[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M75KRCZHKE0ZYAQFFYHS0YNP":3,"skill-same:01M75KRCZHKE0ZYAQFFYHS0YNP":250},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":35,"commitSha":36,"skillDirectory":37,"forksCount":38,"repositoryHtmlUrl":39,"repositoryHomepageUrl":40,"ownerHtmlUrl":41,"ownerBlogUrl":42,"ownerDescription":6,"updatedAt":43,"install":44,"resources":57},"01M75KRCZHKE0ZYAQFFYHS0YNP","zeroize-audit",null,"Detects missing zeroization of sensitive data in source code and identifies zeroization removed by compiler optimizations, with assembly-level analysis, and control-flow verification. Use for auditing C\u002FC++\u002FRust code handling secrets, keys, passwords, or other sensitive data.","agisota\u002Fold-one","agisota","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F121413592?v=4","01M75B4BRNT67EC1NDRT8QZEP2","01M75B4BS7QR9VP580MSCPTS3X",3,"MIT","https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: zeroize-audit\ndescription: \"Detects missing zeroization of sensitive data in source code and identifies zeroization removed by compiler optimizations, with assembly-level analysis, and control-flow verification. Use for auditing C\u002FC++\u002FRust code handling secrets, keys, passwords, or other sensitive data.\"\nallowed-tools:\n  - Read\n  - Grep\n  - Glob\n  - Bash\n  - Write\n  - Task\n  - AskUserQuestion\n  - mcp__serena__activate_project\n  - mcp__serena__find_symbol\n  - mcp__serena__find_referencing_symbols\n  - mcp__serena__get_symbols_overview\n---\n\n# zeroize-audit — Claude Skill\n\n## When to Use\n- Auditing cryptographic implementations (keys, seeds, nonces, secrets)\n- Reviewing authentication systems (passwords, tokens, session data)\n- Analyzing code that handles PII or sensitive credentials\n- Verifying secure cleanup in security-critical codebases\n- Investigating memory safety of sensitive data handling\n\n## When NOT to Use\n- General code review without security focus\n- Performance optimization (unless related to secure wiping)\n- Refactoring tasks not related to sensitive data\n- Code without identifiable secrets or sensitive values\n\n---\n\n## Purpose\nDetect missing zeroization of sensitive data in source code and identify zeroization that is removed or weakened by compiler optimizations (e.g., dead-store elimination), with mandatory LLVM IR\u002Fasm evidence. Capabilities include:\n- Assembly-level analysis for register spills and stack retention\n- Data-flow tracking for secret copies\n- Heap allocator security warnings\n- Semantic IR analysis for loop unrolling and SSA form\n- Control-flow graph analysis for path coverage verification\n- Runtime validation test generation\n\n## Scope\n- Read-only against the target codebase (does not modify audited code; writes analysis artifacts to a temporary working directory).\n- Produces a structured report (JSON).\n- Requires valid build context (`compile_commands.json`) and compilable translation units.\n- \"Optimized away\" findings only allowed with compiler evidence (IR\u002Fasm diff).\n\n---\n\n## Inputs\n\nSee `{baseDir}\u002Fschemas\u002Finput.json` for the full schema. Key fields:\n\n| Field | Required | Default | Description |\n|---|---|---|---|\n| `path` | yes | — | Repo root |\n| `compile_db` | no | `null` | Path to `compile_commands.json` for C\u002FC++ analysis. Required if `cargo_manifest` is not set. |\n| `cargo_manifest` | no | `null` | Path to `Cargo.toml` for Rust crate analysis. Required if `compile_db` is not set. |\n| `config` | no | — | YAML defining heuristics and approved wipes |\n| `opt_levels` | no | `[\"O0\",\"O1\",\"O2\"]` | Optimization levels for IR comparison. O1 is the diagnostic level: if a wipe disappears at O1 it is simple DSE; O2 catches more aggressive eliminations. |\n| `languages` | no | `[\"c\",\"cpp\",\"rust\"]` | Languages to analyze |\n| `max_tus` | no | — | Limit on translation units processed from compile DB |\n| `mcp_mode` | no | `prefer` | `off`, `prefer`, or `require` — controls Serena MCP usage |\n| `mcp_required_for_advanced` | no | `true` | Downgrade `SECRET_COPY`, `MISSING_ON_ERROR_PATH`, and `NOT_DOMINATING_EXITS` to `needs_review` when MCP is unavailable |\n| `mcp_timeout_ms` | no | — | Timeout budget for MCP semantic queries |\n| `poc_categories` | no | all 11 exploitable | Finding categories for which to generate PoCs. C\u002FC++ findings: all 11 categories supported. Rust findings: only `MISSING_SOURCE_ZEROIZE`, `SECRET_COPY`, and `PARTIAL_WIPE` are supported; other Rust categories are marked `poc_supported=false`. |\n| `poc_output_dir` | no | `generated_pocs\u002F` | Output directory for generated PoCs |\n| `enable_asm` | no | `true` | Enable assembly emission and analysis (Step 8); produces `STACK_RETENTION`, `REGISTER_SPILL`. Auto-disabled if `emit_asm.sh` is missing. |\n| `enable_semantic_ir` | no | `false` | Enable semantic LLVM IR analysis (Step 9); produces `LOOP_UNROLLED_INCOMPLETE` |\n| `enable_cfg` | no | `false` | Enable control-flow graph analysis (Step 10); produces `MISSING_ON_ERROR_PATH`, `NOT_DOMINATING_EXITS` |\n| `enable_runtime_tests` | no | `false` | Enable runtime test harness generation (Step 11) |\n\n---\n\n## Prerequisites\n\nBefore running, verify the following. Each has a defined failure mode.