cristiano.hoshikawa/openclaw-oci-pptx

Fork 0

mirror of https://github.com/hoshikawa2/openclaw-oci-pptx.git synced 2026-03-03 16:09:35 +00:00

Files

Cristiano Hoshikawa 612fb30af6 bug fix. PROMPT OK

2026-03-01 09:50:06 -03:00

21 KiB

Raw Permalink Blame History

Generate PowerPoint Presentations with OpenClaw and Oracle Cloud Generative AI

Enterprise AI Power, Open Ecosystem, Zero Compromise

The rapid evolution of AI orchestration tools has reshaped how companies build intelligent systems. Among these tools, OpenClaw has emerged as a powerful open-source platform designed to simplify the creation of AI agents, conversational workflows, and multi-channel integrations.

OpenClaw is not just another wrapper around LLM APIs. It is:

Modular
Plugin-driven
Open-source
OpenAI-compatible
Community-powered

Its OpenAI-compatible design makes it instantly interoperable with the entire AI tooling ecosystem — SDKs, automation frameworks, browser clients, bots, and custom agent pipelines.

And because it is open source, innovation happens in public.

There is an active and growing community contributing:

New plugins
Messaging integrations (WhatsApp, web, etc.)
Tool execution engines
Agent frameworks
Workflow automation patterns
Performance optimizations

This means OpenClaw evolves continuously — without vendor lock-in.

But while agility and innovation are essential, enterprises require something more:

Security
Governance
Compliance
Regional data sovereignty
Observability
Controlled network exposure
Predictable scalability

This is where Oracle Cloud Infrastructure (OCI) Generative AI becomes the strategic enterprise choice.

⸻

The Power of Ecosystem + Enterprise Security

OpenClaw: Open Ecosystem Advantage

Because OpenClaw is:

Open-source
Community-driven
Plugin-extensible
OpenAI-protocol compatible

You benefit from:

Rapid innovation
Transparent architecture
Community-tested integrations
Zero dependency on a single SaaS provider
Full customization capability

You are not locked into one AI vendor. You control your orchestration layer.

This flexibility is critical in a world where models evolve rapidly and enterprises need adaptability.

⸻

OCI Generative AI: Enterprise Trust Layer

Oracle Cloud Infrastructure adds what large organizations require:

Fine-grained IAM control
Signed API requests (no exposed API keys)
Dedicated compartments
Private VCN networking
Sovereign cloud regions
Enterprise SLAs
Monitoring & logging integration
Production-ready inference endpoints

OCI Generative AI supports powerful production-grade models such as:

Cohere Command
LLaMA family
Embedding models
Custom enterprise deployments
OpenAI-compatible models via mapping

This creates a secure AI backbone inside your own tenancy.

⸻

Why This Combination Is Strategically Powerful

By implementing a local OpenAI-compatible gateway backed by OCI:

OpenClaw continues to behave exactly as designed — while inference happens securely inside Oracle Cloud.

You gain:

Full OpenAI protocol compatibility
Enterprise security boundaries
Cloud tenancy governance
Scalable AI inference
Ecosystem extensibility
Open-source flexibility

Without rewriting your agents. Without breaking plugins. Without sacrificing innovation.

Why Use OCI Generative AI?

Oracle Cloud Infrastructure provides:

Enterprise security (IAM, compartments, VCN)
Flexible model serving (ON_DEMAND, Dedicated)
High scalability
Cost control
Regional deployment control
Native integration with Oracle ecosystem

By building an OpenAI-compatible proxy, we combine:

OpenClaw flexibility + OCI enterprise power

OpenClaw + OCI Generative AI Gateway and PPTX Template Builder

About the tutorial

OpenAI-compatible endpoint

This tutorial is based on Integrating OpenClaw with Oracle Cloud Generative AI (OCI) tutorial and explains how to integrate OpenClaw with Oracle Cloud Infrastructure (OCI) Generative AI by building an OpenAI-compatible API gateway using FastAPI.

