Octos is a self-hosted AI assistant that runs on a user's computer and can connect to different AI providers, execute code, browse the web, remember information, schedule jobs, and create documents through browser, terminal, and messaging interfaces. The catalogue entries extend or configure this assistant's workflows.
Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add octos-org/octos --skill pptx-runtimegit clone --depth 1 https://github.com/octos-org/octosWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/octos-org/octos/pptx-runtime)<a href="https://agentmods.dev/skills/octos-org/octos/pptx-runtime"><img src="https://agentmods.dev/badge/skills/octos-org/octos/pptx-runtime/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/octos-org/octos/pptx-runtime"><img src="https://agentmods.dev/badge/skills/octos-org/octos/pptx-runtime.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00025 | $0.02886 |
| Opus 5.5 | $0.00010 | $0.01154 |
| Sonnet 5.5 | $0.00005 | $0.00577 |
| Haiku 4.5 | $0.00003 | $0.00289 |
Grade A, and why
pptx-runtime scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured today.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 206 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Call pptx_tool with an action. Read the pptx skill and its references/rust-runtime.md
before making a deck. This binary provides mechanics, not content, visual decisions,
an image generator or another task scheduler. Use the existing image-generation tool
and peer sessions selected for the task. Read the returned PNG files with the host's
image-capable read_file tool; paths alone are not visual review.
Choose local text targets and same-role sizes from the artwork before fitting.
patch merges partial set.format values by default and reports effective
before/after formatting; use explicit format_mode:"replace" to reset overrides.
split_text partitions one native text into independently placed fragments,
retains the canonical copy, and validates complete non-overlapping copy_range
coverage. It cannot replace a written symbol with a claimed raster equivalent.
See the production skill's references/rust-runtime.md for these schemas.
cycle also emits native QA for every text object. evidence_record independently
remeasures the actual PPTX and cannot clear machine failures through model
pass/dismissal claims. Use qa_check for a standalone check (CLI exit 2 means
fail/unknown). Call action: qa_help for the embedded constraint schemas,
repair feedback, coverage limits and the distinction between mechanical and visual
acceptance. No extra model-written harness or checking script is needed.
Read native_qa.diagnosis before deciding to repair again. It separates measured
constraint violations, visual questions and measurement gaps. An unsupported
owner boundary is not automatically a misplaced label; a raster-contact question
is not automatically an intentional overlay. Diagnosis changes neither acceptance
nor the existing exit policy. See references/qa-gates.md.
For readable whole-page coverage, call review_pack with cycle and a fresh
output directory. Supply previous_cycle after a revision. It writes separate,
unresized current/blank/design quadrant images independent of detector findings,
matched before/current contextual crops for changed objects, and the persistent,
new and no-longer-detected mechanical findings. The report never labels a blank
as a design reference. Read the images using the host's image-capable tool and
give the pack to the existing independent peer; generating it is not inspection.
pages optionally limits an intermediate check; final review must cover all pages.
Changed acceptance targets or measurement criteria cannot count as repaired defects.
review_pack also produces a detector-independent inventory of every text
object, matched native-pixel crops, small view_batches, and an unreviewed
object-review-template.json. Follow object visual review.
evidence_record requires current object-level observations for ownership,
containment, alignment, direction and typography; missing/stale observations stay
unknown. Whole-page claims cannot substitute for this evidence. Use the selected
model's own vision; no additional vision service is introduced.
New tasks separately require reference fidelity for objects with design images. Readable text cannot waive new interference or changed layout. Rust binds the original/blank/current evidence and registers confirmed defects before repair. See review lifecycle for structured observations, safe local background repairs, third-session arbitration, and bounded exit.
The index also includes relation_signals: local surface directions/offsets and
canonical-text OCR reference candidates. Treat them as measured questions, not
automatic defects. intentional requires a measured support_id from that
signal's intentional_options; a prose explanation alone cannot close it.
No supported explanation means unknown or defect, never an invented pass.
Raster OCR/ink overlap is collected automatically. Native glyph contact with
blank-image contrast edges is measured without OCR and also checked afresh by
the final native QA gate. Read the returned contact points and local crops;
contact is a candidate interference, not proof that the artwork is an arrow or
that an overlap is unintended. Keep unresolved overlays unknown, not silently
passed or automatically erased.
