hook-development
This skill should be used when the user asks to "create a hook", "add a PreToolUse/PostToolUse/Stop hook", "validate tool use", "wire command hooks", "use ${PLUGIN_ROOT}", "set up event-driven automation", "block dangerous commands", or mentions hook events (PreToolUse, PostToolUse, Stop, SubagentStart, SubagentStop, SessionStart, UserPromptSubmit). Provides comprehensive guidance for creating and implementing DeepSeek Harness plugin hooks with focus on the command-hook contract and the matcher model.
- License 许可
- MIT
- 最近更新
- 2026-11-14
- Stars
- ★ 21
- Forks
- 2
01 前言 Front Matter
- name
- hook-development
- description
- This skill should be used when the user asks to "create a hook", "add a PreToolUse/PostToolUse/Stop hook", "validate tool use", "wire command hooks", "use ${PLUGIN_ROOT}", "set up event-driven automation", "block dangerous commands", or mentions hook events (PreToolUse, PostToolUse, Stop, SubagentStart, SubagentStop, SessionStart, UserPromptSubmit). Provides comprehensive guidance for creating and implementing DeepSeek Harness plugin hooks with focus on the command-hook contract and the matcher model.
02 资源文件 10
- 脚本examples/load-context.sh1.6 KB
- 脚本examples/validate-bash.sh1.3 KB
- 脚本examples/validate-write.sh1.2 KB
- 参考文档references/advanced.md9.7 KB
- 参考文档references/migration.md7.5 KB
- 参考文档references/patterns.md6.1 KB
- 脚本scripts/hook-linter.sh4.1 KB
- 脚本scripts/README.md3.6 KB
- 脚本scripts/test-hook.sh5.2 KB
- 脚本scripts/validate-hook-schema.sh4.3 KB
03 SKILL.md 正文
Hook Development for DeepSeek Harness Plugins
Overview
Hooks are event-driven automation scripts that execute in response to DeepSeek Harness events. Use hooks to validate operations, enforce policies, add context, and integrate external tools into workflows.
Key capabilities:
- Validate tool calls before execution (PreToolUse)
- React to tool results (PostToolUse)
- Enforce completion standards (Stop, SubagentStop)
- Load project context (SessionStart)
- Automate workflows across the development lifecycle
Hook Types
Command Hooks (The Supported Type)
Execute bash commands for deterministic checks:
{
"type": "command",
"command": "bash ${PLUGIN_ROOT}/scripts/validate.sh",
"timeout": 60
}
Use for:
- Fast deterministic validations
- File system operations
- External tool integrations
- Performance-critical checks
Hook Configuration Formats
Plugin hooks.json Format
For plugin hooks in com.deepseek.harness/hooks/hooks.json, use wrapper format:
{
"description": "Brief explanation of hooks (optional)",
"hooks": {
"PreToolUse": [...],
"Stop": [...],
"SessionStart": [...]
}
}
Key points:
descriptionfield is optionalhooksfield is required wrapper containing actual hook events- This is the plugin-specific format
Example:
{
"description": "Validation hooks for code quality",
"hooks": {
"PreToolUse": [
{
"matcher": "Write",
"hooks": [
{
"type": "command",
"command": "${PLUGIN_ROOT}/hooks/validate.sh"
}
]
}
]
}
}
Settings Format (Direct)
For a standalone hook file (for example a project .agents/hooks/hooks.json), use the bare event-table format:
{
"PreToolUse": [...],
"Stop": [...],
"SessionStart": [...]
}
Key points:
- No wrapper - events directly at top level
- No description field
- This is the settings format
Important: The examples below show the hook event structure that goes inside either format. For plugin hooks.json, wrap these in {"hooks": {...}}.
Hook Events
PreToolUse
Execute before any tool runs. Use to approve, deny, or modify tool calls.
Example:
{
"PreToolUse": [
{
"matcher": "Write|Edit",
"hooks": [
{
"type": "command",
"command": "bash ${PLUGIN_ROOT}/scripts/validate-write.sh",
"timeout": 10
}
]
}
]
}
Output for PreToolUse:
{
"hookSpecificOutput": {
"permissionDecision": "allow|deny|ask",
"updatedInput": {"field": "modified_value"}
},
"systemMessage": "Explanation for the agent"
}
PostToolUse
Execute after tool completes. Use to react to results, provide feedback, or log.
Example:
{
"PostToolUse": [
{
"matcher": "Edit",
"hooks": [
{
"type": "command",
"command": "bash ${PLUGIN_ROOT}/scripts/check-edit-result.sh",
"timeout": 10
}
]
}
]
}
Output behavior:
- Exit 0: stdout shown in transcript
- Exit 2: stderr fed back to the agent
- systemMessage included in context
Stop
Execute when main agent considers stopping. Use to validate completeness.