Instead of modifying OpenClaw's core, we expose an OpenAI-compatible endpoint (/v1/chat/completions) that internally routes requests to OCI Generative AI.

This approach provides:

✅ Full OpenClaw compatibility
✅ Control over OCI model mapping
✅ Support for streaming responses
✅ Enterprise-grade OCI infrastructure
✅ Secure request signing via OCI SDK

Secure Enterprise Prompting

In this material, you have a prompt file that will be incorporate into the OCI OpenAPI Proxy (oci_openai_proxy.py). The prompt (pptx_runner_policy_strict.txt) was created to generate a automatic PowerPoint presentation based on any web documentation (github, docs.oracle.com). This example demonstrates a more enterprise secure way to use the OCI IAM to control the cloud resources like Object Storage and LLM.

See the prompt that is incorporated in the OCI OpenAI Proxy:

Whenever the user requests PPTX generation with external material (link, file, or text):

----------------------------------------------
STEP 0 – FIXED WORKING DIRECTORY (MANDATORY)
----------------------------------------------

All operations MUST occur inside:
    $HOME/.openclaw/workspace/openclaw_folder

Execute:
    cd $HOME/.openclaw/workspace/openclaw_folder

----------------------------------------------
STEP 1 – PREPARATION (MANDATORY)
----------------------------------------------

The file generate_openclaw_ppt_template.py is located in $HOME/.openclaw/workspace/openclaw_folder
The file read_url is located in $HOME/.openclaw/workspace/openclaw_folder
The file read_file is located in $HOME/.openclaw/workspace/openclaw_folder

Required:

    read_url for links
    read_file for local files

GITHUB LINK HANDLING (REQUIRED)

    If the link contains:
    github.com/.../blob/...
    Automatically convert to:
    raw.githubusercontent.com/USER/REPO/BRANCH/PATH
    BEFORE calling read_url.

    Example:
    Original:
    https://github.com/user/repo/blob/main/app.py
    Convert to:
    https://raw.githubusercontent.com/user/repo/main/app.py
    Then call:
    read_url <raw_url>

    If the returned content contains <html or <script>, extract only visible text, removing HTML tags.

    * If the content cannot be read successfully → ABORT.

MANDATORY PIPELINE:

    1) Save material to file:
        (exec read_url <url> > $HOME/.openclaw/workspace/openclaw_folder/material_raw.txt)

    2) Analyze material_raw.txt and generate content.json explicitly:
        (exec cat > $HOME/.openclaw/workspace/openclaw_folder/content.json << 'EOF'
        <valid JSON only>
        EOF)

        Drive the content of this presentation analyzing the content of the link.

        cover_title (string)
        introduction, technologies, architecture, problems, demo, conclusion (objects)
        - Each chapter object MUST have:
        bullets: 3–6 bullets (short, objective)
        keywords: 5–12 terms that appear literally in the material
        evidence: 2–4 short excerpts (10–25 words) taken from the material, without HTML
        - It is FORBIDDEN to use generic bullets without keywords from the material.
        - VALIDATION: if it is not possible to extract at least 20 unique keywords from the total material → ABORT.

    3) Validate JSON:
        (exec python -m json.tool $HOME/.openclaw/workspace/openclaw_folder/content.json)

        Only after successful validation:
        (exec export OCI_LINK_DEMO="<url>")
        (exec python generate_openclaw_ppt_template.py)

----------------------------------------------
STEP 2 – MODIFICATION VALIDATION [STRICT VERSION]
----------------------------------------------

Before running:

    - Verify that each chapter contains at least 1 literal keyword from the material.
    - Verify that at least 8 keywords appear in 4 or more slides.
    - Verify that each chapter contains at least 1 piece of evidence.
    If it fails → ABORT.

