docs(cybergym): simplify README into a script-driven walkthrough - #60
Merged
Conversation
Replace the long command-heavy README with a four-step flow backed by small scripts under scripts/: - config.sh: single source for paths, model, server, and 4h timeout - setup.sh: one-time env/install/data/image setup - start_server.sh: run the CyberGym submission server - run_subset.sh (renamed from run_10_tasks.sh): run the subset - validate.sh: post-validate submitted PoCs with a pass/fail summary Generate the CyberGym API key into the gitignored .env instead of hardcoding the upstream default (avoids secret-scanner flags). Default the per-task hard timeout to 4h so the agent's soft timeout can fire. Keep the direct runner invocation in the README to show how the agent is called. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
llm_config.yaml only names the env var (api_key_env); the key itself lives in .env. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Cleans up the CyberGym example README, which had grown into a long, command-heavy transcript that was hard to follow. Replaces it with a four-step, script-driven walkthrough backed by small scripts under
examples/cybergym/scripts/, reasoned through from a first-time user's perspective.What changed
scripts/config.sh— single source of truth for paths, model, server, and the per-task timeout; every other script sources it.scripts/setup.sh— one-time setup (venv, install/clone CyberGym, fetch the subset via Git LFS, pull the Docker images, install this runner, build the agent image).scripts/start_server.sh— starts the CyberGym submission server.scripts/run_10_tasks.sh→scripts/run_subset.sh— renamed and slimmed to source shared config.scripts/validate.sh— post-validates submitted PoCs and prints a per-task pass/fail summary (collapses the two old, partly-redundant validation sections into one).setup.sh→start_server.sh→run_subset.sh→validate.sh, keeping the direct runner invocation inline to show how the agent is called.Notable fixes surfaced along the way
setup.shnow generates a random key into the gitignored.env; server and validation read it from there. Only local/submit-fix,/query-poc,/verify-agent-pocsuse it — the agent's own/submit-vulis public.TIMEOUTnow defaults to 14400s.curlin validation (verify_agent_result.pyalready POSTs to/verify-agent-pocs).CI notes
.shfiles need no SPDX header (matches the priorrun_10_tasks.sh).🤖 Generated with Claude Code