feat(skills): add /pentest bundled skill#137
Merged
emal-avala merged 1 commit intomainfrom Apr 18, 2026
Merged
Conversation
Adds a white-box penetration test workflow as a bundled skill, runnable via /pentest [target-dir]. The skill runs five phases in order: recon, slice, vulnerability analysis, exploit-or-discard, and reporting. The exploit-or-discard gate requires a concrete proof-of-concept for every finding; anything that cannot be demonstrated from code inspection alone is demoted to INFO or dropped, keeping the output free of theoretical findings. The skill writes a severity-grouped markdown report (CRITICAL / HIGH / MEDIUM / LOW / INFO) with file:line, CWE mapping, vulnerable snippet, fix, impact, and PoC per finding. Dynamic verification is constrained to local development instances — the skill explicitly rejects running against production. - crates/lib/src/skills/mod.rs: register the bundled skill body - README.md: add /pentest row to the bundled skills table - CHANGELOG.md: note under Unreleased All 6 skills_integration tests pass.
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
Adds a
/pentestbundled skill that runs a five-phase white-box penetration test in-session: recon → slice → vuln analysis → exploit-or-discard → report. Modeled on the phase structure used by mature autonomous pentest pipelines, but runs entirely in the agent loop rather than shelling out.The exploit-or-discard gate is the load-bearing piece: every finding must ship with a reproducible PoC (curl command, payload string, or UI repro steps). Findings that cannot be demonstrated from code inspection alone get demoted to INFO or dropped — no theoretical findings pollute the report. Dynamic verification is explicitly constrained to local development instances.
Output is a severity-grouped markdown report (CRITICAL / HIGH / MEDIUM / LOW / INFO) with file:line, CWE mapping, vulnerable snippet, fix, impact, and PoC per finding, plus a ship-readiness verdict.
Complements the existing
/security-reviewskill:security-reviewis a diff-scoped CWE pass over pending changes;pentestis a full-repo or subsystem-scoped workflow with a stricter proof gate.Changes
crates/lib/src/skills/mod.rs— register the bundled skill bodyREADME.md— add/pentestrow to the bundled skills tableCHANGELOG.md— note under UnreleasedTest plan
cargo test -p agent-code-lib --test skills_integration— all 6 tests passcargo build -p agent-code-lib— clean/pentest <some-subdir>in a local agent session and confirm phase 1 recon runs as expected