Submit authored object and relation observations through
review_submit with pack, observations and a fresh output; Rust binds hashes
and checks disposition consistency. OCR-assisted anchors currently require macOS
Vision; surface measurements work without OCR. Missing evidence is not approval.
What ships with it
60 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- build.rs 758 B
- Cargo.toml 1.5 KB
- licenses/DeckCraft-MIT.txt 1.1 KB
- licenses/DeckCraft-NOTICE.txt 6.4 KB
- licenses/Krilla-ICC.txt 6.4 KB
- licenses/Ubuntu-Font-License.txt 4.6 KB
- manifest.json 17 KB
- references/codex-parity.md 5.6 KB
- references/evaluation.md 1.7 KB
- references/object-visual-review.md 7.2 KB
- references/placement.md 3.6 KB
- references/qa-gates.md 21 KB
- references/review-lifecycle.md 11 KB
- src/artwork_contact.rs 5.2 KB
- src/authoring.rs 10 KB
- src/check_feedback.rs 3.2 KB
- src/comparison_review.rs 24 KB
- src/composite_ink.rs 11 KB
- src/deck.rs 15 KB
- src/evidence.rs 31 KB
- src/feedback.rs 26 KB
- src/fidelity.rs 16 KB
- src/finding_guard.rs 17 KB
- src/fit.rs 29 KB
- src/fs.rs 4.0 KB
- src/geometry.rs 5.3 KB
- src/intent_evidence.rs 6.7 KB
- src/label_diff.rs 19 KB
- src/lib.rs 9.1 KB
- src/main.rs 3.3 KB
- src/object_review.rs 25 KB
- src/ocr.rs 6.6 KB
- src/operations.rs 5.4 KB
- src/placement.rs 26 KB
- src/qa.rs 34 KB
- src/raster_overlap.rs 7.4 KB
- src/region_layout.rs 14 KB
- src/relations.rs 21 KB
- src/review_reuse.rs 11 KB
- src/review_routes.rs 11 KB
- src/review.rs 33 KB
- src/spec.rs 21 KB
- src/typography.rs 4.9 KB
- src/visual_diff.rs 11 KB
- src/visual_proof.rs 5.3 KB
- tests/artwork_contact.rs 6.7 KB
- tests/authoring_operations.rs 13 KB
- tests/check_feedback.rs 2.2 KB
- tests/common/mod.rs 1.3 KB
- tests/comparison_coverage.rs 14 KB
- tests/composite_ink.rs 4.5 KB
- tests/feedback_parity.rs 28 KB
- tests/fitting_policy.rs 7.8 KB
- tests/geometry_contract.rs 1.8 KB
- tests/intent_evidence.rs 4.4 KB
- tests/label_comparison.rs 7.7 KB
- tests/object_review.rs 15 KB
- tests/placement.rs 9.1 KB
- tests/qa_diagnostics.rs 4.2 KB
- tests/qa_gates.rs 25 KB
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- today First seen · 206 lines · 25 tokens per session scan A 584c7d9ea403
pptx-runtime is a skill published in the GitHub repository octos-org/octos (766 stars, last pushed today), licensed Apache-2.0. It adds 25 tokens to every session and 2,886 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-11-11.
pptx-runtime compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| pptx-runtimeoctos-org/octos | Skill | 766 | today | Apache-2.0 | A | 25 | Claude Code, Codex |
| kamitw93/Kami | Skill | 13k | yesterday | MIT | A | 78 | Claude Code |
| solidworks-engineering-drawingwzyn20051216/solidworks-automation-skill | Skill | 1.9k | yesterday | MIT | A | 58 | Claude Code, Codex |
| document-designjamditis/claude-skills-journalism | Skill | 452 | 7d ago | MIT | A | 31 | Claude Code, Codex |
| design-export-repairOneWave-AI/claude-skills | Skill | 425 | 22d ago | MIT | A | 219 | Claude Code, Codex |
| architectural-drawing-parserTerminalSkills/skills | Skill | 197 | 3d ago | Apache-2.0 | A | 62 | Claude Code, Codex |
| paperlab_book_typesetting_releaseequinor/neqsim | Skill | 168 | yesterday | Apache-2.0 | A | 55 | Claude Code, Codex |