Push Tools — Security skill for Claude Code
Audit an agent's tools.py and upload it via MCP after approval; verify registration.
How to install Push Tools
Installs to ~/.claude/skills/atishayganesh-sarvam-kit-push-tools/SKILL.md
mkdir -p ~/.claude/skills/atishayganesh-sarvam-kit-push-tools && curl -fsSL https://raw.githubusercontent.com/AtishayGanesh/sarvam-kit/HEAD/commands/push-tools.md -o ~/.claude/skills/atishayganesh-sarvam-kit-push-tools/SKILL.md Restart Claude Code, or start a new session, for it to be picked up.
What Push Tools does
description: Audit an agent's tools.py and upload it via MCP after approval; verify registration. argument-hint: ""
Load sarvam-agent-authoring. Run `uv run sarvam tools audit agents/$1/tools/tools.py`; fix findings until PASS. Show the diff vs the last uploaded copy, ask for approval, then MCP `upload_agent_code`. Afterwards read the agent back (MCP agents get) and confirm every tool class is registered with the expected name and description. Do not commit unless I ask.
Alternatives in Security
- Maestro Odyssey — Long-running iterative cycle — one entry, seven modes (debug improve planex review security defensive ui) 530 ★
- Memory Self Review — Mine recent agent history (claude-mem + usage stats) for recurring failures and repeated patterns, audit MEMOR 297 ★
- Mobile Resume — Audit an unfinished mobile app, create a completion plan, then implement remaining work after approval 107 ★
Full documentation available on GitHub
View Source RepositoryRelated Skills
Meta Ads Agent
Approval-gated Meta Ads agent for Claude Code: reporting, creative upload, targeting research, experiments, gu
Push Back
Verify a claim against code+docs+memory before acting. Run when a finding/audit/recommendation looks suspect,
Claude Code Ship Gate
Pre-push quality gate for Claude Code that runs above git, so git push --no-verify can't skip it. Tests, code
Fix Traceability
Audit and repair one ticket's TMS traceability after explicit approval.
AgentDoctor
Local CLI that audits coding-agent configuration for security, instructions, context, and MCP — no API key or
Logic Fix All
Autonomous audit-and-fix — after consent, scan the target, fix every logic issue found (all severities), verif
Related Agents
Mestre Implementer
Implement explicitly approved scope, verify relevant behavior, and create coherent owned local milestone commi
Dynamic Pricing
Propose and (after approval) push rate and minimum-length-of-stay changes for hotel rooms in the next 30 days.
Safeguard
Security audit specialist. Use proactively before release, after touching auth, payments, file upload, or user