structure</output_format>
<constraints>Rules and limitations</constraints>
<avoid>Command-specific DON'Ts</avoid>
<examples>Sample inputs/outputs</examples>Documentation Policy
Claude Sentient automates documentation based on task context:
Auto-Update Triggers
| Change Type | STATUS.md | CHANGELOG.md | DECISIONS.md |
| Feature added | Auto | Confirm | Only if architectural |
| Bug fixed | Auto | Confirm | -- |
| Refactoring | Auto | -- | If significant |
| Breaking change | Auto | Confirm (required) | Required |
| Config change | Auto | -- | -- |
Automation Levels
| Level | When | Examples |
| Fully Auto | Low risk, high value | Rule loading, STATUS.md updates |
| Auto + Confirm | Significant changes | CHANGELOG entries, version bumps |
| On Request | User preference | Full docs rewrites, ADRs |
Rule Auto-Loading
During /cs-loop INIT, rules are loaded by task keywords.
Full mapping: .claude/rules/_index.md
Skill Chaining
Commands can invoke each other via the Skill tool:
| From | To | When |
/cs-plan | /cs-loop | After plan approval, user chooses to execute |
/cs-status | /cs-loop | When pending tasks exist, user chooses to continue |
/cs-validate | /cs-loop | When issues found, user chooses to auto-fix |
/cs-loop | /cs-init | When no CLAUDE.md detected during INIT |
/cs-init | /cs-loop | After creating CLAUDE.md, user chooses to start working |
/cs-loop | /cs-team | When team eligibility detected, user approves team mode |
/cs-team | /cs-loop | When team completes, fallback to solo for remaining work |
/cs-assess | /cs-loop | When issues found, user chooses to fix them |
/cs-review | /cs-loop | When PR changes needed, user chooses to fix |
/cs-loop | /cs-docs | At COMMIT, when changed files need doc sync check |
/cs-docs | /cs-loop | After generating doc, user chooses to implement against spec |
/cs-learn | — | Standalone, appends to learnings.md |
/cs-mcp | — | Standalone, registers/validates MCP servers |
/cs-ui | /cs-loop | When UI issues found, user chooses to fix |
/cs-deploy | — | Standalone, read-only deployment readiness check |
/cs-sessions | /cs-loop | When user chooses to resume a previous session |
/cs-multi | — | Standalone, configures per-phase model routing |
/cs-debug | /cs-loop | When root cause requires implementation changes, user chooses to fix |
/cs-log | /cs-loop | When logging gaps/issues found, user chooses to fix |
/cs-batch | /cs-loop | Each worktree agent runs /cs-loop for its work stream |
/cs-batch | /cs-plan | When argument is a task description, creates plan first |
/cs-schedule | — | Standalone, wraps native /schedule with CS command integration |
Adding a New Command
- Create
.claude/commands/cs-{name}.mdwith frontmatter - Use XML tags (
<role>,<task>,<steps>,<avoid>, etc.) - Add
<avoid>section with command-specific anti-patterns - Update
cs-validate.mdrequired commands list - Update root
CLAUDE.mdcommands table - Update
README.md,CHANGELOG.md,install.sh,install.ps1
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
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.