\n\n**C\u002FC++ prerequisites:**\n\n| Prerequisite | Failure mode if missing |\n|---|---|\n| `compile_commands.json` at `compile_db` path | Fail fast — do not proceed |\n| `clang` on PATH | Fail fast — IR\u002FASM analysis impossible |\n| `uvx` on PATH (for Serena) | If `mcp_mode=require`: fail. If `mcp_mode=prefer`: continue without MCP; downgrade affected findings per Confidence Gating rules. |\n| `{baseDir}\u002Ftools\u002Fextract_compile_flags.py` | Fail fast — cannot extract per-TU flags |\n| `{baseDir}\u002Ftools\u002Femit_ir.sh` | Fail fast — IR analysis impossible |\n| `{baseDir}\u002Ftools\u002Femit_asm.sh` | Warn and skip assembly findings (STACK_RETENTION, REGISTER_SPILL) |\n| `{baseDir}\u002Ftools\u002Fmcp\u002Fcheck_mcp.sh` | Warn and treat as MCP unavailable |\n| `{baseDir}\u002Ftools\u002Fmcp\u002Fnormalize_mcp_evidence.py` | Warn and use raw MCP output |\n\n**Rust prerequisites:**\n\n| Prerequisite | Failure mode if missing |\n|---|---|\n| `Cargo.toml` at `cargo_manifest` path | Fail fast — do not proceed |\n| `cargo check` passes | Fail fast — crate must be buildable |\n| `cargo +nightly` on PATH | Fail fast — nightly required for MIR and LLVM IR emission |\n| `uv` on PATH | Fail fast — required to run Python analysis scripts |\n| `{baseDir}\u002Ftools\u002Fvalidate_rust_toolchain.sh` | Warn — run preflight manually. Checks all tools, scripts, nightly, and optionally `cargo check`. Use `--json` for machine-readable output, `--manifest` to also validate the crate builds. |\n| `{baseDir}\u002Ftools\u002Femit_rust_mir.sh` | Fail fast — MIR analysis impossible (`--opt`, `--crate`, `--bin\u002F--lib` supported; `--out` can be file or directory) |\n| `{baseDir}\u002Ftools\u002Femit_rust_ir.sh` | Fail fast — LLVM IR analysis impossible (`--opt` required; `--crate`, `--bin\u002F--lib` supported; `--out` must be `.ll`) |\n| `{baseDir}\u002Ftools\u002Femit_rust_asm.sh` | Warn and skip assembly findings (`STACK_RETENTION`, `REGISTER_SPILL`). Supports `--opt`, `--crate`, `--bin\u002F--lib`, `--target`, `--intel-syntax`; `--out` can be `.s` file or directory. |\n| `{baseDir}\u002Ftools\u002Fdiff_rust_mir.sh` | Warn and skip MIR-level optimization comparison. Accepts 2+ MIR files, normalizes, diffs pairwise, and reports first opt level where zeroize\u002Fdrop-glue patterns disappear. |\n| `{baseDir}\u002Ftools\u002Fscripts\u002Fsemantic_audit.py` | Warn and skip semantic source analysis |\n| `{baseDir}\u002Ftools\u002Fscripts\u002Ffind_dangerous_apis.py` | Warn and skip dangerous API scan |\n| `{baseDir}\u002Ftools\u002Fscripts\u002Fcheck_mir_patterns.py` | Warn and skip MIR analysis |\n| `{baseDir}\u002Ftools\u002Fscripts\u002Fcheck_llvm_patterns.py` | Warn and skip LLVM IR analysis |\n| `{baseDir}\u002Ftools\u002Fscripts\u002Fcheck_rust_asm.py` | Warn and skip Rust assembly analysis (`STACK_RETENTION`, `REGISTER_SPILL`, drop-glue checks). Dispatches to `check_rust_asm_x86.py` (production) or `check_rust_asm_aarch64.py` (**EXPERIMENTAL** — AArch64 findings require manual verification). |\n| `{baseDir}\u002Ftools\u002Fscripts\u002Fcheck_rust_asm_x86.py` | Required by `check_rust_asm.py` for x86-64 analysis; warn and skip if missing |\n| `{baseDir}\u002Ftools\u002Fscripts\u002Fcheck_rust_asm_aarch64.py` | Required by `check_rust_asm.py` for AArch64 analysis (**EXPERIMENTAL**); warn and skip if missing |\n\n**Common prerequisite:**\n\n| Prerequisite | Failure mode if missing |\n|---|---|\n| `{baseDir}\u002Ftools\u002Fgenerate_poc.py` | Fail fast — PoC generation is mandatory |\n\n---\n\n## Approved Wipe APIs\n\nThe following are recognized as valid zeroization. Configure additional entries in `{baseDir}\u002Fconfigs\u002F`.\n\n**C\u002FC++**\n- `explicit_bzero`\n- `memset_s`\n- `SecureZeroMemory`\n- `OPENSSL_cleanse`\n- `sodium_memzero`\n- Volatile wipe loops (pattern-based; see `volatile_wipe_patterns` in `{baseDir}\u002Fconfigs\u002Fdefault.yaml`)\n- In IR: `llvm.memset` with volatile flag, volatile stores, or non-elidable wipe call\n\n**Rust**\n- `zeroize::Zeroize` trait (`zeroize()` method)\n- `Zeroizing\u003CT>` wrapper (drop-based)\n- `ZeroizeOnDrop` derive macro\n\n---\n\n## Finding Capabilities\n\nFindings are grouped by required evidence. Only attempt findings for which the required tooling is available.