baseline SKILL.md skillA

baseline from skillberry-ai/cap-evolve is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.

Install it, pinned
mdr add skillberry-ai/cap-evolve/baseline@git:20260928.31fe1ff

Pins git:20260928.31fe1ff in mdr.lock. Pin the fingerprint instead.

Audit grade
A, 17 of 17 checks passed
Kind
SKILL.md skill
Repository
skillberry-ai/cap-evolve
Path
skills/phases/baseline/SKILL.md
Versions recorded
5
Latest version
git:20260928.31fe1ff, committed Sep 28, 2026
First version recorded
Jun 16, 2026
Size
4,508 bytes
Repository stars
66
License
Apache-2.0

Establish the starting point. Use after implement-and-check and before any algorithm. Creates the run directory, freezes the seeded train/val/test split (written once), scores the unmodified seed capability on val, and records it as the candidate every algorithm must beat. Reports the remaining headroom so a saturated seed stops the run before it spends budget.

Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.

What the file says

# baseline — freeze splits, score the seed

baseline is the first phase that touches data, so it owns the run's one
irreversible decision: **the split**. It writes `splits.json` once (seeded), scores
the *unmodified* seed capability on val, and records that score as the bar every
algorithm must beat.

Run `implement-and-check` first. baseline re-runs that check itself and exits
non-zero before creating a run dir if it is red — a split frozen against a broken
adapter poisons every number measured afterwards.

## Why it matters
- **Fair comparison point.** Every algorithm hill-climbs *against* the baseline
  val score; a candidate that does not beat it is not progress.
- **Headroom.** The printed JSON carries `headroom` (`1 - val`) and
  `headroom_verdict`: `saturated` means the seed is already at the ceiling and
  further iterations buy noise — stop; `floor` (val at 0) usually means a
  mis-wired adapter rather than a hard task — re-check before spending budget;
  `ok` means proceed. The same verdict is logged as a `headroom` event so the
  orchestrator can stop on it with no human reading the number.

## Splitting choices
…

Read the whole file at its exact version.

How to install

The exact contents, by fingerprint
mdr add skillberry-ai/cap-evolve/baseline@sha256:9851c88a732c0ee9

The line at the top of this page pins the label git:20260928.31fe1ff, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.

Badge

mdr badge

[![mdr](https://markdownregistry.com/badge/art_u6itnhmgx6vgbcu2.svg)](https://markdownregistry.com/a/art_u6itnhmgx6vgbcu2)

2 badge image fetches in 30 days. What the badge shows.

Versions

The registry has recorded 5 versions of this file, between Jun 16, 2026 and Sep 28, 2026, about one every 26 days.

versioncommittedcommitsizeaudit
git:20260928.31fe1ff latest2026-09-28 31fe1ff 4,508 BA view · diff
git:20260822.beab97e2026-08-22 beab97e 4,288 BA view · diff
git:20260619.818894d2026-06-19 818894d 4,536 BA view · diff
git:20260617.92b5c652026-06-17 92b5c65 3,832 BA view · diff
git:20260616.306ff982026-06-16 306ff98 3,507 BA view

Audit of the latest version

A  17 of 17 checks passed. Fixed automatic checks with no AI involved, so the same checks give the same answer every run.
  • pass: Frontmatter block present
  • pass: Frontmatter declares a name
  • pass: Frontmatter declares a description
  • pass: Size between 200 bytes and 200 KB (4508 bytes)
  • pass: No zero-width or bidi control characters
  • pass: No instruction hidden inside an HTML comment
  • pass: No link to an exfiltration or paste host
  • pass: No credential-shaped string
  • pass: No instruction to send local credentials anywhere
  • pass: No text hidden with inline styles
  • pass: No prompt-injection phrasing
  • pass: No curl or wget piped into a shell
  • pass: No recursive delete of root, home or parent
  • pass: No instruction to read or print local credentials
  • pass: No base64 blob over 200 characters
  • pass: No link to a raw IP address
  • pass: No script tag

Source

GitHub

skillberry-ai/cap-evolve · 66 stars · license Apache-2.0 · pushed 2026-10-21 · branch main

API

GET https://markdownregistry.com/api/v1/artifacts/art_u6itnhmgx6vgbcu2
GET https://markdownregistry.com/api/v1/resolve?ref=skillberry-ai/cap-evolve/baseline
GET https://markdownregistry.com/api/v1/blob/9851c88a732c0ee99858918a05a54d654b8f3db8a394f78c363d500de0687f93

Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.

Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.

More from skillberry-ai/cap-evolve

CLAUDE.md CLAUDE.md file
skillberry-ai/cap-evolve · CLAUDE.md
git:20261005.6af27e2 · audit A · 66 stars
my_skill SKILL.md skill
skillberry-ai/cap-evolve · examples/tau2_custom/blackbox/seed_capability/my_skill/SKILL.md
git:20260924.bafd81b · audit C · 66 stars
my_skill SKILL.md skill
skillberry-ai/cap-evolve · examples/tau2_sierra_airline/blackbox/seed_capability/my_skill/SKILL.md
git:20261013.01104a5 · audit C · 66 stars
arithmetic-answers SKILL.md skill
skillberry-ai/cap-evolve · examples/toy_skill/seed_capability/SKILL.md · Answers a short arithmetic question with just the number. Use when the user asks to compute a sum, difference, or…
git:20260822.f0fb098 · audit A · 66 stars
llms.txt llms.txt file
skillberry-ai/cap-evolve · llms.txt
git:20260924.bafd81b · audit A · 66 stars
agent-optimize SKILL.md skill
skillberry-ai/cap-evolve · skills/algorithms/agent-optimize/SKILL.md · Free-form optimization algorithm for agent orchestration mode: the conversational agent owns the whole search — reading…
git:20261013.db330d5 · audit A · 66 stars
evograph SKILL.md skill
skillberry-ai/cap-evolve · skills/algorithms/evograph/SKILL.md · Deprecated agent-mode algorithm (evo-graph port): a weakness-graph search that dispatched one solver agent per failure…
git:20260904.802badb · audit A · 66 stars
gepa SKILL.md skill
skillberry-ai/cap-evolve · skills/algorithms/gepa/SKILL.md · Runs the GEPA optimization loop (arXiv:2507.19457) — sample-efficient reflective Pareto search. A cheap train-minibatch…
git:20260822.02a1a1d · audit A · 66 stars
hill-climb SKILL.md skill
skillberry-ai/cap-evolve · skills/algorithms/hill-climb/SKILL.md · Runs a global hill-climb optimization loop where the parent is always the current best candidate and the val…
git:20261012.bf14385 · audit A · 66 stars
skillopt SKILL.md skill
skillberry-ai/cap-evolve · skills/algorithms/skillopt/SKILL.md · Runs the SkillOpt single-lineage optimization loop, which organizes a hill-climb into epochs over mini-batches of train…
git:20260822.02a1a1d · audit A · 66 stars
mcp-tool SKILL.md skill
skillberry-ai/cap-evolve · skills/capabilities/mcp-tool/SKILL.md · Optimize the tool surface of an EXTERNAL MCP server — one the agent talks to but does not implement. Use when an agent…
git:20260822.10e53e1 · audit A · 66 stars
skill-package SKILL.md skill
skillberry-ai/cap-evolve · skills/capabilities/skill-package/SKILL.md · Optimize an Agent Skill package itself — its SKILL.md (frontmatter + body), its references, and its bundled scripts…
git:20261008.03e3727 · audit A · 66 stars

Every file in skillberry-ai/cap-evolve

Browse by kind, by grade A, or by owner.