Example:
{
"Stop": [
{
"hooks": [
{
"type": "command",
"command": "bash ${PLUGIN_ROOT}/scripts/check-completeness.sh",
"timeout": 30
}
]
}
]
}
Decision output:
{
"decision": "approve|block",
"reason": "Explanation",
"systemMessage": "Additional context"
}
SubagentStop
Execute when subagent considers stopping. Use to ensure subagent completed its task.
Similar to Stop hook, but for subagents. Observe only — it cannot block or add context.
UserPromptSubmit
Execute when user submits a prompt. Use to add context, validate, or block prompts. Takes no matcher.
Example:
{
"UserPromptSubmit": [
{
"hooks": [
{
"type": "command",
"command": "bash ${PLUGIN_ROOT}/scripts/check-prompt.sh",
"timeout": 10
}
]
}
]
}
SessionStart
Execute when DeepSeek Harness session begins. Use to load context and set environment.
Example:
{
"SessionStart": [
{
"matcher": "*",
"hooks": [
{
"type": "command",
"command": "bash ${PLUGIN_ROOT}/scripts/load-context.sh"
}
]
}
]
}
Special capability: Persist environment variables using $DSH_ENV_FILE:
echo "export PROJECT_TYPE=nodejs" >> "$DSH_ENV_FILE"
See examples/load-context.sh for complete example.
Events without a DeepSeek Harness counterpart
The runtime runs ten events: SessionStart, UserPromptSubmit, PreToolUse, PostToolUse, PostToolUseFailure, Stop, SubagentStart, SubagentStop, PreCompact and SessionEnd. PostToolUse fires when a tool call succeeds and PostToolUseFailure when it fails, matching the reference split. PreCompact runs but cannot veto a committed compaction, and SessionEnd runs from session teardown and discards its output. The source format also carries PostCompact, Notification, PermissionRequest and other events with no interception point here; those are skipped with a diagnostic, so do not build on them.
Hook Output Format
Standard Output (All Hooks)
{
"continue": true,
"suppressOutput": false,
"systemMessage": "Message for the agent"
}
continue: If false, halt processing (default true)suppressOutput: Hide output from transcript (default false)systemMessage: Message shown to the agent
Exit Codes
0- Success (stdout shown in transcript)2- Blocking error (stderr fed back to the agent)- Other - Non-blocking error
Hook Input Format
All hooks receive JSON via stdin with common fields:
{
"session_id": "abc123",
"transcript_path": "/path/to/transcript.txt",
"cwd": "/current/working/dir",
"permission_mode": "ask|allow",
"hook_event_name": "PreToolUse"
}
Event-specific fields:
- PreToolUse/PostToolUse:
tool_name,tool_input,tool_result - UserPromptSubmit:
user_prompt - Stop/SubagentStop:
reason
Access fields in prompts using $TOOL_INPUT, $TOOL_RESULT, $USER_PROMPT, etc.
Environment Variables
Available in all command hooks:
$DSH_PROJECT_DIR- Project root path$PLUGIN_ROOT- Plugin directory (use for portable paths)$DSH_ENV_FILE- SessionStart only: persist env vars here$DSH_REMOTE- Set if running in remote context
Always use ${PLUGIN_ROOT} in hook commands for portability:
{
"type": "command",
"command": "bash ${PLUGIN_ROOT}/scripts/validate.sh"
}
Plugin Hook Configuration
In plugins, define hooks in com.deepseek.harness/hooks/hooks.json, wrapped in a top-level "hooks" object:
{
"hooks": {
"PreToolUse": [
{
"matcher": "Write|Edit",
"hooks": [
{
"type": "command",
"command": "bash ${PLUGIN_ROOT}/scripts/validate-write.sh",
"timeout": 10
}
]
}
],
"SessionStart": [
{
"hooks": [
{
"type": "command",
"command": "bash ${PLUGIN_ROOT}/scripts/load-context.sh",
"timeout": 10
}
]
}
]
}
}
Plugin hooks merge with user's hooks and run in parallel.
Matchers
Tool Name Matching
Exact match:
"matcher": "Write"
Multiple tools:
"matcher": "Read|Write|Edit"
Wildcard (all tools):
"matcher": "*"
Regex patterns:
"matcher": "mcp__.*__delete.*" // All MCP delete tools
Note: Matchers are case-sensitive.
