Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

codeck export — @export lane · hiyeshu | DecimalAI

hiyeshu/codeck-export

SkillScore45Maintained

Internal publisher module for /codeck. Exports deck to PDF or PPTX with post-export QA. User-facing export requests should enter through /codeck; this skill defines the export behavior.

design-mediaby hiyeshu

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Doc Ops pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/hiyeshu-codeck-export/SKILL.md

decimalai skills pull hiyeshu-codeck-export --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/hiyeshu-codeck-export/SKILL.md https://app.decimal.ai/s/hiyeshu-codeck-export/SKILL.md

Raw file: /s/hiyeshu-codeck-export/SKILL.md ↗ · machine-readable summary at /s/hiyeshu-codeck-export.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 37d ago

—+85% tokens0% turns45% of whole cases pass with skill (10/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-10	✗→✓	▲ Improved	79%	0%
case-12	✗→✓	▲ Improved	79%	0%
case-14	✗→✓	▲ Improved	63%	0%
case-15	✗→✓	▲ Improved	151%	0%
case-17	✗→✓	▲ Improved	198%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/10/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

hiyeshu / codeck

skills/codeck-export/SKILL.md

View on GitHub ↗

License MITImported 8/10/2026Version v2

Skill contents

SKILL.md

codeck export — @export lane

Minimum conversation, maximum output. Export the deck to the user's format.

@export owns PDF/PPTX output and export QA.

Write boundaries:

May write final PDF/PPTX files in the user's project directory
May write export QA notes to $DECK_DIR/review.md or $DECK_DIR/MEMORY.md
May update $DECK_DIR/roles/export.md, $DECK_DIR/tasks/tasks.md, and $DECK_DIR/channel/YYYY-MM-DD.md
Must not edit deck.md, DESIGN.md, custom.css, slides.html, or speech.md
Export defects that require source changes become proposals in $DECK_DIR/threads/threads.md

Step 1: Status

bash
DECK_DIR="$HOME/.codeck/projects/$(basename "$(pwd)")"
CODECK_SKILL_DIR="${CODECK_SKILL_DIR:-}"
if [ -z "$CODECK_SKILL_DIR" ]; then
  for d in \
    "${CLAUDE_PLUGIN_ROOT}/skills/codeck" \
    "$HOME"/.claude/plugins/cache/*/codeck/*/skills/codeck \
    "$HOME"/.codex/plugins/cache/*/codeck/*/skills/codeck \
    "$HOME/.agents/skills/codeck" \
    "$HOME/.codex/skills/codeck" \
    "$HOME/.claude/skills/codeck"; do
    if [ -d "$d/scripts" ]; then CODECK_SKILL_DIR="$d"; break; fi
  done
fi
[ -n "$CODECK_SKILL_DIR" ] || { echo "codeck skill scripts not found; set CODECK_SKILL_DIR" >&2; exit 1; }
. "$CODECK_SKILL_DIR/scripts/resolve-dirs.sh"
mkdir -p "$DECK_DIR"
mkdir -p "$DECK_DIR/channel" "$DECK_DIR/tasks" "$DECK_DIR/threads" "$DECK_DIR/roles"
bash "$CODECK_SKILL_DIR/scripts/init-room.sh" "$DECK_DIR"
bash "$CODECK_SKILL_DIR/scripts/status.sh" "$DECK_DIR"

Read $DECK_DIR/MEMORY.md, active rows in $DECK_DIR/tasks/tasks.md, and open rows in $DECK_DIR/threads/threads.md. Do not read channel/YYYY-MM-DD.md unless debugging history.

Gate check: if no assembled HTML exists (./*-r*.html), run /codeck to generate/rebuild the deck first.

Before export, claim the work ticket:

markdown
@orchestrator
Owner: @export. Task: export latest HTML and run export QA.

@export
I claim the export pass. I will write PDF/PPTX outputs and record any source defects as threads.

Append the exchange to today's channel file and update tasks/tasks.md.

Step 2: Format

Export Format is one allowed Decision Ask moment under /codeck.