----------------------------------------------
STEP 3 – EXECUTION
----------------------------------------------

Only now execute:

    SET THE ENVIRONMENT VARIABLE WITH THE URL PASSED AS A BASIS FOR DOCUMENTATION AND THE FILE NAME GENERATED WITH CONTENT READ FROM THE LINK:

    `export OCI_LINK_DEMO=<link passed as documentation>`
    `export OCI_CONTENT_FILE=<NAME OF THE GENERATED FILE>`
    `python $HOME/.openclaw/workspace/openclaw_folder/generate_openclaw_ppt_template.py`

----------------------------------------------
STEP 4 – UPLOAD
----------------------------------------------

    First, delete the file in object storage: `openclaw_oci_presentation.pptx`

    And only then upload it to Object Storage: `oci os object put \
    --bucket-name hoshikawa_template \
    --file` $HOME/.openclaw/workspace/openclaw_folder/openclaw_oci_presentation.pptx \
    --force

----------------------------------------------
STEP 5 – GENERATE PRE-AUTH LINK
----------------------------------------------
    oci os preauth-request create ...

The custom prompt is incorporated here:

PROMPT_PATH = os.path.expanduser("pptx_runner_policy_strict.txt")
def load_runner_policy():
    if os.path.exists(PROMPT_PATH):
        with open(PROMPT_PATH, "r", encoding="utf-8") as f:
            return f.read()
    return ""
RUNNER_POLICY = load_runner_policy()

RUNNER_PROMPT = (
        RUNNER_POLICY + "\n\n"
                        "You are a Linux execution agent.\n"
                        "\n"
                        "OUTPUT CONTRACT (MANDATORY):\n"
                        "- You must output EXACTLY ONE of the following per response:\n"
                        "  A) (exec <command>)\n"
                        "  B) (done <final answer>)\n"
                        "\n"
                        "STRICT RULES:\n"
                        "1) NEVER output raw commands without (exec <command>). Raw commands will be ignored.\n"
                        "2) NEVER output explanations, markdown, code fences, bullets, or extra text.\n"
                        "3) If you need to create multi-line files, you MUST use heredoc inside (exec <command>), e.g.:\n"
                        "   (exec cat > file.py << 'EOF'\n"
                        "   ...\n"
                        "   EOF)\n"
                        "4) If the previous tool result shows an error, your NEXT response must be (exec <command>) to fix it.\n"
                        "5) When the artifact is created successfully, end with (done ...).\n"
                        "\n"
                        "REMINDER: Your response must be only a single parenthesized block."
)

PPTX Builder

A PPTX builder will generate a professional PowerPoint deck from a template (.pptx) + a structured content.json

The goal is to keep OpenClaw fully compatible with the OpenAI protocol while moving inference to OCI and enabling artifact generation (PPTX) using a repeatable, governed pipeline.

Architecture

OpenClaw
  ↓ (OpenAI protocol)
OpenAI-compatible Gateway (FastAPI)
  ↓ (signed OCI REST)
OCI Generative AI (chat endpoint)
  ↓
LLM response

(Optional)
Material (URL / file / text)
  ↓
content.json (validated / governed)
  ↓
PPTX Builder (template + content.json)
  ↓
openclaw_oci_presentation.pptx

Project structure

project/
 ├── oci_openai_proxy.py                 # FastAPI OpenAI-compatible gateway -> OCI GenAI
 ├── pptx_runner_policy_strict.txt       # Strict policy for extracting/validating material -> content.json
 ├── openclaw.json                       # Example OpenClaw config using the gateway
 └── README.md
 AND these files:
 ├── generate_openclaw_ppt_template.py   # PPTX generator (template + content.json)
 ├── read_url_and_read_file.sh           # Helper script to create read_url/read_file in OpenClaw workspace
 └── template_openclaw_oci_clean.pptx    # You MUST have one template here

 
 Move these files to:
 $HOME/.openclaw/workspace/openclaw_folder
 ├── generate_openclaw_ppt_template.py   # PPTX generator (template + content.json)
 ├── read_url_and_read_file.sh           # Helper script to create read_url/read_file in OpenClaw workspace
 └── template_openclaw_oci_clean.pptx    # You MUST have one template here

Part A — OpenAI-compatible Gateway (OpenClaw → OCI GenAI)

Why OCI Generative AI?

