discover
Initialize evo for the current repository by exploring the codebase, proposing unexplored optimization dimensions, constructing the benchmark inside a baseline worktree, and running the first experiment. Use when the user invokes /evo:discover, mentions setting up evo, wants to instrument a codebase for autonomous optimization, or asks to start a new evo run on a project.
npx skills add evo-hq/evo --skill discover --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
Instructs the agent to start a new evo workspace for the current repository, instrument the codebase for autonomous optimization, propose unexplored optimization dimensions, assemble a benchmark inside a baseline worktree, and execute the first experiment.
How it works
- Provides an explicit flow for starting evo in a project: initialize workspace, explore the repo to understand optimization targets and metrics, identify a benchmark (or propose dimensions if ambiguous), and prepare for the first experiment.
- References a structured command set and references to sub-skills and references that guide instrumentation, benchmarking, and experiment construction, including guidance on how to wire benchmarks (SDK vs inline instrumentation) and how to handle instrumentation mode via evo init.
- Specifies that the runtime will dispatch subagents and sub-skills (e.g., evo:benchmark-reviewer, evo:ideator, evo:verifier) as part of the workflow, and emphasizes not committing evo artifacts to main and keeping all evo work inside a baseline worktree.
- Includes a verification step to ensure the evo CLI version matches the skill’s required version and describes fallback actions if mismatches occur.
When to use it
Use when the user invokes /evo:discover, mentions setting up evo, wants to instrument a codebase for autonomous optimization, or asks to start a new evo run on a project.
What it can touch
- The skill outlines handling of repositories and worktrees and describes the instrumentation decision (SDK vs inline) to be passed to evo init via --instrumentation-mode <sdk|inline>.
- It references commands and files in the evo workflow (evo init, evo new, instrumentation contracts, and inline instrumentation helpers), but does not itself expose or modify specific files here.
Caveats
- Requires evo-hq-cli 0.8.0 to be installed exactly as shown; potential drift causes instructions to stop until alignment is restored.
- Baseline worktree concept means changes are isolated from main; main remains byte-identical to the pre-evo state.
- Infra setup is not user-invocable; if a remote backend is needed, follow provider-matrix guidance rather than invoking directly from this skill.
# Discover Internal procedure for `evo:discover`. The user only sees the user-facing prompts, the dashboard URL, and the baseline score -- everything else is the agent's choreography. ## Evo surface General guidance on the skills and tools available in evo. Each line is a triggering condition: if you're about to do X, pull/dispatch/read this. Don't preload -- act when the trigger fires. **Always have a sense of the skill before jumping into its references.** A skill body carries the decision-making; references are concrete contracts that assume a decision has been made. ``` evo plugin │ ├── Main thread (the orchestrator -- you, inside /evo:discover or /evo:optimize) │ │ │ ├── Skills (Skill tool) │ │ ├── evo:discover starting a new evo workspace / instrumenting a project │ │ ├── evo:optimize after discover commits the baseline -- drives the loop. │ │ │ Args: subagents=N (read sizing-the-round FIRST), │ │ │ autonomous, subagents-only, budget=N, stall=N │ │ ├── evo:ship after the loop stops -- distills the best valid │ │ │ experiment into a mergeable change (PR if remote, │ │ │ else merge) + a mergeability report │ │ ├── evo:finetuning task is finetuning / post-training / training a m
- Evo surface
- Host conventions
- Mid-run user directives (evo direct)
- 0. Verify the evo CLI is in sync with this skill
- Guiding principles
- 1. Explore the repo
- 2. Look for the obvious benchmark
- 3. Propose unexplored optimization dimensions (only if step 2 was ambiguous)
- 4. Ask the user to pick the benchmark
- 5. Ask the user for instrumentation mode
- 6. Prepare main (without committing to it)
- 6a. Detect (don't auto-commit) dirty or untracked dependencies
- 6b. Add local-only git excludes
- 7. Initialize the workspace
evo --version evo init --name "<short project name>" \ evo init \ evo config runtime set --prepare "uv sync" --before-run "make reset-test-state" --prefix "uv run" evo config runtime show evo env load .env --all evo env load .env --allow KEY1,KEY2 evo env show evo config set task-skills finetuning # any weight-update / training task Test-suite gate: pytest already exits non-zero on failures (use uv run --with if pytest isn't already a dep)
What does the discover skill do?
Initialize evo for the current repository by exploring the codebase, proposing unexplored optimization dimensions, constructing the benchmark inside a baseline worktree, and running the first experiment. Use when the user invokes /evo:discover, mentions setting up evo, wants to instrument a codebase for autonomous optimization, or asks to start a new evo run on a project.
How do I install it?
Run `npx skills add evo-hq/evo --skill discover --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From evo-hq/evo, a repository with 1,359 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.