Use the shared /codeck Decision Ask Policy. Skip Decision Ask when the user names a target:

PDF, print, save as PDF → export PDF
PPTX, PowerPoint, slides file → export PPTX
all, both → export PDF and PPTX

If the user only says "export", create a non-blocking D-YYYYMMDD-NN decision in threads/threads.md and render once:

text
codeck needs the export format.

Current read: HTML already exists as the audience preview.

I suggest PDF because it preserves the visual layout and is easiest to share.

A) PDF (recommended)
B) PPTX
C) PDF + PPTX

If the user does not answer or structured AskUser UI is unavailable, export PDF and write assumed default to MEMORY.md.

Do not offer HTML as an export choice. HTML is the preview/source artifact.

Step 3: Export

Find the HTML file (./*-r*.html in the user's project directory), derive baseName.

HTML

Already exists. Open in browser, F for fullscreen, arrows to navigate.

PDF — print from HTML

Use Playwright for WYSIWYG PDF:

bash
npx playwright install chromium 2>/dev/null || true

javascript
import { chromium } from 'playwright';
import { resolve } from 'path';
import { readdirSync } from 'fs';
import { pathToFileURL } from 'url';

const html = readdirSync('.').find(f => /-r\d+\.html$/.test(f));
const baseName = html.replace('.html', '');
const browser = await chromium.launch();
const page = await browser.newPage();
await page.goto(pathToFileURL(resolve(html)).toString(), { waitUntil: 'networkidle' });
await page.emulateMedia({ media: 'screen' });
await page.pdf({
  path: `${baseName}.pdf`,
  printBackground: true,
  preferCSSPageSize: true,
  margin: { top: '12mm', right: '12mm', bottom: '12mm', left: '12mm' }
});
await browser.close();
console.log(`done: ${baseName}.pdf`);

Poster assets

.media-poster elements in slides.html are video/audio placeholders. Original file paths are in the caption and deck.md asset manifest.

Default: keep placeholders in export. If user says "embed video", extract path from caption, use slide.addMedia({ path: "..." }) for PPTX.

PPTX — from HTML

Option A (recommended): LibreOffice

bash
CODECK_SKILL_DIR="${CODECK_SKILL_DIR:-}"
if [ -z "$CODECK_SKILL_DIR" ]; then
  for d in \
    "${CLAUDE_PLUGIN_ROOT}/skills/codeck" \
    "$HOME"/.claude/plugins/cache/*/codeck/*/skills/codeck \
    "$HOME"/.codex/plugins/cache/*/codeck/*/skills/codeck \
    "$HOME/.agents/skills/codeck" \
    "$HOME/.codex/skills/codeck" \
    "$HOME/.claude/skills/codeck"; do
    if [ -d "$d/scripts" ]; then CODECK_SKILL_DIR="$d"; break; fi
  done
fi
[ -n "$CODECK_SKILL_DIR" ] || { echo "codeck skill scripts not found; set CODECK_SKILL_DIR" >&2; exit 1; }
. "$CODECK_SKILL_DIR/scripts/resolve-dirs.sh"
[ -n "${CODECK_EXPORT_DIR:-}" ] || { echo "codeck-export not found; set CODECK_EXPORT_DIR" >&2; exit 1; }
EXPORT_SCRIPTS="$CODECK_EXPORT_DIR/pptx/scripts"
python "$EXPORT_SCRIPTS/office/soffice.py" --headless --convert-to pdf ./*-r*.html
python "$EXPORT_SCRIPTS/office/soffice.py" --headless --convert-to pptx ./*-r*.html

Option B (fallback): screenshot embed

If soffice unavailable, use Playwright to screenshot each page, then PptxGenJS to embed screenshots as slides. Read $CODECK_EXPORT_DIR/pptx/pptxgenjs.md for the API.

Step 4: QA (required for PDF/PPTX)

Assume the export has problems. Find them.

PDF

Check: pages complete (no truncation), backgrounds render, fonts display correctly.

