Skip to content
Agentmart
Submit a tool

absorb

Use when you need a capability an open-source project already provides — adopt it and re-engineer it as a native skill instead of inventing a duplicate.

Skill by Tcuzzo

IndexedMIT★ 2
Write a review

Chalk Labs

Need help deploying absorb?

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.

Book a call

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/tcuzzo-absorb

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.

We use your email only for this newsletter. Privacy

About

Use when you need a capability an open-source project already provides — adopt it and re-engineer it as a native skill instead of inventing a duplicate. Trigger words: absorb, adopt, port, re-engineer, ingest a repo, prior art, capability port, make this native.

No 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

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/absorb

What's inside

Skill name
absorb
Description
Use when you need a capability an open-source project already provides — adopt it and re-engineer it as a native skill instead of inventing a duplicate. Trigger words: absorb, adopt, port, re-engineer, ingest a repo, prior art, capability port, make this native.
License
MIT
Instructions
70 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Absorb — Adopt Prior Art, Don't Reinvent It
**Effort:** light — a repo ingest (metadata probe + shallow clone) and a cross-family grade of the port. Removes: reinventing a capability prior art already solved — the from-scratch duplicate you'd owe every bug for.

**Capability is king.** A repo is a vehicle for a capability. When you need something
an existing project already does, do not build a duplicate from scratch, and do not
clone-and-paste. Find the best prior art, extract the capability, re-engineer it to
fit your harness, and cite the scaffold. The citation is a fact, not a decoration.

## When to run

- You are asked to add a capability (a tool, skill, agent, pipeline) that open source
  likely already solves.
- You are about to `git clone` and copy code verbatim — stop; this is the path instead.
- Skip it for a single snippet, a config value, or a fact lookup. Just read those.

## Steps

1. **Hunt prior art first.** Search before you build. A duplicate you invent is worse
   than a scaffold you adopt: you inherit zero field testing and you owe all the bugs.
2. **Ingest beyond the README.** Pull the project metadata (license, activity,
   language) from the platform API. Shallow-clone into a scratch directory. Read the
   code and the tests. The README is marketing; the code is the truth.
3. **Run the trust gates.**
   - *License:* permissive (MIT / Apache / BSD / MPL) = safe to re-engineer.
     Copyleft (GPL / AGPL) = technique-only — re-engineer the idea, never copy the
     code. No license = treat as all-rights-reserved, technique-only.
     Non-commercial terms = a blocker; take it to your human.
   - *Shady scan:* grep for cloak / spam / fake-review / scam patterns. Flag loudly.
   - *No wild installs:* never `pip install` / `npm install` an unvetted dependency
     (typo-squatting is a real supply-chain attack). Re-engineer as thin code over
     your own primitives instead.
   - *Is the capability real?* Verify claims against independent evidence. A seller's
     blog is a claim, not evidence. Verdict: real / hype / scam / unverifiable.
   - *Bounded egress:* anything the adopted version fetches must be throttled,
     cached, and killable.
4. **Deconstruct into a capability map.** For every ability the project provides,
   record: what it does, how, its load-bearing seams, its bloat or risk, what you can
   reuse from your own stack, and whether it lands native or behind a thin adapter.
   Every capability is **preserved or refuted with evidence**. A silently dropped
   capability is a defect.
5. **Write the re-engineering spec.** The seams to build, the bloat you are dropping
   (recorded loudly, never silently), and one failing contract test per capability
   that asserts a real side-effect — a file, a database row, real output. Mock only a
   paid external API's transport, never the logic.
6. **Reconstruct red-first.** Commit the failing tests, then build until green across
   the whole seam. A model from a different family than the builder grades the result
   — the builder never grades its own work.
7. **Cite and record.** Write the scaffold credit where the capability now lives:
   author, project, license, what is borrowed (the scaffold) and what is yours (the
   re-engineering). Never invent a credit. Never strip one.

## Hard rules — any one of these fails the skill

- Copying code verbatim instead of re-engineering the capability.
- Building a duplicate without ever searching for prior art.
- Trusting the README or a marketing page over the code.
- Installing a wild dependency instead of re-engineering the technique.
- Copying copyleft or unlicensed code (technique-only, always).
- Dropping a capability without a written refutation.
- Mock theater in a capability test — the test must touch a real side-effect.
- Shipping without the scaffold citation.

## Works well with

- [red-first](../red-first/SKILL.md) — the contract tests that guard each capability.
- [sniper-testing](../sniper-testing/SKILL.md) — real side-effects, no mock theater.
- [blind-tribunal](../blind-tribunal/SKILL.md) — cross-family grading of the port.
- [decision-bar](../decision-bar/SKILL.md) — license blockers and taste calls go to your human; everything else executes.

From Tcuzzo/HydraAgent_public, MIT.

More from Tcuzzo/HydraAgent_public

Signals

Stars
2
Forks
0
Open issues
0
Last commit (this folder)
Oct 6, 2026 (25 days ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from Tcuzzo

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 absorb 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.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying absorb?

CallGet help
Cite this page

absorb skill. Agent Mart. https://agentmart.dev/skills/tcuzzo-absorb (last updated 2026-10-24)