Common Patterns
// All MCP tools
"matcher": "mcp__.*"
// Specific plugin's MCP tools
"matcher": "mcp__plugin_asana_.*"
// All file operations
"matcher": "Read|Write|Edit"
// Bash commands only
"matcher": "Bash"
Security Best Practices
Input Validation
Always validate inputs in command hooks:
#!/bin/bash
set -euo pipefail
input=$(cat)
tool_name=$(echo "$input" | jq -r '.tool_name')
# Validate tool name format
if [[ ! "$tool_name" =~ ^[a-zA-Z0-9_]+$ ]]; then
echo '{"decision": "deny", "reason": "Invalid tool name"}' >&2
exit 2
fi
Path Safety
Check for path traversal and sensitive files:
file_path=$(echo "$input" | jq -r '.tool_input.file_path')
# Deny path traversal
if [[ "$file_path" == *".."* ]]; then
echo '{"decision": "deny", "reason": "Path traversal detected"}' >&2
exit 2
fi
# Deny sensitive files
if [[ "$file_path" == *".env"* ]]; then
echo '{"decision": "deny", "reason": "Sensitive file"}' >&2
exit 2
fi
See examples/validate-write.sh and examples/validate-bash.sh for complete examples.
Quote All Variables
# GOOD: Quoted
echo "$file_path"
cd "$DSH_PROJECT_DIR"
# BAD: Unquoted (injection risk)
echo $file_path
cd $DSH_PROJECT_DIR
Set Appropriate Timeouts
{
"type": "command",
"command": "bash script.sh",
"timeout": 10
}
Defaults: 60s per hook when the hook sets no timeout
Performance Considerations
Parallel Execution
All matching hooks run in parallel:
{
"PreToolUse": [
{
"matcher": "Write",
"hooks": [
{"type": "command", "command": "check1.sh"},
{"type": "command", "command": "check2.sh"}
]
}
]
}
Design implications:
- Hooks don't see each other's output
- Non-deterministic ordering
- Design for independence
Optimization
- Use command hooks for quick deterministic checks
- Keep hook commands small — heavy reasoning belongs in the session, where the model can delegate
- Cache validation results in temp files
- Minimize I/O in hot paths
Temporarily Active Hooks
Create hooks that activate conditionally by checking for a flag file or configuration:
Pattern: Flag file activation
#!/bin/bash
# Only active when flag file exists
FLAG_FILE="$DSH_PROJECT_DIR/.enable-strict-validation"
if [ ! -f "$FLAG_FILE" ]; then
# Flag not present, skip validation
exit 0
fi
# Flag present, run validation
input=$(cat)
# ... validation logic ...
Pattern: Configuration-based activation
#!/bin/bash
# Check configuration for activation
CONFIG_FILE="$DSH_PROJECT_DIR/.agents/plugin-config.json"
if [ -f "$CONFIG_FILE" ]; then
enabled=$(jq -r '.strictMode // false' "$CONFIG_FILE")
if [ "$enabled" != "true" ]; then
exit 0 # Not enabled, skip
fi
fi
# Enabled, run hook logic
input=$(cat)
# ... hook logic ...
Use cases:
- Enable strict validation only when needed
- Temporary debugging hooks
- Project-specific hook behavior
- Feature flags for hooks
Best practice: Document activation mechanism in plugin README so users know how to enable/disable temporary hooks.
Hook Lifecycle and Limitations
Hooks Load at Session Start
Important: Hooks are loaded when DeepSeek Harness session starts. Changes to hook configuration require restarting DeepSeek Harness.
Cannot hot-swap hooks:
- Editing
com.deepseek.harness/hooks/hooks.jsonwon't affect current session - Adding new hook scripts won't be recognized
- Changing hook commands won't update
- Must restart DeepSeek Harness: exit and run
dshagain
To test hook changes:
- Edit hook configuration or scripts
- Exit DeepSeek Harness session
- Restart:
dsh - New hook configuration loads
- Test hooks with
dsh --debug
Hook Validation at Startup
Hooks are validated when DeepSeek Harness starts:
- Invalid JSON in hooks.json causes loading failure
- Missing scripts cause warnings
- Syntax errors reported in debug mode
Use /hooks command to review loaded hooks in current session.
Debugging Hooks
Enable Debug Mode
dsh --debug
Look for hook registration, execution logs, input/output JSON, and timing information.
Test Hook Scripts
Test command hooks directly:
echo '{"tool_name": "Write", "tool_input": {"file_path": "/test"}}' | \
bash ${PLUGIN_ROOT}/scripts/validate.sh
echo "Exit code: $?"