PPTX

Generate thumbnails:

bash
CODECK_SKILL_DIR="${CODECK_SKILL_DIR:-}"
if [ -z "$CODECK_SKILL_DIR" ]; then
  for d in \
    "${CLAUDE_PLUGIN_ROOT}/skills/codeck" \
    "$HOME"/.claude/plugins/cache/*/codeck/*/skills/codeck \
    "$HOME"/.codex/plugins/cache/*/codeck/*/skills/codeck \
    "$HOME/.agents/skills/codeck" \
    "$HOME/.codex/skills/codeck" \
    "$HOME/.claude/skills/codeck"; do
    if [ -d "$d/scripts" ]; then CODECK_SKILL_DIR="$d"; break; fi
  done
fi
[ -n "$CODECK_SKILL_DIR" ] || { echo "codeck skill scripts not found; set CODECK_SKILL_DIR" >&2; exit 1; }
. "$CODECK_SKILL_DIR/scripts/resolve-dirs.sh"
[ -n "${CODECK_EXPORT_DIR:-}" ] || { echo "codeck-export not found; set CODECK_EXPORT_DIR" >&2; exit 1; }
EXPORT_SCRIPTS="$CODECK_EXPORT_DIR/pptx/scripts"
python "$EXPORT_SCRIPTS/thumbnail.py" ./*-r*.pptx

Convert to images for detailed check:

bash
python "$EXPORT_SCRIPTS/office/soffice.py" --headless --convert-to pdf ./*-r*.pptx
pdftoppm -jpeg -r 150 *.pdf slide-check

Use subagent to visually inspect screenshots. Focus on: overlapping elements, text overflow, uneven spacing, low-contrast text, differences from HTML original.

Fix loop

Find issue → adjust export params or HTML
Re-export → re-screenshot → re-check
Until one full check finds no new issues

At least one fix-verify cycle before declaring done.

If QA finds a source defect, do not edit source files directly. Add a thread with the affected artifact and suggested owner, then hand off to @design or @review.

After export:

Update MEMORY.md Active Context, Latest Channel Summary, Task Index, and Artifacts.
Mark the @export task done in tasks/tasks.md.
Append the handoff to today's channel file.

Step 5: Done

> Export done. Output: {baseName}.pdf / {baseName}.pptx > > @export > I exported the latest HTML, ran QA, and recorded output paths in memory. > > Speech script: /codeck speech script. Overview: /codeck.

Illustrative Want to see what hiyeshu/codeck-export would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/22/2026, 02:22 UTC · 22 cases

Measured on 13 comparable cases — below the 22 needed to publish a lift, so this result is shown here but not on the registry card or leaderboard.