OCI provides what enterprises usually need:

IAM & compartments
Signed requests (no API key leakage)
Regional control / sovereignty
VCN options
Observability integration
Production-grade inference endpoints

By putting an OpenAI-compatible API in front of OCI, you get:

✅ OpenClaw compatibility
✅ Model mapping (OpenAI names → OCI modelIds)
✅ Streaming compatibility (simulated if OCI returns full text)
✅ Governance inside your tenancy

Requirements

Python 3.10+ (recommended)
OCI config file (~/.oci/config) + API key
Network access to OCI GenAI endpoint

Install dependencies:


pip install fastapi uvicorn requests oci pydantic

Configuration (environment variables)

The gateway reads OCI configuration using environment variables (defaults shown):


export OCI_CONFIG_FILE="$HOME/.oci/config"
export OCI_PROFILE="DEFAULT"
export OCI_COMPARTMENT_ID="ocid1.compartment.oc1..."
export OCI_GENAI_ENDPOINT="https://inference.generativeai.<region>.oci.oraclecloud.com"

Run the server


uvicorn oci_openai_proxy:app --host 0.0.0.0 --port 8050

Test with curl


curl http://127.0.0.1:8050/v1/chat/completions   -H "Content-Type: application/json"   -d '{
    "model": "gpt-5",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Or if you want to generate a PPTX direct by the oci_openai_proxy.py:


curl http://127.0.0.1:8050/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
        "model": "gpt-5",
        "messages": [
            {"role": "user", "content": "generate pptx from https://github.com/hoshikawa2/flexcube-14.5 in portuguese"}
        ],
        "temperature": 0.2
      }'

OpenClaw configuration (openclaw.json)

Point OpenClaw to the gateway:

baseUrl → your local gateway (port 8050)
api → openai-completions
model id → must match a MODEL_MAP key inside oci_openai_proxy.py

Example provider block:

{
  "models": {
    "providers": {
      "openai-compatible": {
        "baseUrl": "http://127.0.0.1:8050/v1",
        "apiKey": "sk-test",
        "api": "openai-completions"
      }
    }
  }
}

Part B — PPTX generation from a template (Template → Deck)

What it does

generate_openclaw_ppt_template.py builds a fixed 7-slide strategic deck:

Cover
Intro (use case)
Technologies
Architecture
Problems
Demo (includes the source link)
Conclusion

The deck is generated from:

a PPTX template (with expected layouts),
a content.json file,
and a OCI_LINK_DEMO link (material source shown on the Demo slide).

Inputs

1) PPTX template

You MUST have a PowerPoint template named template_openclaw_oci_clean.pptx with some master layout slides.

Default expected layout names inside the template:

Cover 1 - Full Image
Full Page - Light

You can change the template by passing --template or PPTX_TEMPLATE_PATH.

2) content.json

content.json must contain:

cover_title (string)
introduction, technologies, architecture, problems, demo, conclusion (objects)

Each section object must include:

bullets: 3–6 short bullets
keywords: 5–12 keywords that appear literally in the material
evidence: 2–4 short excerpts (10–25 words) extracted from the material (no HTML)

The strict validation rules are described in pptx_runner_policy_strict.txt.

Configure paths

Create a folder named openclaw_folder inside the $HOME/.openclaw/workspace.


cd $HOME/.openclaw
mkdir openclaw_folder
cd openclaw_folder

Put these files into the openclaw_folder:

generate_openclaw_ppt_template.py
read_url_and_read_file.sh 
template_openclaw_oci_clean.pptx (Your PPTX template if you have)

Run this command only one time:

bash read_url_and_read_file.sh

This will generate the read_url and read_file tools.

You can run everything without hardcoded paths using either CLI flags or environment variables.

