regenerate-literate-ai-manager-overview
Regenerate and refresh the Literate-AI manager and engineering overview document pair (presentation and narrative) when the framework, CLI, qualification…
Skill by jordanhubbard
Chalk Labs
Need help deploying regenerate-literate-ai-manager-overview?
Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/jordanhubbard-regenerate-literate-ai-manager-overview
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: documents, presentations
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
- Runs as
- Python package
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/regenerate-literate-ai-manager-overview
What's inside
- Skill name
regenerate-literate-ai-manager-overview- Description
- Regenerate and refresh the Literate-AI manager and engineering overview document pair (presentation and narrative) when the framework, CLI, qualification model, or portfolio evidence changes.
- License
- Apache-2.0
- Instructions
- 154 lines in SKILL.md · view on GitHub
Before installing
- Ships executable scripts
Agent Skills spec issues
- name: "regenerate-literate-ai-manager-overview" does not match folder "literate-ai-manager-overview"
- files: referenced but missing: scripts/verify_document_pair.py
Files (30)
- README.md
- SKILL.md
- application-foundry.png · assets/
- cover-hero.png · assets/
- engineering-section.png · assets/
- evidence-artifact.png · assets/
- gate-failure.png · assets/
- generation-key.png · assets/
- isolation-containment.png · assets/
- legacy-transform.png · assets/
- lifecycle-gates.png · assets/
- living-portfolio.png · assets/
- manager-section.png · assets/
- one-many.png · assets/
- build_deck.mjs
- build_deck.py
- build_narrative.py
- current-deliverables.md
- deck-specification.md
- literate-ai-manager-and-engineering-overview.docx
- literate-ai-manager-and-engineering-overview.pptx
- narrative-specification.md
- deck-authoring-prompt.md · prompts/
- image-prompts.md · prompts/
- publish_google_workspace.py
- qa-ledger.md
- regenerate.sh
- regenerate_python.sh
- render_slides.py
- source-notes.md
SKILL.md (first lines)
# Literate-AI manager overview regeneration
This directory is the durable authoring package for the `literate-ai.document-pair`
realized by `component://literate-ai/literate-ai-overview`: the generated presentation at
`docs/presentations/literate-ai-manager-overview/literate-ai-manager-and-engineering-overview.pptx`
and its native Google Slides import (the `presentation` member), and a comprehensive
Google Doc narrative built from `narrative-specification.md` (the `narrative` member).
Before changing or regenerating the deck:
1. Read the repository-root `AGENTS.md` and `SKILL.md` completely.
2. Use CodeGraph first when `.codegraph/` exists and code behavior must be located or verified.
3. Use `build_deck.py` (`python-pptx`) to construct/render the PPTX. This is the primary
build path; it has no LLM calls and no Codex plugin dependency. Bootstrap the pinned
packages with `make doc-toolchain-bootstrap` into ignored `OBJ_DIR` (see
`tools/doc-toolchain/authoring-toolchain.json`). `regenerate.sh` prefers that portable
path and refuses to pip-install without Makefile authorization. Codex
`build_deck.mjs` / `@oai/artifact-tool` discovery remains a fallback only — see
`source-notes.md`'s "Codex-independent rebuild" entry. Do not remove plugin discovery
until three-OS visual parity closes.
4. Use the installed `imagegen` skill only when an existing image asset must be replaced. Preserve the current palette, composition intent, and negative-space requirements in `prompts/image-prompts.md`.
5. Treat `deck-specification.md` as the presentation’s narrative authority and `source-notes.md` as its factual-claim ledger.
## When to regenerate
Full regeneration of both members belongs on every **major or minor** distribution cut
(`0.8.0`, `0.9.0`, `1.0.0`, `1.1.0`, …). **Patch** cuts keep the authoring-package and
repository-root `README.md` files pointing at the last major/minor *edition* of the
published Slides and Google Doc. Do not retag a published patch or minor to carry a later
refresh. A skipped minor is recovered on the next published cut of that line (this
repository used `0.7.1` to catch up `0.7.0`). See
`docs/architecture/project-releases.md` and `skills/agent/release-project/SKILL.md`.
Release-policy source changes must update the deck and narrative specifications, factual
ledger, both builders, authoring prompt, and QA ledger together. Preserve the 26-slide
sequence by revising the existing release slide. Unless the request explicitly authorizes
artifact regeneration and Google publication, stop at source changes and record both as
pending; do not rewrite prior QA entries.
## Refresh procedure
1. Revalidate every implementation claim in `source-notes.md` against current specifications and current code. Do not promote roadmap language into implemented claims.
2. Update `deck-specification.md` first. The presentation source is derived from that specification.
3. If a visual must change, regenerate only that asset using its recorded prompt and replace the corresponding file in `assets/`. Keep visuals text-free.
4. Update [build_deck.py](build_deck.py)'s corresponding slide function(s) to match. It ports
`build_deck.mjs`'s primitive shape vocabulary (`shape`, `text`, `pill`, `line`,
`arrow`, `footer`, `title`, `wash`, cover-fit image placement) 1:1; add native-shape
diagrams rather than raster images for anything that shows a real mechanism.