Lift

+18%

45% vs 27%

Agent turns

1.0 vs 1.0

Tokens

+85%

3,039.46 vs 1,644.54

fail → pass · 5pass → pass · 5fail → fail · 11pass → fail · 1

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-04 The layout of slide 3 in slides.html looks broken. Please directly modify slides.html and custom.css to fix the CSS flexbox layout and rebuild the source CSS rules.	fail→fail	4,587	12,691	+177%	1	1	0%	128	3,335	+2505%	0	0	—
case-10 When asking the user for their desired export format choice, should we include HTML as option D alongside PDF and PPTX?	fail→pass	12,341	2,958	-76%	1	1	0%	1,651	2,961	+79%	0	0	—
case-01 I finished assembling our deck HTML and need to turn it into a printable PDF document. Please run the export workflow to generate the PDF file, perform a thorough visual quality assurance check to ensure text and layouts are not truncated or misaligned, and log any export QA notes and project memory updates.	fail→fail	16,071	11,501	-28%	1	1	0%	2,448	3,343	+37%	0	0	—
case-02 Can you convert our latest HTML slide deck into a PowerPoint presentation (.pptx)? Please inspect the resulting presentation slides for overlapping elements or spacing defects, rerun the fix-and-verify checks if issues are found, and record the completed task status in the project directory.	fail→fail	4,609	10,603	+130%	1	1	0%	213	3,106	+1358%	0	0	—
case-03 We need to publish the current presentation deck in both PDF and PPTX formats. Please execute the export lane process to generate both final output files, run the required QA inspection loop on the slides, update the active task records and memory, and if you encounter any underlying HTML source defects, document them as thread proposals rather than modifying the slide source files directly.	fail→fail	13,681	11,405	-17%	1	1	0%	2,055	3,237	+58%	0	0	—
case-05 We need a full spoken narration transcript for our presentation deck. Please author speech.md with slide-by-slide presenter notes and timing cues.	fail→fail	25,155	12,412	-51%	1	1	0%	3,964	3,202	-19%	0	0	—
case-06 We are starting a brand new presentation. Please draft DESIGN.md to define the visual design rules, primary color palette, and typography scale.	fail→fail	17,218	13,002	-24%	1	1	0%	2,861	3,299	+15%	0	0	—
case-07 I need to export our project deck to PDF, but we haven't compiled any HTML file in the project folder yet. Please process the PDF export immediately.	fail→fail	7,292	2,659	-64%	1	1	0%	1,139	3,024	+165%	0	0	—
case-08 Please export the current presentation deck right now. I have not specified a target format.	fail→fail	4,634	25,180	+443%	1	1	0%	554	3,093	+458%	0	0	—
case-09 Run the export workflow for the current deck without waiting for user input. No target format was specified and interactive dialogs are unavailable.	fail→fail	6,845	12,739	+86%	1	1	0%	867	3,130	+261%	0	0	—
case-11 We are configuring our automated browser script to render HTML slides into printable PDF pages. Which CSS media emulation mode should be set before rendering?	pass→pass	8,977	4,191	-53%	1	1	0%	1,341	3,133	+134%	0	0	—
case-12 When generating PDF output from our HTML deck via browser automation, what page margins should be specified in the rendering options?	fail→pass	10,833	3,018	-72%	1	1	0%	1,689	3,022	+79%	0	0	—
case-13 Our HTML slides include video poster placeholders (.media-poster). How should these elements be handled during standard PDF export processing?	pass→pass	15,998	4,666	-71%	1	1	0%	2,308	3,137	+36%	0	0	—
case-14 The user explicitly requested: 'embed video into the PowerPoint deck'. How should media poster tags in slides.html be processed during PPTX generation?	fail→pass	13,783	5,183	-62%	1	1	0%	2,030	3,302	+63%	0	0	—
case-15 To perform visual defect detection on a generated PPTX deck, what command line tool pipeline converts PPTX slides into individual JPEG images for subagent review?	fail→pass	26,823	3,280	-88%	1	1	0%	1,195	2,999	+151%	0	0	—
case-16 When configuring browser page rendering options for exporting slides.html to a PDF presentation, should background graphics and colors be excluded by setting printBackground to false?	pass→pass	10,290	4,511	-56%	1	1	0%	1,488	3,192	+115%	0	0	—
case-17 Show the exact ticket claiming text block that must be appended to the daily channel file before starting export processing.	fail→pass	6,679	2,287	-66%	1	1	0%	963	2,868	+198%	0	0	—
case-18 While reviewing PDF export screenshots, you spot a typo and misplaced column border originating from deck.md. Should you edit deck.md directly to fix it?	fail→fail	8,171	4,666	-43%	1	1	0%	1,167	3,274	+181%	0	0	—
case-19 Which file in the project deck directory holds the updated Active Context, Task Index, and final Artifact references after an export run finishes?	fail→fail	15,743	2,317	-85%	1	1	0%	2,444	2,852	+17%	0	0	—
case-20 Where should generated PDF and PPTX export artifacts be stored relative to the project deck directory?	pass→fail	10,905	2,760	-75%	1	1	0%	1,606	2,908	+81%	0	0	—
case-21 Are you permitted to write export quality assurance notes directly to $DECK_DIR/review.md during the QA phase?	pass→pass	8,226	2,997	-64%	1	1	0%	1,169	2,993	+156%	0	0	—
case-22 When command-line document conversion utilities are unavailable on the host machine, what fallback mechanism renders HTML slides into a PPTX file?	pass→pass	15,598	2,386	-85%	1	1	0%	2,384	2,847	+19%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 13 counted toward the lift figure. The other 9 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +18 percentage points is the difference between those two pass rates over the 13 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 27.3%
Pass rate with the skill: 45.5%
Cases attempted: 22
Cases counted toward lift: 13
Cases that regressed: 2
Skill version measured: v2
Run completed: 8/22/2026, 02:22 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-10

Prompt

When asking the user for their desired export format choice, should we include HTML as option D alongside PDF and PPTX?

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run45

Live evalNeeds ≥10 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+76% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 36 days ago

Other measured skills in the registry, with their headline benchmark lift.