agentic-tool-integration

A development guide for connecting a new tool to an AI agent, including the registry that lists available tools and the code paths that execute them.

In plain words
What is it for?
It is for adding or changing command-line, MCP, or API tools that an agent can call, while checking dependencies, registration, and dispatch logic.
Why use it?
It prevents a tool from being visible in one mode but missing, ignored, or broken when the agent runs parallel plans or uses another execution path.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/samugit83/redamon/agentic-tool-integration
Any agent
npx skills add samugit83/redamon --skill agentic-tool-integration
Clone the repo
git clone --depth 1 https://github.com/samugit83/redamon

Made for: Claude Code, Codex.

Per session 135 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,381 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00135 $0.01381
Opus 5 $0.00068 $0.00691
Sonnet 5 $0.00027 $0.00276
Haiku 4.5 $0.00014 $0.00138

Measured 3d ago against content hash 7def707dc34e, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

agentic-tool-integration scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/agentic-tool-integration/SKILL.md · 98 lines

How it starts

The opening of the file, as written. The whole thing — 98 lines — stays where its author put it; the contents beside it link to each section on GitHub.

When to Use

  • Adding any tool the agent uses during a chat session (CLI wrapper, MCP tool, or API tool).

For a recon-pipeline tool (runs during a scan, writes the graph), use recon-tool-integration instead. For the per-setting/default multi-layer wiring, use project-settings-cascade. For a whole new attack skill (not a tool), use builtin-agent-skill.


Critical Rules

  • NEVER import a package not already in the agent image. agentic/ is baked into the redamon-agent image; a missing import crash-loops the container. Confirm it is in agentic/requirements.txt or the Dockerfile first.
  • NEVER add a tool without a TOOL_REGISTRY entry in agentic/prompts/tool_registry.py. The registry is the single source of truth the LLM reads; an unregistered tool is invisible to the agent. All four fields (purpose, when_to_use, args_format, description) are required, and the description must clear the 100-char minimum in agentic/tests/test_tool_registry_completeness.py, which fails on any unregistered or stub entry.
  • NEVER forget the dispatch branch for a non-MCP / API tool. PhaseAwareToolExecutor.execute() at agentic/tools.py:2070 has hardcoded if/elif dispatch. A Type D (API) tool or an MCP tool needing key injection MUST add an elif; without it the tool is registered but never dispatched. MCP tools with no key injection go through the else branch automatically.
  • NEVER edit execute_tool_node.py without mirroring it in execute_plan_node.py. execute_tool_node.py (single tool) and execute_plan_node.py (parallel plans) duplicate long-running detection and session/listener handlers. Update one only and the tool works interactively but silently misbehaves in parallel plan execution.
  • NEVER rely on the Prisma default to reach existing projects. The agentToolPhaseMap default applies to NEW projects only. Existing projects need a jsonb UPDATE or the agent never sees the tool there. See project-settings-cascade.
  • ALWAYS add the tool to TOOL_PHASE_MAP in agentic/project_settings.py. If it sends attack traffic: also add it to DANGEROUS_TOOLS (frozenset, project_settings.py:22), a per-tool section in agentic/prompts/stealth_rules.py, and the right list in CATEGORY_TOOL_MAP (_check_roe_blocked, execute_plan_node.py:47).

Read the full file on GitHub · 98 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 3d ago First seen · 98 lines · 135 tokens per session scan A 7def707dc34e

Subscribe to this mod's changes

agentic-tool-integration is a skill published in the GitHub repository samugit83/redamon (2,372 stars, last pushed 2d ago), licensed MIT. It adds 135 tokens to every session and 1,381 once invoked, about $0.0007 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other skills, from other repositories

workflow-orchestrator

Module Loop and Iteration Skill for orchestrating multi-phase penetration testing workflows. Use when coordinating sequential tool execution, managing dependencies between reconnaissance and vulnerability scanning modules, implementing adaptive fallback strategies, or managing workflow state across iterations.…

SHAdd0WTAka/Zen-Ai-Pentest · 85 tokens

poc-validator

Proof-of-Concept (PoC) Validation Skill for security exploit verification. Use when validating exploitability of vulnerabilities, generating tailored payloads, executing exploits in sandboxed environments, or verifying successful exploitation (e.g., SQL injection, CVE exploitation, command injection). Triggers on…

SHAdd0WTAka/Zen-Ai-Pentest · 86 tokens

Reconnaissance & OSINT Automation

Passive and active reconnaissance, subdomain enumeration, DNS analysis, technology fingerprinting, and OSINT data correlation for authorized security assessments.

Masriyan/Claude-Code-CyberSecurity-Skill · 33 tokens

research-orchestrator

Parallel research with 4 specialized agents (Official, Practical, Edge Cases, Future).

SHAdd0WTAka/Zen-Ai-Pentest · 22 tokens

tdd-agent

Autonomous test-driven development agent that writes code to make tests pass.

SHAdd0WTAka/Zen-Ai-Pentest · 17 tokens

analyzing-prefetch-files-for-execution-history

Parse Windows Prefetch files to determine program execution history including run counts, timestamps, and referenced files for forensic investigation.

26zl/cybersec-toolkit · 34 tokens