Discover
Instruments a repo for evo's autonomous optimization: benchmark, baseline, first run
Test report
- Verdict
- Works with setup
- Score
- Tested
- Jul 20, 2026
- Environment
- Claude Code 2.x (agent harness)
- Upstream re-checked
- Aug 10, 2026 · f26276a
Fetched the full repo tree and the 45KB SKILL.md plus scripts/validate_result.py and inline_instrumentation.py raw. All body-referenced files (constructing-benchmark.md, sdk_python.py/node.js, inline_instrumentation.py/.js, proposing-dimensions.md, sizing-the-round.md, instrumentation-contract.md, validate_result.py) exist in the tree; validate_result.py is a benign local JSON checker and no exfiltration/base64/hidden-payload patterns appeared. The skill is a detailed, honest orchestration procedure but every step drives the external `evo` CLI (step 0 hard-checks `evo --version` == 0.8.0 and directs the user to `uv tool install evo-hq-cli==0.8.0`), so copying files alone leaves it inert (-1 install). Trigger phrasings tested: SHOULD -> "/evo:discover", "set up evo on this repo to start optimizing my agent", "instrument this codebase so evo can autonomously optimize it"; SHOULD NOT -> "help me discover new music playlists", "run the optimization loop" (that's the sibling optimize skill) — all 5 judged correctly. Telemetry is opt-out with explicit privacy redaction rules, and an "[EVO DIRECTIVE]" banner section grants user-authority to the plugin's own runtime channel — transparent, not injection.
Scored on four weighted criteria — install, triggering, output vs. baseline, docs. How scoring works
- Installs cleanly 4/5
- Triggers reliably 5/5
- Output vs. baseline 5/10
- Docs & honesty 4/5
What Discover does
The setup skill for the evo plugin: it explores a codebase, proposes optimization dimensions, constructs an instrumented benchmark inside a baseline worktree, and runs the first experiment. Triggers on /evo:discover, "set up evo", or "instrument this repo for autonomous optimization". Requires the external evo-hq-cli, so it cannot run as a pure prompt.
How to install Discover
Add plugin: /plugin marketplace add evo-hq/evo then /plugin install evo. Then install the runtime CLI: uv tool install evo-hq-cli==0.8.0 (or pipx install evo-hq-cli==0.8.0).
Skills live in ~/.claude/skills/ (global) or .claude/skills/
(per-project). Restart Claude Code after installing.
Commands — how to trigger Discover
-
/discoverInstruments a repo for evo's autonomous optimization: benchmark, baseline, first run
It also activates on plain-language prompts like these:
-
Set up evo to optimize this repository -
Run evo:discover to set up this codebase -
Start a new autonomous optimization run
Frequently asked questions
- Is the Discover skill free?
- Yes. The skill itself is free from evo-hq/evo. SkillProof publishes the install command and an independent test verdict at no cost.
- Does Discover work with Claude Code?
- We tested it with Claude Code 2.x (agent harness) on Jul 20, 2026. Verdict: Works with setup. Fetched the full repo tree and the 45KB SKILL.md plus scripts/validate_result.py and inline_instrumentation.py raw. All body-referenced files (constructing-benchmark.md, sdk_python.py/node.js, inline_instrumentation.py/.js, proposing-dimensions.md, sizing-the-round.md, instrumentation-contract.md, validate_result.py) exist in the tree; validate_result.py is a benign local JSON checker and no exfiltration/base64/hidden-payload patterns appeared. The skill is a detailed, honest orchestration procedure but every step drives the external `evo` CLI (step 0 hard-checks `evo --version` == 0.8.0 and directs the user to `uv tool install evo-hq-cli==0.8.0`), so copying files alone leaves it inert (-1 install). Trigger phrasings tested: SHOULD -> "/evo:discover", "set up evo on this repo to start optimizing my agent", "instrument this codebase so evo can autonomously optimize it"; SHOULD NOT -> "help me discover new music playlists", "run the optimization loop" (that's the sibling optimize skill) — all 5 judged correctly. Telemetry is opt-out with explicit privacy redaction rules, and an "[EVO DIRECTIVE]" banner section grants user-authority to the plugin's own runtime channel — transparent, not injection.
- What is the Discover SkillProof Score?
- 7.2/10 — installs cleanly 4/5, triggers reliably 5/5, output vs. baseline 5/10, docs & honesty 4/5.
- How do I install Discover?
- Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.
- Can I use Discover with Cursor, Copilot, Gemini CLI, Codex or other AI tools?
- The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.