\n\n| Finding ID | Description | Requires | PoC Support |\n|---|---|---|---|\n| `MISSING_SOURCE_ZEROIZE` | No zeroization found in source | Source only | Yes (C\u002FC++ + Rust) |\n| `PARTIAL_WIPE` | Incorrect size or incomplete wipe | Source only | Yes (C\u002FC++ + Rust) |\n| `NOT_ON_ALL_PATHS` | Zeroization missing on some control-flow paths (heuristic) | Source only | Yes (C\u002FC++ only) |\n| `SECRET_COPY` | Sensitive data copied without zeroization tracking | Source + MCP preferred | Yes (C\u002FC++ + Rust) |\n| `INSECURE_HEAP_ALLOC` | Secret uses insecure allocator (malloc vs. secure_malloc) | Source only | Yes (C\u002FC++ only) |\n| `OPTIMIZED_AWAY_ZEROIZE` | Compiler removed zeroization | IR diff required (never source-only) | Yes |\n| `STACK_RETENTION` | Stack frame may retain secrets after return | Assembly required (C\u002FC++); LLVM IR `alloca`+`lifetime.end` evidence (Rust); assembly corroboration upgrades to `confirmed` | Yes (C\u002FC++ only) |\n| `REGISTER_SPILL` | Secrets spilled from registers to stack | Assembly required (C\u002FC++); LLVM IR `load`+call-site evidence (Rust); assembly corroboration upgrades to `confirmed` | Yes (C\u002FC++ only) |\n| `MISSING_ON_ERROR_PATH` | Error-handling paths lack cleanup | CFG or MCP required | Yes |\n| `NOT_DOMINATING_EXITS` | Wipe doesn't dominate all exits | CFG or MCP required | Yes |\n| `LOOP_UNROLLED_INCOMPLETE` | Unrolled loop wipe is incomplete | Semantic IR required | Yes |\n\n---\n\n## Agent Architecture\n\nThe analysis pipeline uses 11 agents across 8 phases, invoked by the orchestrator (`{baseDir}\u002Fprompts\u002Ftask.md`) via `Task`. Agents write persistent finding files to a shared working directory (`\u002Ftmp\u002Fzeroize-audit-{run_id}\u002F`), enabling parallel execution and protecting against context pressure.\n\n| Agent | Phase | Purpose | Output Directory |\n|---|---|---|---|\n| `0-preflight` | Phase 0 | Preflight checks (tools, toolchain, compile DB, crate build), config merge, workdir creation, TU enumeration | `{workdir}\u002F` |\n| `1-mcp-resolver` | Phase 1, Wave 1 (C\u002FC++ only) | Resolve symbols, types, and cross-file references via Serena MCP | `mcp-evidence\u002F` |\n| `2-source-analyzer` | Phase 1, Wave 2a (C\u002FC++ only) | Identify sensitive objects, detect wipes, validate correctness, data-flow\u002Fheap | `source-analysis\u002F` |\n| `2b-rust-source-analyzer` | Phase 1, Wave 2b (Rust only, parallel with 2a) | Rustdoc JSON trait-aware analysis + dangerous API grep | `source-analysis\u002F` |\n| `3-tu-compiler-analyzer` | Phase 2, Wave 3 (C\u002FC++ only, N parallel) | Per-TU IR diff, assembly, semantic IR, CFG analysis | `compiler-analysis\u002F{tu_hash}\u002F` |\n| `3b-rust-compiler-analyzer` | Phase 2, Wave 3R (Rust only, single agent) | Crate-level MIR, LLVM IR, and assembly analysis | `rust-compiler-analysis\u002F` |\n| `4-report-assembler` | Phase 3 (interim) + Phase 6 (final) | Collect findings from all agents, apply confidence gates; merge PoC results and produce final report | `report\u002F` |\n| `5-poc-generator` | Phase 4 | Craft bespoke proof-of-concept programs (C\u002FC++: all categories; Rust: MISSING_SOURCE_ZEROIZE, SECRET_COPY, PARTIAL_WIPE) | `poc\u002F` |\n| `5b-poc-validator` | Phase 5 | Compile and run all PoCs | `poc\u002F` |\n| `5c-poc-verifier` | Phase 5 | Verify each PoC proves its claimed finding | `poc\u002F` |\n| `6-test-generator` | Phase 7 (optional) | Generate runtime validation test harnesses | `tests\u002F` |\n\nThe orchestrator reads one per-phase workflow file from `{baseDir}\u002Fworkflows\u002F` at a time, and maintains `orchestrator-state.json` for recovery after context compression. Agents receive configuration by file path (`config_path`), not by value.\n\n### Execution flow\n\n```\nPhase 0: 0-preflight agent — Preflight + config + create workdir + enumerate TUs\n           → writes orchestrator-state.json, merged-config.yaml, preflight.json\nPhase 1: Wave 1:  1-mcp-resolver              (skip if mcp_mode=off OR language_mode=rust)\n         Wave 2a: 2-source-analyzer           (C\u002FC++ only; skip if no compile_db)  ─┐ parallel\n         Wave 2b: 2b-rust-source-analyzer     (Rust only; skip if no cargo_manifest) ─┘\nPhase 2: Wave 3:  3-tu-compiler-analyzer x N  (C\u002FC++ only; parallel per TU)\n         Wave 3R: 3b-rust-compiler-analyzer   (Rust only; single crate-level agent)\nPhase 3: Wave 4:  4-report-assembler          (mode=interim → findings.json; reads all agent outputs)\nPhase 4: Wave 5:  5-poc-generator             (C\u002FC++: all categories; Rust: MISSING_SOURCE_ZEROIZE, SECRET_COPY, PARTIAL_WIPE; other Rust findings: poc_supported=false)\nPhase 5: PoC Validation & Verification\n           Step 1: 5b-poc-validator agent      (compile and run all PoCs)\n           Step 2: 5c-poc-verifier agent       (verify each PoC proves its claimed finding)\n           Step 3: Orchestrator presents verification failures to user via AskUserQuestion\n           Step 4: Orchestrator merges all results into poc_final_results.json\nPhase 6: Wave 6: 4-report-assembler           (mode=final → merge PoC results, final-report.md)\nPhase 7: Wave 7: 6-test-generator             (optional)\nPhase 8: Orchestrator — Return final-report.md\n```\n\n## Cross-Reference Convention\n\nIDs are namespaced per agent to prevent collisions during parallel execution:\n\n| Entity | Pattern | Assigned By |\n|---|---|---|\n| Sensitive object (C\u002FC++) | `SO-0001`–`SO-4999` | `2-source-analyzer` |\n| Sensitive object (Rust) | `SO-5000`–`SO-9999` (Rust namespace) | `2b-rust-source-analyzer` |\n| Source finding (C\u002FC++) | `F-SRC-NNNN` | `2-source-analyzer` |\n| Source finding (Rust) | `F-RUST-SRC-NNNN` | `2b-rust-source-analyzer` |\n| IR finding (C\u002FC++) | `F-IR-{tu_hash}-NNNN` | `3-tu-compiler-analyzer` |\n| ASM finding (C\u002FC++) | `F-ASM-{tu_hash}-NNNN` | `3-tu-compiler-analyzer` |\n| CFG finding | `F-CFG-{tu_hash}-NNNN` | `3-tu-compiler-analyzer` |\n| Semantic IR finding | `F-SIR-{tu_hash}-NNNN` | `3-tu-compiler-analyzer` |\n| Rust MIR finding | `F-RUST-MIR-NNNN` | `3b-rust-compiler-analyzer` |\n| Rust LLVM IR finding | `F-RUST-IR-NNNN` | `3b-rust-compiler-analyzer` |\n| Rust assembly finding | `F-RUST-ASM-NNNN` | `3b-rust-compiler-analyzer` |\n| Translation unit | `TU-{hash}` | Orchestrator |\n| Final finding | `ZA-NNNN` | `4-report-assembler` |\n\nEvery finding JSON object includes `related_objects`, `related_findings`, and `evidence_files` fields for cross-referencing between agents.\n\n---\n\n## Detection Strategy\n\nAnalysis runs in two phases. For complete step-by-step guidance, see `{baseDir}\u002Freferences\u002Fdetection-strategy.md`.\n\n| Phase | Steps | Findings produced | Required tooling |\n|---|---|---|---|\n| Phase 1 (Source) | 1–6 | `MISSING_SOURCE_ZEROIZE`, `PARTIAL_WIPE`, `NOT_ON_ALL_PATHS`, `SECRET_COPY`, `INSECURE_HEAP_ALLOC` | Source + compile DB |\n| Phase 2 (Compiler) | 7–12 | `OPTIMIZED_AWAY_ZEROIZE`, `STACK_RETENTION`*, `REGISTER_SPILL`*, `LOOP_UNROLLED_INCOMPLETE`†, `MISSING_ON_ERROR_PATH`‡, `NOT_DOMINATING_EXITS`‡ | `clang`, IR\u002FASM tools |\n\n\\* requires `enable_asm=true` (default)\n† requires `enable_semantic_ir=true`\n‡ requires `enable_cfg=true`\n\n---\n\n\n## Output Format\n\nEach run produces two outputs:\n\n1. **`final-report.md`** — Comprehensive markdown report (primary human-readable output)\n2. **`findings.json`** — Structured JSON matching `{baseDir}\u002Fschemas\u002Foutput.json` (for machine consumption and downstream tools)\n\n### Markdown Report Structure\n\nThe markdown report (`final-report.md`) contains these sections:\n\n- **Header**: Run metadata (run_id, timestamp, repo, compile_db, config summary)\n- **Executive Summary**: Finding counts by severity, confidence, and category\n- **Sensitive Objects Inventory**: Table of all identified objects with IDs, types, locations\n- **Findings**: Grouped by severity then confidence. Each finding includes location, object, all evidence (source\u002FIR\u002FASM\u002FCFG), compiler evidence details, and recommended fix\n- **Superseded Findings**: Source findings replaced by CFG-backed findings\n- **Confidence Gate Summary**: Downgrades applied and overrides rejected\n- **Analysis Coverage**: TUs analyzed, agent success\u002Ffailure, features enabled\n- **Appendix: Evidence Files**: Mapping of finding IDs to evidence file paths\n\n### Structured JSON\n\nThe `findings.json` file follows the schema in `{baseDir}\u002Fschemas\u002Foutput.json`. Each `Finding` object:\n\n```json\n{\n  \"id\": \"ZA-0001\",\n  \"category\": \"OPTIMIZED_AWAY_ZEROIZE\",\n  \"severity\": \"high\",\n  \"confidence\": \"confirmed\",\n  \"language\": \"c\",\n  \"file\": \"src\u002Fcrypto.c\",\n  \"line\": 42,\n  \"symbol\": \"key_buf\",\n  \"evidence\": \"store volatile i8 0 count: O0=32, O2=0 — wipe eliminated by DSE\",\n  \"compiler_evidence\": {\n    \"opt_levels\": [\"O0\", \"O2\"],\n    \"o0\": \"32 volatile stores targeting key_buf\",\n    \"o2\": \"0 volatile stores (all eliminated)\",\n    \"diff_summary\": \"All volatile wipe stores removed at O2 — classic DSE pattern\"\n  },\n  \"suggested_fix\": \"Replace memset with explicit_bzero or add compiler_fence(SeqCst) after the wipe\",\n  \"poc\": {\n    \"file\": \"generated_pocs\u002FZA-0001.c\",\n    \"makefile_target\": \"ZA-0001\",\n    \"compile_opt\": \"-O2\",\n    \"requires_manual_adjustment\": false,\n    \"validated\": true,\n    \"validation_result\": \"exploitable\"\n  }\n}\n```\n\nSee `{baseDir}\u002Fschemas\u002Foutput.json` for the full schema and enum values.\n\n---\n\n## Confidence Gating\n\n### Evidence thresholds\n\nA finding requires at least **2 independent signals** to be marked `confirmed`. With 1 signal, mark `likely`. With 0 strong signals (name-pattern match only), mark `needs_review`.\n\nSignals include: name pattern match, type hint match, explicit annotation, IR evidence, ASM evidence, MCP cross-reference, CFG evidence, PoC validation.