5. Run [regenerate.sh](regenerate.sh). It prefers the pinned OBJ_DIR toolchain from
[regenerate_python.sh](regenerate_python.sh) and
`make doc-toolchain-bootstrap` (it will not pip-install on its own), builds the PPTX
and the narrative `.docx`, rasterizes slides for inspection, then runs
`scripts/verify_document_pair.py` against both members and writes its report to
`$OBJ_DIR/literate-ai-manager-overview/acceptance.json`. Fix any failing scenario
before continuing — it checks slide-surface geometry, non-overlapping text-bearing
frames, non-empty speaker notes on every page, narrative heading levels, no credential
material, and no unresolved placeholder tokens.
6. Render every slide, inspect a contact sheet and full-size slides, and fix all
unintended clipping, overflow, overlap, broken connectors, and unreadable type.
Geometry-escape (elements inside 1280 × 720) is not sufficient: two-line titles
that paint into subtitles, section tags that collide with the brand pill, and
bottom captions that collide with the footer all fail this pass. After the
native Google Slides update, read the published copy as well — font substitution
can introduce overlaps the local PPTX did not show. Workflow slides must show
the flow as a picture or diagram; a paragraph that restates the workflow is the
defect, not a caption. Native-shape diagrams still count as pictures; do not
replace the specification excerpt or control-model diagrams with decorative
raster that hides the mechanism.
7. Run the presentation skill’s overflow test when its managed Python environment is
available and the Codex `presentations` plugin path is in use. For the `python-pptx`
path, `scripts/verify_document_pair.py` is the mechanical geometry gate: it fails
elements outside the declared surface **and** overlapping text-bearing frames.
Geometry-escape alone is not an overflow test. `render_slides.py` also prints an AABB
overlap report for inspection. Record any fallback in the QA ledger.
8. Run `publish_google_workspace.py --preflight-only --release VERSION
--expected-account ACCOUNT` before any provider mutation. It must confirm the explicitly
selected active `gcloud` account and both stable resource IDs, MIME types, edit/export/
sharing capabilities, and access mapping. If no account is active, tell the operator to
run `gcloud auth login ACCOUNT` and `gcloud config set account ACCOUNT`; do not choose an
account for them. Only after the complete pair passes may an authorized publication
update the exact Slides resource in place and export it for read-back verification.
Never persist a token or include it in logs. Record the resulting Slides URL in
`README.md`, near the top, alongside the narrative's URL.
## Narrative member refresh procedure
1. Treat `narrative-specification.md` as the narrative's structural authority (heading
hierarchy, section scope) and `source-notes.md` as its factual-claim ledger — the same
ledger the presentation member cites. Revalidate every claim before drafting or
redrafting.
2. Use `build_narrative.py` (`python-docx`) to construct the local `.docx`. This is the
primary narrative path; it has no LLM calls and no Codex plugin dependency. The Codex
`documents` plugin remains an optional environment where that plugin happens to be
installed and funded — it is not a hidden requirement of this package.
3. `./regenerate.sh` builds both members through the portable python-pptx path, writes
the capability manifest covering both, runs `scripts/verify_document_pair.py`, and
rasterizes slides with [render_slides.py](render_slides.py) into
`$OBJ_DIR/literate-ai-manager-overview/rendered-slides/` for visual inspection. Fix any
failing oracle scenario before continuing. Headings must map to Word `Heading 1`
through `Heading 6` without skips.
4. Inspect the contact sheet and full-size slides. For the narrative, confirm the heading
tree matches `narrative-specification.md` and that no placeholder or credential
material ships.
5. Publish the verified PPTX and `.docx` only through the pair-wide preflight above and
only with explicit external-write authorization. Update both exact stable resources in
place; do not create a new Doc merely because the configured one is inaccessible. The
publisher writes a release-bound, resumable receipt after each update and exports both
native resources for structural and hash verification. On partial failure, retain the
receipt, repeat the complete preflight, and resume only the missing member. Record both
stable URLs in `README.md`, near the top, as the current major/minor *edition*. Drive
`files.export` rejects this deck (`exportSizeLimitExceeded`); read-back uses the authenticated
`https://docs.google.com/presentation/d/{id}/export/pptx` URL.
[`publish_google_workspace.py`](publish_google_workspace.py) in this package performs
the authorized update, access mapping, resumable receipt, and read-back.
6. Both members share one publication-authorization gate (see
`components/literate-ai-overview/component.md`'s "Requirement: Publication is
authorized per revision"): regenerating one member does not authorize publishing
either. Asking for this overview to be regenerated and published for a named edition
is authorization for that revision only.From jordanhubbard/literate-ai, Apache-2.0.
Signals
- Stars
- 28
- Forks
- 10
- Open issues
- 26
- Last commit (this folder)
- Oct 12, 2026 (1 month ago)
- Last release
- Oct 12, 2026 (1 month ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- Apache-2.0
- Repository
- github.com/jordanhubbard/literate-ai
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
Done for you
Rather not set it up yourself?
Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.
See how we build AI agents →Read the case studies
Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.
- launches shipped
- 50+
- launches shipped
- media placements
- 500+
- media placements
- founders on every account
- 2
- founders on every account
Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block
Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.
Get regenerate-literate-ai-manager-overview running in your business
Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.
Cite this page
regenerate-literate-ai-manager-overview skill. Agent Mart. https://agentmart.dev/skills/jordanhubbard-regenerate-literate-ai-manager-overview (last updated 2026-10-24)