Automated audit: This PR was generated by NLPM, a natural language programming linter, running via claude-code-action. Please evaluate the diff on its merits.
About this audit
Orca was audited by NLPM, a natural language programming linter that scores Claude Code skills, agents, and configuration files on a 100-point scale. The audit scanned 9 NL artifacts and produced an overall score of 96/100 — a strong result that reflects well-written, self-contained skills.
This issue summarises the findings that warranted follow-up. Quality/style issues (vague quantifiers etc.) are listed for information only; the PRs below address only concrete bugs and low/medium security improvements.
Bug fixes
Bug #1 — Broken doc references in skills/orca-cli/SKILL.md (Priority: Medium)
The ## References section (lines 568–570) lists three documentation files that do not exist in the repository:
docs/orca-cli-focused-v1-status.md
docs/orca-cli-v1-spec.md
docs/orca-runtime-layer-design.md
Agents following the skill will hit dead links when trying to resolve ambiguous CLI behavior, silently losing the intended guidance. The skill is otherwise fully self-contained.
PR: #973
Security improvements (Medium / Low only)
The audit also detected one HIGH pattern (the postinstall script in package.json rebuilds native modules — standard Electron practice, not a vulnerability) and two lower-severity items that are safe to address via PR.
Security Fix #1 — ORCA_ELECTRON_VITE_CLI override lacks an environment guard (Medium)
config/scripts/run-electron-vite-dev.mjs allows the ORCA_ELECTRON_VITE_CLI env var to substitute the electron-vite binary unconditionally. The comment documents this as a test affordance, but without a guard the override is active in dev and CI runs where a compromised environment could inject an arbitrary binary path.
PR: #974
Security Fix #2 — Critical native deps use ^ semver constraints (Low)
node-pty, @parcel/watcher, and electron — all compiled native modules listed in pnpm.onlyBuiltDependencies — use ^ constraints. This allows silent minor/patch upgrades that could ship a different native binary or a vulnerable version. Notably, node-pty already has a patch file targeting 1.1.0 exactly, which a ^ bump could silently skip.
PR: #975
Quality notes (informational, no PRs)
These are style-level findings from the scoring run. No action required — listed here for completeness.
| File |
Issue |
Score impact |
skills/orca-cli/SKILL.md |
6 vague quantifiers ("significant checkpoint", "meaningful progress", etc.) |
-12 |
.agents/skills/typescript/SKILL.md |
"complex type definitions", "when applicable" |
-6 |
.agents/skills/auto-review-fix/SKILL.md |
"relevant to that review type" |
-4 |
.agents/skills/react-useeffect/SKILL.md |
"Expensive calculations", "when possible" |
-4 |
.agents/skills/auto-pr-merge/SKILL.md |
"when appropriate" |
-2 |
.agents/skills/auto-submit/SKILL.md |
"fails catastrophically" (no observable threshold) |
-2 |
CLAUDE.md |
"non-obvious constraint" (subjective threshold) |
-2 |
Thank you for maintaining a well-structured, high-quality set of NL artifacts. Feel free to close any PR you disagree with — the diff should stand on its own merits.
About this audit
Orca was audited by NLPM, a natural language programming linter that scores Claude Code skills, agents, and configuration files on a 100-point scale. The audit scanned 9 NL artifacts and produced an overall score of 96/100 — a strong result that reflects well-written, self-contained skills.
This issue summarises the findings that warranted follow-up. Quality/style issues (vague quantifiers etc.) are listed for information only; the PRs below address only concrete bugs and low/medium security improvements.
Bug fixes
Bug #1 — Broken doc references in
skills/orca-cli/SKILL.md(Priority: Medium)The
## Referencessection (lines 568–570) lists three documentation files that do not exist in the repository:docs/orca-cli-focused-v1-status.mddocs/orca-cli-v1-spec.mddocs/orca-runtime-layer-design.mdAgents following the skill will hit dead links when trying to resolve ambiguous CLI behavior, silently losing the intended guidance. The skill is otherwise fully self-contained.
PR: #973
Security improvements (Medium / Low only)
The audit also detected one HIGH pattern (the
postinstallscript inpackage.jsonrebuilds native modules — standard Electron practice, not a vulnerability) and two lower-severity items that are safe to address via PR.Security Fix #1 —
ORCA_ELECTRON_VITE_CLIoverride lacks an environment guard (Medium)config/scripts/run-electron-vite-dev.mjsallows theORCA_ELECTRON_VITE_CLIenv var to substitute the electron-vite binary unconditionally. The comment documents this as a test affordance, but without a guard the override is active in dev and CI runs where a compromised environment could inject an arbitrary binary path.PR: #974
Security Fix #2 — Critical native deps use
^semver constraints (Low)node-pty,@parcel/watcher, andelectron— all compiled native modules listed inpnpm.onlyBuiltDependencies— use^constraints. This allows silent minor/patch upgrades that could ship a different native binary or a vulnerable version. Notably,node-ptyalready has a patch file targeting1.1.0exactly, which a^bump could silently skip.PR: #975
Quality notes (informational, no PRs)
These are style-level findings from the scoring run. No action required — listed here for completeness.
skills/orca-cli/SKILL.md.agents/skills/typescript/SKILL.md.agents/skills/auto-review-fix/SKILL.md.agents/skills/react-useeffect/SKILL.md.agents/skills/auto-pr-merge/SKILL.md.agents/skills/auto-submit/SKILL.mdCLAUDE.mdThank you for maintaining a well-structured, high-quality set of NL artifacts. Feel free to close any PR you disagree with — the diff should stand on its own merits.