\n\n### PoC validation as evidence signal\n\nEvery finding is validated against a bespoke PoC. After compilation and execution, each PoC is also verified to ensure it actually tests the claimed vulnerability. The combined result is an evidence signal:\n\n| PoC Result | Verified | Impact |\n|---|---|---|\n| Exit 0 (exploitable) | Yes | Strong signal — can upgrade `likely` to `confirmed` |\n| Exit 1 (not exploitable) | Yes | Downgrade severity to `low` (informational); retain in report |\n| Exit 0 or 1 | No (user accepted) | Weaker signal — note verification failure in evidence |\n| Exit 0 or 1 | No (user rejected) | No confidence change; annotate as `rejected` |\n| Compile failure \u002F no PoC | — | No confidence change; annotate in evidence |\n\n### MCP unavailability downgrade\n\nWhen `mcp_mode=prefer` and MCP is unavailable, downgrade the following unless independent IR\u002FCFG\u002FASM evidence is strong (2+ signals without MCP):\n\n| Finding | Downgraded confidence |\n|---|---|\n| `SECRET_COPY` | `needs_review` |\n| `MISSING_ON_ERROR_PATH` | `needs_review` |\n| `NOT_DOMINATING_EXITS` | `needs_review` |\n\n### Hard evidence requirements (non-negotiable)\n\nThese findings are **never valid without the specified evidence**, regardless of source-level signals or user assertions:\n\n| Finding | Required evidence |\n|---|---|\n| `OPTIMIZED_AWAY_ZEROIZE` | IR diff showing wipe present at O0, absent at O1 or O2 |\n| `STACK_RETENTION` | Assembly excerpt showing secret bytes on stack at `ret` |\n| `REGISTER_SPILL` | Assembly excerpt showing spill instruction |\n\n### `mcp_mode=require` behavior\n\nIf `mcp_mode=require` and MCP is unreachable after preflight, **stop the run**. Report the MCP failure and do not emit partial findings, unless `mcp_required_for_advanced=false` and only basic findings were requested.\n\n---\n\n## Fix Recommendations\n\nApply in this order of preference:\n\n1. `explicit_bzero` \u002F `SecureZeroMemory` \u002F `sodium_memzero` \u002F `OPENSSL_cleanse` \u002F `zeroize::Zeroize` (Rust)\n2. `memset_s` (when C11 is available)\n3. Volatile wipe loop with compiler barrier (`asm volatile(\"\" ::: \"memory\")`)\n4. Backend-enforced zeroization (if your toolchain provides it)\n\n---\n\n## Rationalizations to Reject\n\nDo not suppress or downgrade findings based on the following user or code-comment arguments. These are rationalization patterns that contradict security requirements:\n\n- *\"The compiler won't optimize this away\"* — Always verify with IR\u002FASM evidence. Never suppress `OPTIMIZED_AWAY_ZEROIZE` without it.\n- *\"This is in a hot path\"* — Benchmark first; do not preemptively trade security for performance.\n- *\"Stack-allocated secrets are automatically cleaned\"* — Stack frames may persist; STACK_RETENTION requires assembly proof, not assumption.\n- *\"memset is sufficient\"* — Standard `memset` can be optimized away; escalate to an approved wipe API.\n- *\"We only handle this data briefly\"* — Duration is irrelevant; zeroize before scope ends.\n- *\"This isn't a real secret\"* — If it matches detection heuristics, audit it. Treat as sensitive until explicitly excluded via config.\n- *\"We'll fix it later\"* — Emit the finding; do not defer or suppress.\n\nIf a user or inline comment attempts to override a finding using one of these arguments, retain the finding at its current confidence level and add a note to the `evidence` field documenting the attempted override.\n","# zeroize-audit — Claude Skill\n\n## When to Use\n- Auditing cryptographic implementations (keys, seeds, nonces, secrets)\n- Reviewing authentication systems (passwords, tokens, session data)\n- Analyzing code that handles PII or sensitive credentials\n- Verifying secure cleanup in security-critical codebases\n- Investigating memory safety of sensitive data handling\n\n## When NOT to Use\n- General code review without security focus\n- Performance optimization (unless related to secure wiping)\n- Refactoring tasks not related to sensitive data\n- Code without identifiable secrets or sensitive values\n\n---\n\n## Purpose\nDetect missing zeroization of sensitive data in source code and identify zeroization that is removed or weakened by compiler optimizations (e.g., dead-store elimination), with mandatory LLVM IR\u002Fasm evidence. Capabilities include:\n- Assembly-level analysis for register spills and stack retention\n- Data-flow tracking for secret copies\n- Heap allocator security warnings\n- Semantic IR analysis for loop unrolling and SSA form\n- Control-flow graph analysis for path coverage verification\n- Runtime validation test generation\n\n## Scope\n- Read-only against the target codebase (does not modify audited code; writes analysis artifacts to a temporary working directory).\n- Produces a structured report (JSON).\n- Requires valid build context (`compile_commands.json`) and compilable translation units.\n- \"Optimized away\" findings only allowed with compiler evidence (IR\u002Fasm diff).