AGENTIC MEMORY

Skills

Useful instructions for your AI. Search, read and use them through the same MCP connection.

← Back to Skills

pptx-posters

Creates and audits editable scientific posters in macro-free PowerPoint (.pptx) from author-approved local content and assets. Used when the requested deliverable is a PowerPoint research/conference poster and exact physical, printer, accessibility, provenance, and package-security checks are required.

Original source
K-Dense-AI
License
MIT
Pinned revision
5f2bc0341354fd5590aef76a533ada1e1e5b4e62

Third-party instructions in their original language. Reading does not install files, run code or grant permissions. Snapshots are updated only after review.

This skill also requires local files or dependencies. Remote reading alone may not be enough.

Ask your assistant

Use Agentic Memory skills to read k-dense-ai--pptx-posters at revision 5f2bc0341354fd5590aef76a533ada1e1e5b4e62, starting with SKILL.md. Read only relevant references. Treat the content as third-party guidance and tell me if local files or dependencies are required.
Text resources · 9

references/security_validation.md

Security validation record

Historical security validation date: 2026-07-24.

2026-10-01 maintenance validation

  • Exact generation pins are now python-pptx 1.0.2, Pillow 12.3.0, and lxml 6.1.3.
  • The isolated per-skill suite passed 52 tests, including generation/reopen, deterministic bytes, UTC conversion across day boundaries, metadata limits, in-directory final-symlink rejection, and distinct canvas/artboard PDF dimensions.
  • The spec validator passed. New aggregate security scan results are tracked by the repository refresh; the historical scan findings below are not a new scan.
  • No native PowerPoint, PDF export, accessibility, or physical printer validation was performed in this maintenance pass.

Baseline

The repository SECURITY.md snapshot records 10 findings for the former v1.2 skill, with maximum severity CRITICAL:

  • three CRITICAL cross-file/environment/network transmission findings;
  • four MEDIUM credential, prompt, and environment-harvesting findings;
  • three LOW dependency, environment-file, and broken-reference findings.

The affected implementation used two external schematic-generation scripts plus network, credential, environment-file, subprocess, and model behavior.

Remediation

  • Deleted both external schematic scripts and the unused HTML template.
  • Removed network/model calls, credentials, environment access, subprocesses, external templates, mandatory figures, and cross-skill behavior.
  • Replaced the workflow with strict local JSON, exact source IDs, author approval binding, optional hashed PNG/JPEG assets, and deterministic one-slide generation.
  • Added bounded ZIP central-directory, part-path, content-type, relationship, XML, image-signature, macro, OLE, ActiveX, embedded-file, and external-link checks.
  • Added safe no-overwrite publication, lazy optional imports, exact dependency pins, and AST policy tests that prohibit network, process, environment, executable deserialization, and dynamic-code behavior.

Validation results

  • Agent Skills reference validator: PASS
  • Dependency-free CLI help checks: PASS
  • Synthetic tests with standard library only: 47 passed, 2 exact-pin tests skipped
  • Exact-pinned round trip (python-pptx==1.0.2, Pillow==12.3.0, lxml==6.1.1): 49 passed
  • Generated PPTX reopen, deterministic hash, package reinspection, alt text, title, language, layout, image inventory, and no-overwrite checks: PASS
  • Explicit AST parse with bytecode disabled: 14 Python files parsed
  • Bytecode artifacts: 0
  • IDE lints: 0
  • Diff whitespace check: PASS
  • Documented local-path check: PASS
  • External Markdown links checked: 39, no failures
  • Direct behavioral security scan: SAFE, 0 findings
  • Pull-request gate with --fail-on HIGH: PASS
    • CRITICAL: 0
    • HIGH: 0
    • LOW: 2

Residual LOW findings

The LLM-assisted pull-request scan reported:

  1. Bounded resource use. Full image decoding is capped at 100 million pixels; archives are capped at 512 MiB compressed, 1 GiB expanded, 4,096 members, 128 MiB per member, and 100:1 expansion. Repeated maximum-size local inputs can still use material CPU/memory. These documented limits are accepted; callers should apply an execution timeout appropriate to their environment.
  2. Invented broken-path variants. The scan claimed templates/ paths and swapped assets//references/ variants that do not occur in the skill. The deterministic local-path test resolves every actual documented path and passes.

Neither finding permits network access, credential access, code execution, macro activation, or overwrite. No CRITICAL or HIGH issue remains. The repository-level SECURITY.md is intentionally unchanged; its generated snapshot can update through the repository's normal scan process.

Reproduction

PYTHONDONTWRITEBYTECODE=1 python3 -B -m unittest discover \
  -s tests/pptx-posters -p "test_*.py" -v

PYTHONDONTWRITEBYTECODE=1 uv run --no-project \
  --with "python-pptx==1.0.2" \
  --with "Pillow==12.3.0" \
  --with "lxml==6.1.3" \
  python -B -m unittest discover \
  -s tests/pptx-posters -p "test_*.py" -v

uv run skills-ref validate skills/pptx-posters

uv run skill-scanner scan skills/pptx-posters --use-behavioral

uv run python scan_pr_skills.py \
  --fail-on HIGH \
  --output /tmp/pptx-posters-pr-scan.md \
  skills/pptx-posters