Skill Verified

openai-slides-skill

Creates and edits PowerPoint presentation files with slide layouts, text formatting, images, charts, and structured content using PptxGenJS and Python libraries.

77

/ 100 · Grade C

C = 70–79

I need to create professional PowerPoint presentations, convert existing slides from screenshots or PDFs, modify slide content while maintaining editability, or fix layout issues in presentation decks.

document generationpresentation toolsoffice automationcontent creationfile processing
Publisher: openaiVersion: latestCertified: Mar 28, 2026Expires: Mar 28, 2027Source ↗

openai-slides-skill earned Verified status with a trust score of 77/100 (Grade C). Adversarial testing produced 1 finding (1 high). Security scan flagged 8 findings. Tier is Verified rather than Certified due to unmitigated findings above severity thresholds.


Trust Score Breakdown

Eight weighted signals composing the aggregate trust score

security scan
0% × 15w
0.0
supply chain
100% × 10w
10.0
adversarial
83% × 25w
20.8
provenance
80% × 20w
16.0
consumer confirm
100% × 10w
10.0
behavioral pass
100% × 10w
10.0
contract accuracy
100% × 6w
6.0
uptime
100% × 4w
4.0

Scheme v2.0 · Weights provisional · Consumer confirmations and uptime use pipeline-derived baselines.


Findings

Security scan results, adversarial testing, and pipeline review

Security Scan — Cisco Skill Scanner

cisco-skill-scannerFAIL
staticbehavioralllmmeta
critical: 2high: 2medium: 3low: 1

Finding details

criticalcommand injectionstatic

Pattern detected: require("child_process")

criticalcommand injectionstatic

Pattern detected: spawnSync(

highdata exfiltrationstatic

Pattern detected: fs.readFileSync(

highcommand injectionllm

Multiple Python scripts use subprocess.run() with user-controlled input without proper sanitization. The render_slides.py script constructs shell commands using file paths that could contain malicious characters, and ensure_raster_image.py executes external tools (inkscape, gs, heif-convert, JxrDecApp) with user-provided file paths. An attacker could craft malicious filenames to inject shell commands.

mediumcommand injectionllm

The JavaScript helper files use Node.js require() to load child_process modules and filesystem modules. While not directly executing commands in the provided code, the presence of these imports in a skill package creates potential for command injection if the helpers are modified or if additional code uses these modules unsafely.

mediumdata exfiltrationllm

The JavaScript helper modules have unrestricted access to the file system through Node.js fs module. The code reads files using fs.readFileSync() and require.resolve(), which could potentially be used to access sensitive files on the system if file paths are not properly validated.

mediumunauthorized tool usestatic

Pattern detected: sudo apt-get install

lowskill discovery abusellm

The skill manifest is missing optional metadata fields including license, compatibility, and allowed-tools. While these fields are optional per the agent skills specification, their absence makes it harder to assess the skill's intended scope and restrictions.

Adversarial Testing — 3 categories, 1 findings

prompt injection chainscapability squattingcontext poisoning
highWarnprompt injection chainssuspected

The skill instructions reference user-provided content (screenshots, PDFs, reference decks, slide content) that gets processed and incorporated into slide generation without explicit data-boundary markers. The workflow section states 'Inspect the request and determine whether you are creating a new deck, recreating an existing deck, or editing one' and 'Recreate Or Edit Existing Slides' section mentions processing 'reference PDF' and 'reference slide' content. User content could potentially contain instruction-like directives that override the skill's behavior during the slide creation process.

Methodology v1.0 · 3 categories · ~55 attack patterns

Pipeline Review — 3 findings

criticalsecurityBLOCKING

cisco_skill_scanner: critical finding — command_injection

highsecurity

cisco_skill_scanner: high finding — data_exfiltration

highsecurity

cisco_skill_scanner: high finding — command_injection


Behavioral Fingerprint

Runtime performance baseline for drift detection

Samples

8

Error rate

0.0%

Peak memory

— MB

Avg CPU

—%

Response time distribution

p50: 26713msp95: 28685msp99: 28685ms

Output size distribution

p50: 4.9 KBp95: 5.2 KBmean: 3.6 KB

Fingerprint v1.0 · Baseline: Mar 28, 2026 · Status: baseline


Interface

Skill triggers and instruction summary

Activation

This skill activates when tasks involve creating, editing, or recreating PowerPoint presentation slide decks (.pptx files).

This skill handles slide deck creation, modification, layout validation, and rendering using PptxGenJS with bundled helper utilities.

Instructions: 48Files: 4Format: markdown

Does

Use PptxGenJS for slide authoring instead of python-pptx for generation

Work in task-local directories and copy final artifacts after validation

Copy bundled pptxgenjs_helpers folder into workspace and import locally

Set slide size upfront with 16:9 default unless source material indicates otherwise

Set theme fonts explicitly and use helper functions for text sizing

Run validation scripts to check for overflow and layout issues

Deliver both .pptx file and source .js file with any required assets

Include overlap and bounds checking warnings in generated JavaScript

Render source decks first when recreating or editing existing slides

Does not

Use python-pptx for deck generation unless task is inspection-only

Rely on PowerPoint defaults for typography when it matters

Use PptxGenJS fit or autoFit for text sizing

Use literal bullet characters instead of bullet options

Use PptxGenJS built-in image sizing instead of helper functions


Scope & Permissions

What this capability can and cannot access — derived from pipeline analysis

creates files

yes

deletes files

no

modifies files

yes

accesses env variables

no

invokes external tools

yes

makes network requests

no


Known Failure Modes

Documented edge cases and recovery behaviors

when when layout issues are detected

then the agent fixes all unintentional overlap and out-of-bounds warnings before delivery

when when fonts are missing or substituted

then the agent reports and addresses font issues using detect_font.py

when when slide content overflows canvas

then the agent uses slides_test.py to detect and fix overflow issues


Badge & Integration

Embed certification status in your README, docs, or CI pipeline

Fidensa Verified badge for openai-slides-skill
badge SVG →attestation API →integration guide →

Certification Notes

Provenance observations from the pipeline

publisher

Publisher "openai" is not verified — first certification from this publisher

provenance

No SECURITY.md or SECURITY.txt file found — no published vulnerability reporting process

provenance

Single contributor — no peer review evidence in commit history

provenance

Repository is 3 days old — recently created

provenance

Package description appears to be boilerplate or template text


Signed Artifact

Certification provenance and verification metadata

Content hashsha256:982dfb1129598cca163c4aa60599e5ee69885d982f37156b59053b3aa397eed2
Key IDkms-9db4ed3b9f53
CertifiedMar 28, 2026
ExpiresMar 28, 2027
Pipeline version1.0
Statusvalid