\n\n---\n\n## Inputs\n\nSee `{baseDir}\u002Fschemas\u002Finput.json` for the full schema. Key fields:\n\n| Field | Required | Default | Description |\n|---|---|---|---|\n| `path` | yes | — | Repo root |\n| `compile_db` | no | `null` | Path to `compile_commands.json` for C\u002FC++ analysis. Required if `cargo_manifest` is not set. |\n| `cargo_manifest` | no | `null` | Path to `Cargo.toml` for Rust crate analysis. Required if `compile_db` is not set. |\n| `config` | no | — | YAML defining heuristics and approved wipes |\n| `opt_levels` | no | `[\"O0\",\"O1\",\"O2\"]` | Optimization levels for IR comparison. O1 is the diagnostic level: if a wipe disappears at O1 it is simple DSE; O2 catches more aggressive eliminations. |\n| `languages` | no | `[\"c\",\"cpp\",\"rust\"]` | Languages to analyze |\n| `max_tus` | no | — | Limit on translation units processed from compile DB |\n| `mcp_mode` | no | `prefer` | `off`, `prefer`, or `require` — controls Serena MCP usage |\n| `mcp_required_for_advanced` | no | `true` | Downgrade `SECRET_COPY`, `MISSING_ON_ERROR_PATH`, and `NOT_DOMINATING_EXITS` to `needs_review` when MCP is unavailable |\n| `mcp_timeout_ms` | no | — | Timeout budget for MCP semantic queries |\n| `poc_categories` | no | all 11 exploitable | Finding categories for which to generate PoCs. C\u002FC++ findings: all 11 categories supported. Rust findings: only `MISSING_SOURCE_ZEROIZE`, `SECRET_COPY`, and `PARTIAL_WIPE` are supported; other Rust categories are marked `poc_supported=false`. |\n| `poc_output_dir` | no | `generated_pocs\u002F` | Output directory for generated PoCs |\n| `enable_asm` | no | `true` | Enable assembly emission and analysis (Step 8); produces `STACK_RETENTION`, `REGISTER_SPILL`. Auto-disabled if `emit_asm.sh` is missing. |\n| `enable_semantic_ir` | no | `false` | Enable semantic LLVM IR analysis (Step 9); produces `LOOP_UNROLLED_INCOMPLETE` |\n| `enable_cfg` | no | `false` | Enable control-flow graph analysis (Step 10); produces `MISSING_ON_ERROR_PATH`, `NOT_DOMINATING_EXITS` |\n| `enable_runtime_tests` | no | `false` | Enable runtime test harness generation (Step 11) |\n\n---\n\n## Prerequisites\n\nBefore running, verify the following. Each has a defined failure mode.\n\n**C\u002FC++ prerequisites:**\n\n| Prerequisite | Failure mode if missing |\n|---|---|\n| `compile_commands.json` at `compile_db` path | Fail fast — do not proceed |\n| `clang` on PATH | Fail fast — IR\u002FASM analysis impossible |\n| `uvx` on PATH (for Serena) | If `mcp_mode=require`: fail. If `mcp_mode=prefer`: continue without MCP; downgrade affected findings per Confidence Gating rules. |\n| `{baseDir}\u002Ftools\u002Fextract_compile_flags.py` | Fail fast — cannot extract per-TU flags |\n| `{baseDir}\u002Ftools\u002Femit_ir.sh` | Fail fast — IR analysis impossible |\n| `{baseDir}\u002Ftools\u002Femit_asm.sh` | Warn and skip assembly findings (STACK_RETENTION, REGISTER_SPILL) |\n| `{baseDir}\u002Ftools\u002Fmcp\u002Fcheck_mcp.sh` | Warn and treat as MCP unavailable |\n| `{baseDir}\u002Ftools\u002Fmcp\u002Fnormalize_mcp_evidence.py` | Warn and use raw MCP output |\n\n**Rust prerequisites:**\n\n| Prerequisite | Failure mode if missing |\n|---|---|\n| `Cargo.toml` at `cargo_manifest` path | Fail fast — do not proceed |\n| `cargo check` passes | Fail fast — crate must be buildable |\n| `cargo +nightly` on PATH | Fail fast — nightly required for MIR and LLVM IR emission |\n| `uv` on PATH | Fail fast — required to run Python analysis scripts |\n| `{baseDir}\u002Ftools\u002Fvalidate_rust_toolchain.sh` | Warn — run preflight manually. Checks all tools, scripts, nightly, and optionally `cargo check`. Use `--json` for machine-readable output, `--manifest` to also validate the crate builds. |\n| `{baseDir}\u002Ftools\u002Femit_rust_mir.sh` | Fail fast — MIR analysis impossible (`--opt`, `--crate`, `--bin\u002F--lib` supported; `--out` can be file or directory) |\n| `{baseDir}\u002Ftools\u002Femit_rust_ir.sh` | Fail fast — LLVM IR analysis impossible (`--opt` required; `--crate`, `--bin\u002F--lib` supported; `--out` must be `.ll`) |\n| `{baseDir}\u002Ftools\u002Femit_rust_asm.sh` | Warn and skip assembly findings (`STACK_RETENTION`, `REGISTER_SPILL`). Supports `--opt`, `--crate`, `--bin\u002F--lib`, `--target`, `--intel-syntax`; `--out` can be `.s` file or directory. |\n| `{baseDir}\u002Ftools\u002Fdiff_rust_mir.sh` | Warn and skip MIR-level optimization comparison. Accepts 2+ MIR files, normalizes, diffs pairwise, and reports first opt level where zeroize\u002Fdrop-glue patterns disappear. |\n| `{baseDir}\u002Ftools\u002Fscripts\u002Fsemantic_audit.py` | Warn and skip semantic source analysis |\n| `{baseDir}\u002Ftools\u002Fscripts\u002Ffind_dangerous_apis.py` | Warn and skip dangerous API scan |\n| `{baseDir}\u002Ftools\u002Fscripts\u002Fcheck_mir_patterns.py` | Warn and skip MIR analysis |\n| `{baseDir}\u002Ftools\u002Fscripts\u002Fcheck_llvm_patterns.py` | Warn and skip LLVM IR analysis |\n| `{baseDir}\u002Ftools\u002Fscripts\u002Fcheck_rust_asm.py` | Warn and skip Rust assembly analysis (`STACK_RETENTION`, `REGISTER_SPILL`, drop-glue checks). Dispatches to `check_rust_asm_x86.py` (production) or `check_rust_asm_aarch64.py` (**EXPERIMENTAL** — AArch64 findings require manual verification). |\n| `{baseDir}\u002Ftools\u002Fscripts\u002Fcheck_rust_asm_x86.py` | Required by `check_rust_asm.py` for x86-64 analysis; warn and skip if missing |\n| `{baseDir}\u002Ftools\u002Fscripts\u002Fcheck_rust_asm_aarch64.py` | Required by `check_rust_asm.py` for AArch64 analysis (**EXPERIMENTAL**); warn and skip if missing |\n\n**Common prerequisite:**\n\n| Prerequisite | Failure mode if missing |\n|---|---|\n| `{baseDir}\u002Ftools\u002Fgenerate_poc.py` | Fail fast — PoC generation is mandatory |\n\n---\n\n## Approved Wipe APIs\n\nThe following are recognized as valid zeroization. Configure additional entries in `{baseDir}\u002Fconfigs\u002F`.\n\n**C\u002FC++**\n- `explicit_bzero`\n- `memset_s`\n- `SecureZeroMemory`\n- `OPENSSL_cleanse`\n- `sodium_memzero`\n- Volatile wipe loops (pattern-based; see `volatile_wipe_patterns` in `{baseDir}\u002Fconfigs\u002Fdefault.yaml`)\n- In IR: `llvm.memset` with volatile flag, volatile stores, or non-elidable wipe call\n\n**Rust**\n- `zeroize::Zeroize` trait (`zeroize()` method)\n- `Zeroizing\u003CT>` wrapper (drop-based)\n- `ZeroizeOnDrop` derive macro\n\n---\n\n## Finding Capabilities\n\nFindings are grouped by required evidence. Only attempt findings for which the required tooling is available.\n\n| Finding ID | Description | Requires | PoC Support |\n|---|---|---|---|\n| `MISSING_SOURCE_ZEROIZE` | No zeroization found in source | Source only | Yes (C\u002FC++ + Rust) |\n| `PARTIAL_WIPE` | Incorrect size or incomplete wipe | Source only | Yes (C\u002FC++ + Rust) |\n| `NOT_ON_ALL_PATHS` | Zeroization missing on some control-flow paths (heuristic) | Source only | Yes (C\u002FC++ only) |\n| `SECRET_COPY` | Sensitive data copied without zeroization tracking | Source + MCP preferred | Yes (C\u002FC++ + Rust) |\n| `INSECURE_HEAP_ALLOC` | Secret uses insecure allocator (malloc vs. secure_malloc) | Source only | Yes (C\u002FC++ only) |\n| `OPTIMIZED_AWAY_ZEROIZE` | Compiler removed zeroization | IR diff required (never source-only) | Yes |\n| `STACK_RETENTION` | Stack frame may retain secrets after return | Assembly required (C\u002FC++); LLVM IR `alloca`+`lifetime.end` evidence (Rust); assembly corroboration upgrades to `confirmed` | Yes (C\u002FC++ only) |\n| `REGISTER_SPILL` | Secrets spilled from registers to stack | Assembly required (C\u002FC++); LLVM IR `load`+call-site evidence (Rust); assembly corroboration upgrades to `confirmed` | Yes (C\u002FC++ only) |\n| `MISSING_ON_ERROR_PATH` | Error-handling paths lack cleanup | CFG or MCP required | Yes |\n| `NOT_DOMINATING_EXITS` | Wipe doesn't dominate all exits | CFG or MCP required | Yes |\n| `LOOP_UNROLLED_INCOMPLETE` | Unrolled loop wipe is incomplete | Semantic IR required | Yes |\n\n---\n\n## Agent Architecture\n\nThe analysis pipeline uses 11 agents across 8 phases, invoked by the orchestrator (`{baseDir}\u002Fprompts\u002Ftask.md`) via `Task`. Agents write persistent finding files to a shared working directory (`\u002Ftmp\u002Fzeroize-audit-{run_id}\u002F`), enabling parallel execution and protecting against context pressure.\n\n| Agent | Phase | Purpose | Output Directory |\n|---|---|---|---|\n| `0-preflight` | Phase 0 | Preflight checks (tools, toolchain, compile DB, crate build), config merge, workdir creation, TU enumeration | `{workdir}\u002F` |\n| `1-mcp-resolver` | Phase 1, Wave 1 (C\u002FC++ only) | Resolve symbols, types, and cross-file references via Serena MCP | `mcp-evidence\u002F` |\n| `2-source-analyzer` | Phase 1, Wave 2a (C\u002FC++ only) | Identify sensitive objects, detect wipes, validate correctness, data-flow\u002Fheap | `source-analysis\u002F` |\n| `2b-rust-source-analyzer` | Phase 1, Wave 2b (Rust only, parallel with 2a) | Rustdoc JSON trait-aware analysis + dangerous API grep | `source-analysis\u002F` |\n| `3-tu-compiler-analyzer` | Phase 2, Wave 3 (C\u002FC++ only, N parallel) | Per-TU IR diff, assembly, semantic IR, CFG analysis | `compiler-analysis\u002F{tu_hash}\u002F` |\n| `3b-rust-compiler-analyzer` | Phase 2, Wave 3R (Rust only, single agent) | Crate-level MIR, LLVM IR, and assembly analysis | `rust-compiler-analysis\u002F` |\n| `4-report-assembler` | Phase 3 (interim) + Phase 6 (final) | Collect findings from all agents, apply confidence gates; merge PoC results and produce final report | `report\u002F` |\n| `5-poc-generator` | Phase 4 | Craft bespoke proof-of-concept programs (C\u002FC++: all categories; Rust: MISSING_SOURCE_ZEROIZE, SECRET_COPY, PARTIAL_WIPE) | `poc\u002F` |\n| `5b-poc-validator` | Phase 5 | Compile and run all PoCs | `poc\u002F` |\n| `5c-poc-verifier` | Phase 5 | Verify each PoC proves its claimed finding | `poc\u002F` |\n| `6-test-generator` | Phase 7 (optional) | Generate runtime validation test harnesses | `tests\u002F` |\n\nThe orchestrator reads one per-phase workflow file from `{baseDir}\u002Fworkflows\u002F` at a time, and maintains `orchestrator-state.json` for recovery after context compression. Agents receive configuration by file path (`config_path`), not by value.