Conversation
…t pipeline (#3050) Fix five defects in the bundled skill trust and injection scanner pipeline: - bootstrap/mod.rs: chain .with_hub_dirs() before scan_loaded() so the injection scanner receives hub directory context on every startup (#3045) - agent/mod.rs: replace fresh SkillRegistry construction in reload_skills() with registry.reload() to preserve hub_dirs across hot-reload ticks (#3046) - runner.rs + agent/mod.rs: guard trust elevation with a compile-time bundled_skill_names() allowlist; skills with a forged .bundled marker that are not in the allowlist are classified as Hub and a warning is emitted (#3049) - runner.rs, agent/mod.rs: fix misleading comment — migration promotes Quarantined/Verified to Trusted; only Blocked skills are skipped (#3047) - os-automation/SKILL.md: bump version to 1.1 to trigger re-provisioning and restore the missing shell boundary instruction (#3048) Closes #3045, #3046, #3047, #3048, #3049, #3050
This was
linked to
issues
Apr 15, 2026
Closed
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
bootstrap/mod.rs— chain.with_hub_dirs()beforescan_loaded()so the injection scanner receives hub directory context on every startup; the defense added in fix(skills): filter .bundled marker from hub-install packages to prevent trust escalation #3040 was never active in productionagent/mod.rs— replace freshSkillRegistryconstruction inreload_skills()withregistry.reload()to preservehub_dirsacross hot-reload ticksrunner.rs+agent/mod.rs— guard trust elevation with abundled_skill_names()allowlist on both startup and hot-reload paths; forged.bundledmarkers are rejected and loggedrunner.rsandagent/mod.rs)os-automationSKILL.md to version1.1to trigger re-provisioning and restore missing shell boundary instructionTest plan
cargo nextest run --workspace --all-features --lib --bins)cargo +nightly fmt --checkpassescargo clippy --workspace -- -D warningspassescargo run --features full -- --config .local/config/testing.toml— all 5 bundled skills should load astrustedafter startup migrationos-automationSKILL.md on disk matches embedded content after startup (shell boundary instruction present).bundledmarker in hub dir does not elevate skill toTrusted(check warning log)Closes #3045, #3046, #3047, #3048, #3049, #3050