Environment variables

# Optional: where your files live (default: current directory)
export OPENCLAW_WORKDIR="$HOME/.openclaw/workspace/openclaw_folder"

# Template + output
export PPTX_TEMPLATE_PATH="$OPENCLAW_WORKDIR/template_openclaw_oci_clean.pptx"
export PPTX_OUTPUT_PATH="$OPENCLAW_WORKDIR/openclaw_oci_presentation.pptx"

# Content JSON (if not set, defaults to $OPENCLAW_WORKDIR/content.json)
export OCI_CONTENT_FILE="$OPENCLAW_WORKDIR/content.json"

# Source link shown on the Demo slide
export OCI_LINK_DEMO="https://docs.oracle.com/en-us/iaas/Content/generative-ai/home.htm"

CLI usage

python generate_openclaw_ppt_template.py   --template "$PPTX_TEMPLATE_PATH"   --output "$PPTX_OUTPUT_PATH"   --content "$OCI_CONTENT_FILE"   --link "$OCI_LINK_DEMO"

End-to-end pipeline (URL → content.json → PPTX)

A typical (strict) flow:

Read material (URL or local file)
Generate content.json following the strict policy
Validate JSON
Generate PPTX

Helper scripts (read_url / read_file)

The repository includes read_url e read_file.sh to install helper scripts into your OpenClaw workspace.

Example:

bash "read_url e read_file.sh"

Then:

# Read URL
~/.openclaw/workspace/openclaw_folder/read_url "https://example.com" > material_raw.txt

# Read local file
~/.openclaw/workspace/openclaw_folder/read_file "/path/to/file.pdf" > material_raw.txt

Validate JSON

python -m json.tool "$OCI_CONTENT_FILE" >/dev/null

Generate PPTX

python gera_oci_ppt_openclaw_template.py --link "$OCI_LINK_DEMO"

Deploying (common options)

Option 1 — Run locally (developer laptop)

Run the gateway with uvicorn
Generate decks on demand in the workspace folder

Option 2 — Server VM (systemd for gateway)

Create a systemd service (example):

[Unit]
Description=OpenAI-compatible OCI GenAI Gateway
After=network.target

[Service]
WorkingDirectory=/opt/openclaw-oci
Environment=OCI_CONFIG_FILE=/home/ubuntu/.oci/config
Environment=OCI_PROFILE=DEFAULT
Environment=OCI_COMPARTMENT_ID=ocid1.compartment...
Environment=OCI_GENAI_ENDPOINT=https://inference.generativeai.<region>.oci.oraclecloud.com
ExecStart=/usr/bin/python -m uvicorn oci_openai_proxy:app --host 0.0.0.0 --port 8050
Restart=always

[Install]
WantedBy=multi-user.target

Option 3 — Containerize

Put oci_openai_proxy.py inside an image
Mount ~/.oci/config read-only
Pass the same env vars above

(Exact Dockerfile depends on how you manage OCI config and keys in your environment.)

Troubleshooting

PPTX builder errors

Layout not found: your template does not have the expected layout names.
Too few placeholders: your selected layout must have at least 2 text placeholders.
Exactly 7 slides: the generator enforces the fixed structure.

Content issues

If content.json has generic bullets/keywords not present in the material, the strict policy should fail validation.
If you cannot extract enough literal keywords, re-check your material extraction (HTML removal, raw GitHub URL, etc.).

Test the Solution

Go to the openclaw dashboard:

openclaw dashboard

Try this:

generate a pptx based on this material https://github.com/hoshikawa2/openclaw-oci

And you get a temporary OCI Object Storage link:

This is the oci_openai_proxy.py monitoring output:

And the Presentation generated is:

Final Notes

You now have:

✔ OpenClaw fully integrated
✔ OCI Generative AI backend
✔ Streaming compatibility
✔ Enterprise-ready architecture

Reference

Acknowledgments

Author - Cristiano Hoshikawa (Oracle LAD A-Team Solution Engineer)

21 KiB Raw Permalink Blame History Unescape Escape