Validate JSON Output
Ensure hooks output valid JSON:
output=$(./your-hook.sh < test-input.json)
echo "$output" | jq .
Quick Reference
Hook Events Summary
| Event | When | Use For |
|---|---|---|
| PreToolUse | Before tool | Validation, modification |
| PostToolUse | After a tool succeeds | Feedback, logging |
| PostToolUseFailure | After a tool fails | Failure alerts, corrective context |
| UserPromptSubmit | User input | Context, validation |
| Stop | Agent stopping | Completeness check |
| SubagentStart | Subagent starts | Context injection |
| SubagentStop | Subagent done | Task validation |
| SessionStart | Session begins | Context loading |
| PreCompact | Before compaction | Observation only, no veto |
| SessionEnd | Session teardown | Cleanup, output discarded |
A suite that declares PostToolUse and no PostToolUseFailure keeps seeing failed calls through its PostToolUse hooks, and the host log says so once. Declare PostToolUseFailure to take the reference split.
PostCompact, Notification and PermissionRequest have no counterpart in the runtime and are skipped with a diagnostic.
Best Practices
DO:
- ✅ Keep hook commands deterministic and fast
- ✅ Use ${PLUGIN_ROOT} for portability
- ✅ Validate all inputs in command hooks
- ✅ Quote all bash variables
- ✅ Set appropriate timeouts
- ✅ Return structured JSON output
- ✅ Test hooks thoroughly
DON'T:
- ❌ Use hardcoded paths
- ❌ Trust user input without validation
- ❌ Create long-running hooks
- ❌ Rely on hook execution order
- ❌ Modify global state unpredictably
- ❌ Log sensitive information
Additional Resources
Reference Files
For detailed patterns and advanced techniques, consult:
references/patterns.md- Common hook patterns (8+ proven patterns)references/migration.md- Migrating from basic to advanced hooksreferences/advanced.md- Advanced use cases and techniques
Example Hook Scripts
Working examples in examples/:
validate-write.sh- File write validation examplevalidate-bash.sh- Bash command validation exampleload-context.sh- SessionStart context loading example
Utility Scripts
Development tools in scripts/:
validate-hook-schema.sh- Validate hooks.json structure and syntaxtest-hook.sh- Test hooks with sample input before deploymenthook-linter.sh- Check hook scripts for common issues and best practices
External Resources
- Official Docs: the harness hooks documentation
- Examples: See security-guidance plugin in marketplace
- Testing: Use
dsh --debugfor detailed logs - Validation: Use
jqto validate hook JSON output
Implementation Workflow
To implement hooks in a plugin:
- Identify events to hook into (PreToolUse, Stop, SessionStart, etc.)
- Decide which event the check belongs at
- Write hook configuration in
com.deepseek.harness/hooks/hooks.json - For command hooks, create hook scripts
- Use ${PLUGIN_ROOT} for all file references
- Validate configuration with
scripts/validate-hook-schema.sh com.deepseek.harness/hooks/hooks.json - Test hooks with
scripts/test-hook.shbefore deployment - Test in DeepSeek Harness with
dsh --debug - Document hooks in plugin README
Command hooks are the supported type here. Keep them deterministic, fast, and free of side effects.
04 同仓库更多 Skills
verify
Verify that a code change actually does what it's supposed to by running the app and observing behavior. Use when asked to verify a PR, confirm a fix works, test a change manually, check that a feature works, or validate local changes before pushing.
simplify
Review changed code for reuse, quality, and efficiency, then fix any issues found. Use when the user asks to "simplify the code", "clean up the diff", "reduce duplication in my changes", or after finishing a feature before committing.
insight
Analyze this machine's DeepSeek Harness session history and report how sessions actually go — operating style, what works, friction points, and what to try next. Use when the user asks "how have I been using this", "analyze my sessions", "make an insight report", or wants a retrospective.
init
Set up a minimal AGENTS.md (and optionally skills and hooks) for this repo. Use when the user asks to "initialize the project", "generate AGENTS.md", "set up project instructions", or runs /init.
explore
Fast read-only codebase exploration that returns a report instead of filling the main context. Use when the user asks to "explore the codebase", "find where X is handled", "how does Y work", or when a search would take many reads to answer.
automation-recommender
Analyze a codebase and recommend DeepSeek Harness automations (hooks, subagents, skills, plugins, MCP servers). Use when user asks for automation recommendations, wants to optimize their DeepSeek Harness setup, mentions improving DeepSeek Harness workflows, asks how to first set up DeepSeek Harness for a project, or wants to know what DeepSeek Harness features they should use.