Skip to main content

Skills

Snoopy ships a skill package that provides AI agents with zero-context workflow guidance for Reddit monitoring.

Primary skill package​

The skill/snoopy-cli/ package is the installable skill bundle for Snoopy.

  • Location: skill/snoopy-cli/SKILL.md (repository root)
  • Purpose: Guides agents through install, setup, job management, debugging, and MCP integration
  • Format: Agent Skills specification compliant

Use cases​

  • Install and configure Snoopy from scratch
  • Create and manage monitoring jobs
  • Debug failed runs using errors and logs
  • Register Snoopy with agent frameworks (Claude, Cursor, VS Code, etc.)
  • Use MCP server for programmatic access
  • Export qualified results for downstream processing

Core file​

skill/snoopy-cli/SKILL.md contains:

  • Installation and setup instructions (interactive + non-interactive)
  • Deterministic workflow for all operations
  • MCP tool set documentation
  • Argument semantics and constraints
  • Gotchas and sharp edges
  • Failure handling matrix
  • Source evidence map

Companion references​

  • references/command-catalog.md — full command/argument matrix
  • references/troubleshooting.md — failure diagnostics

Installation locations​

Project scope​

  • .agents/skills/snoopy-cli/
  • .github/skills/snoopy-cli/
  • .cursor/skills/snoopy-cli/

User scope​

  • ~/.agents/skills/snoopy-cli/
  • ~/.copilot/skills/snoopy-cli/
  • ~/.claude/skills/snoopy-cli/

Required skill contract​

Each shipped skill must document:

  1. Name: snoopy-cli (kebab-case, matches directory)
  2. Inputs: OpenRouter API key, subreddits, qualification prompt, job name, schedule
  3. Guardrails: always require API key, daemon must be running for scheduled jobs
  4. Outputs: job runs, qualified results, analytics, error reports
  5. Failure modes: missing API key, daemon not running, token truncation, DB locked
  6. Verification prompts: should-trigger and should-not-trigger examples

Publication rules​

  • Canonical page: docs/for-agents/skills.md
  • Links to human docs: docs/getting-started/installation.md, docs/getting-started/quickstart.md
  • Concrete examples: all command examples are copy-paste safe
  • Sync requirements: skill updates must happen in same change as CLI behavior changes

Sync and drift policy​

When MCP tool surface changes, update in the same change:

  1. src/mcp/tools.ts (schemas and contracts)
  2. src/mcp/server.ts (handlers)
  3. docs/for-agents/mcp-server.md (MCP docs)
  4. skill/snoopy-cli/SKILL.md (skill docs)

When agent install targets change, update:

  1. src/agent/install.ts (install logic)
  2. docs/for-agents/agent-setup.md (setup docs)
  3. skill/snoopy-cli/SKILL.md (skill docs)