Skip to content

fix(assail): UnboundedAllocation is Medium, not Critical#128

Closed
hyperpolymath wants to merge 9 commits into
mainfrom
claude/unbounded-allocation-severity
Closed

fix(assail): UnboundedAllocation is Medium, not Critical#128
hyperpolymath wants to merge 9 commits into
mainfrom
claude/unbounded-allocation-severity

Conversation

@hyperpolymath

Copy link
Copy Markdown
Owner

Summary

  • Changes Severity::CriticalSeverity::Medium for the UnboundedAllocation weak point emitted by fn analyze_rust in src/assail/analyzer.rs.
  • This was the only WeakPoint construction for this category; the other occurrences in analyzer.rs are .filter() calls on existing findings and are unaffected.

Motivation: The heuristic keyword match ("potential pattern") was calibrated at Critical but was ~70% of all estate Criticals in the 2026-06-11 estate-loop audit, drowning confirmed findings. Hypatia's ingest already applies a cap to Medium for scans produced by older binaries; this aligns the source so the raw store and PR-scan-comments also reflect the corrected severity estate-wide.

Test plan

  • cargo build --release — clean
  • cargo test --release assail — pass (no test asserts UnboundedAllocation == Critical)
  • cargo test --release unbounded — 6 detection tests green (assert presence, not severity)
  • Pre-existing clippy errors in analyzer.rs confirmed pre-existing; change adds zero new warnings

🤖 Generated with Claude Code

hyperpolymath and others added 9 commits May 31, 2026 00:23
…rse failure)

panic-attack rust-ci.yml + 42 other estate repos that pin
rust-ci-reusable.yml@cc5a372a have been failing with 0s-duration
"workflow file issue" parse errors since 2026-05-26 (when PR #45
introduced the thin-wrapper). cc5a372a IS reachable from
standards/main (verified via git merge-base --is-ancestor), so
this is NOT the orphan-SHA failure mode panic-attack#84 thought
it was fixing.

Empirical: every recent rust-ci run failed at parse time. The
reusable's content at cc5a372a is structurally fine. The simplest
hypothesis is that GH Actions has cached a bad resolution of this
specific SHA — repinning to a fresher merge-commit forces re-fetch.

822fa14e is the current HEAD of standards/main on the file
(standards#299, "pass --locked to cargo check/clippy/test").
Bumping forward to that brings the --locked safety along with
the unblock.

If 822fa14e also fails parse: the issue is not the SHA, and the
caller-side workflow file needs investigation (the workflow's
API-reported name field has been ".github/workflows/rust-ci.yml"
not "Rust CI", suggesting GH never parsed it successfully).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- C001: CodeQL language fixes
- C002: License identifier standardization
- C003: Outdated actions audit
- C004: Pin standards refs to SHA 861b5e9
- C005: Add workflow-level permissions
Heuristic keyword match at ~70% of all estate Criticals (2026-06-11
estate-loop audit), drowning confirmed findings. Medium reflects actual
confirmation strength. Hypatia's ingest already applies the same cap
for scans from older binaries; this aligns the source.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
@hyperpolymath hyperpolymath enabled auto-merge (squash) June 13, 2026 21:03
@hyperpolymath

Copy link
Copy Markdown
Owner Author

Superseded by #129 (claude/unbounded-severity-v2) — rebased onto current main.

auto-merge was automatically disabled June 13, 2026 21:12

Pull request was closed

@hyperpolymath hyperpolymath deleted the claude/unbounded-allocation-severity branch June 13, 2026 21:14
hyperpolymath added a commit that referenced this pull request Jun 13, 2026
## Summary

Changes `Severity::Critical` → `Severity::Medium` for the
`UnboundedAllocation` weak point emitted by `fn analyze_rust` in
`src/assail/analyzer.rs`. This is the only `WeakPoint` construction for
this category; the other occurrences in `analyzer.rs` are `.filter()`
calls on existing findings and are unaffected.

**Motivation:** The heuristic keyword match ("potential pattern") was
calibrated at `Critical` but was ~70% of all estate Criticals in the
2026-06-11 estate-loop audit, drowning confirmed findings. Hypatia's
ingest already caps it to `Medium` for scans produced by older binaries;
this aligns the source so the raw store and PR-scan-comments also
reflect the corrected severity estate-wide.

Rebased onto current main (6a814fa). Supersedes #128.

## Test plan

- [x] `cargo build --release` — clean
- [x] `cargo test --release assail` — pass
- [x] `cargo test --release unbounded` — 6 detection tests green (assert
presence, not severity)
- [x] Change adds no new clippy warnings

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants