Local AIPROPRO REQUIREDFC-AI-011
Deployment Claw
deployment-claw
v0.2.0May 25, 2026Local model deployment — GPU sizing, inference server configs, load balancing, and serving benchmarks
Deploying an LLM without me is like deploying a database without a backup strategy. I care about model serving latency (TTFT, TPOT, time-per-token p50/p95/p99), GPU memory fragmentation under concurrent requests, dynamic batching efficiency, and cold-start time from checkpoint to first token. I will benchmark your vLLM/TGI/TensorRT-LLM setup under load, profile KV-cache memory pressure at max context length, and produce a deployment topology with autoscaling rules and canary rollout configs. I speak Kubernetes, NVIDIA Triton, and the painful reality of CUDA driver mismatches.
OpenClaw
Hermes
Claude Code
Codex
Cursor
Windsurf
Aider
OllamaPRIMARY ACTION
Unlock with ProCOMPATIBLE WITH
OpenClawHermesClaude CodeCodex+4
8 compatible tools4 output formats
When to Use
- Run local models with private workflows
- Tune inference for local hardware
- Choose effective GGUF variants
- Benchmark practical latency and quality
Compatible Frameworks
OpenClawHermesClaude CodeCodexCursorWindsurfAiderOllama8 TOOLS
Quality Gates
- no fake claims
- docker configuration valid
- gpu allocation documented
- inference server tested
4 GATES DEFINED
Expected Outputs
Model Serving Latency Report (TTFT/TPOT p50-p99 with Throughput Curve)Deployment Topology Diagram with Autoscaling Rules and Resource AllocationLoad-Test Results with Saturation Point and Degradation AnalysisCanary Rollout Config with Traffic-Splitting and Health-Check Criteria
Native exports per tool
OpenClaw10 files
openclaw/AGENTS.mdopenclaw/SOUL.mdopenclaw/TOOLS.md+7 moreHermes5 files
hermes/skills/flickclaw/deployment-claw/SKILL.mdhermes/skills/flickclaw/deployment-claw/references/workflow.mdhermes/skills/flickclaw/deployment-claw/references/quality-gates.md+2 moreClaude Code6 files
claude-code/CLAUDE.mdclaude-code/.claude/skills/deployment-claw/SKILL.mdclaude-code/.claude/skills/deployment-claw/references/workflow.md+3 moreCodex5 files
codex/AGENTS.mdcodex/.flickclaw/agents/deployment-claw/codex.mdcodex/.flickclaw/agents/deployment-claw/workflow.md+2 moreCursor3 files
cursor/.cursor/rules/flickclaw-deployment-claw.mdccursor/.cursor/rules/flickclaw-deployment-claw-workflow.mdccursor/.cursor/rules/flickclaw-deployment-claw-quality-gates.mdcWindsurf3 files
windsurf/.windsurf/rules/flickclaw-deployment-claw.mdwindsurf/.windsurf/rules/flickclaw-deployment-claw-workflow.mdwindsurf/.windsurf/rules/flickclaw-deployment-claw-quality-gates.mdAider3 files
aider/CONVENTIONS.mdaider/aider.mdaider/.aider.conf.ymlOllama4 files
ollama/Modelfileollama/system-prompt.mdollama/template.md+1 moreInstall Commands
Install the FlickClaw CLI, then select your AI agent framework below to get the correct install command.
Step 1: Install CLI (one-time)
npm install -g @flickclaw/cli@latestStep 2: Select Framework
OpenClaw
npm exec --yes @flickclaw/cli@latest -- install deployment-claw --target openclawDownload as ZIP
Example Prompt
Try this prompt with Deployment Claw to see what it can do:
Optimize the local model setup for performance. Benchmark current config and suggest improvements for Model Serving Latency Report (TTFT/TPOT p50-p99 with Throughput Curve), Deployment Topology Diagram with Autoscaling Rules and Resource Allocation, Load-Test Results with Saturation Point and Degradation Analysis.Example Output
IllustrativeWhat a typical Deployment Claw report looks like:
# Deployment Claw — Assessment Report **Project**: ollama-deployment **Context**: a local LLM deployment running Llama 3.1 8B on consumer GPU hardware **Generated**: 2026-07-10 ## Executive Summary The Deployment Claw completed its review of ollama-deployment (a local LLM deployment running Llama 3.1 8B on consumer GPU hardware). 3 findings were identified with concrete remediation steps. All quality gates were verified before delivery. ## Findings | # | Severity | Area | Finding | Recommended Action | |---|----------|------|---------|-------------------| | 1 | **P1** | Performance | Inference latency spikes to 8s under concurrency | Enable continuous batching and set max_batch=4 | | 2 | **P2** | Memory | Model consumes 18GB VRAM, headroom insufficient | Switch to Q4_K_M quantization, target <12GB | | 3 | **P2** | Setup | No Modelfile with system prompt defined | Create Modelfile with role, constraints, and templates | ## Quality Gates - [✓] latency_ttft_p95_under_target - [✓] kv_cache_memory_pressure_profile - [✓] cold_start_benchmark_from_checkpoint ## Outputs Generated - **Model Serving Latency Report (TTFT/TPOT p50-p99 with Throughput Curve)**: Included in the report above. - **Deployment Topology Diagram with Autoscaling Rules and Resource Allocation**: Included in the report above. - **Load-Test Results with Saturation Point and Degradation Analysis**: Included in the report above. - **Canary Rollout Config with Traffic-Splitting and Health-Check Criteria**: Included in the report above. ## Validation - [x] All quality gates passed (3/3) - [x] 3 findings documented with severity and remediation - [x] 4 output sections generated - [x] Evidence collected and referenced --- *This is an illustrative example output from FlickClaw. Results vary based on project context.*
RELATED AGENTS