\n\n### Execution flow\n\n```\nPhase 0: 0-preflight agent — Preflight + config + create workdir + enumerate TUs\n           → writes orchestrator-state.json, merged-config.yaml, preflight.json\nPhase 1: Wave 1:  1-mcp-resolver              (skip if mcp_mode=off OR language_mode=rust)\n         Wave 2a: 2-source-analyzer           (C\u002FC++ only; skip if no compile_db)  ─┐ parallel\n         Wave 2b: 2b-rust-source-analyzer     (Rust only; skip if no cargo_manifest) ─┘\nPhase 2: Wave 3:  3-tu-compiler-analyzer x N  (C\u002FC++ only; parallel per TU)\n         Wave 3R: 3b-rust-compiler-analyzer   (Rust only; single crate-level agent)\nPhase 3: Wave 4:  4-report-assembler          (mode=interim → findings.json; reads all agent outputs)\nPhase 4: Wave 5:  5-poc-generator             (C\u002FC++: all categories; Rust: MISSING_SOURCE_ZEROIZE, SECRET_COPY, PARTIAL_WIPE; other Rust findings: poc_supported=false)\nPhase 5: PoC Validation & Verification\n           Step 1: 5b-poc-validator agent      (compile and run all PoCs)\n           Step 2: 5c-poc-verifier agent       (verify each PoC proves its claimed finding)\n           Step 3: Orchestrator presents verification failures to user via AskUserQuestion\n           Step 4: Orchestrator merges all results into poc_final_results.json\nPhase 6: Wave 6: 4-report-assembler           (mode=final → merge PoC results, final-report.md)\nPhase 7: Wave 7: 6-test-generator             (optional)\nPhase 8: Orchestrator — Return final-report.md\n```\n\n## Cross-Reference Convention\n\nIDs are namespaced per agent to prevent collisions during parallel execution:\n\n| Entity | Pattern | Assigned By |\n|---|---|---|\n| Sensitive object (C\u002FC++) | `SO-0001`–`SO-4999` | `2-source-analyzer` |\n| Sensitive object (Rust) | `SO-5000`–`SO-9999` (Rust namespace) | `2b-rust-source-analyzer` |\n| Source finding (C\u002FC++) | `F-SRC-NNNN` | `2-source-analyzer` |\n| Source finding (Rust) | `F-RUST-SRC-NNNN` | `2b-rust-source-analyzer` |\n| IR finding (C\u002FC++) | `F-IR-{tu_hash}-NNNN` | `3-tu-compiler-analyzer` |\n| ASM finding (C\u002FC++) | `F-ASM-{tu_hash}-NNNN` | `3-tu-compiler-analyzer` |\n| CFG finding | `F-CFG-{tu_hash}-NNNN` | `3-tu-compiler-analyzer` |\n| Semantic IR finding | `F-SIR-{tu_hash}-NNNN` | `3-tu-compiler-analyzer` |\n| Rust MIR finding | `F-RUST-MIR-NNNN` | `3b-rust-compiler-analyzer` |\n| Rust LLVM IR finding | `F-RUST-IR-NNNN` | `3b-rust-compiler-analyzer` |\n| Rust assembly finding | `F-RUST-ASM-NNNN` | `3b-rust-compiler-analyzer` |\n| Translation unit | `TU-{hash}` | Orchestrator |\n| Final finding | `ZA-NNNN` | `4-report-assembler` |\n\nEvery finding JSON object includes `related_objects`, `related_findings`, and `evidence_files` fields for cross-referencing between agents.\n\n---\n\n## Detection Strategy\n\nAnalysis runs in two phases. For complete step-by-step guidance, see `{baseDir}\u002Freferences\u002Fdetection-strategy.md`.\n\n| Phase | Steps | Findings produced | Required tooling |\n|---|---|---|---|\n| Phase 1 (Source) | 1–6 | `MISSING_SOURCE_ZEROIZE`, `PARTIAL_WIPE`, `NOT_ON_ALL_PATHS`, `SECRET_COPY`, `INSECURE_HEAP_ALLOC` | Source + compile DB |\n| Phase 2 (Compiler) | 7–12 | `OPTIMIZED_AWAY_ZEROIZE`, `STACK_RETENTION`*, `REGISTER_SPILL`*, `LOOP_UNROLLED_INCOMPLETE`†, `MISSING_ON_ERROR_PATH`‡, `NOT_DOMINATING_EXITS`‡ | `clang`, IR\u002FASM tools |\n\n\\* requires `enable_asm=true` (default)\n† requires `enable_semantic_ir=true`\n‡ requires `enable_cfg=true`\n\n---\n\n\n## Output Format\n\nEach run produces two outputs:\n\n1. **`final-report.md`** — Comprehensive markdown report (primary human-readable output)\n2. **`findings.json`** — Structured JSON matching `{baseDir}\u002Fschemas\u002Foutput.json` (for machine consumption and downstream tools)\n\n### Markdown Report Structure\n\nThe markdown report (`final-report.md`) contains these sections:\n\n- **Header**: Run metadata (run_id, timestamp, repo, compile_db, config summary)\n- **Executive Summary**: Finding counts by severity, confidence, and category\n- **Sensitive Objects Inventory**: Table of all identified objects with IDs, types, locations\n- **Findings**: Grouped by severity then confidence. Each finding includes location, object, all evidence (source\u002FIR\u002FASM\u002FCFG), compiler evidence details, and recommended fix\n- **Superseded Findings**: Source findings replaced by CFG-backed findings\n- **Confidence Gate Summary**: Downgrades applied and overrides rejected\n- **Analysis Coverage**: TUs analyzed, agent success\u002Ffailure, features enabled\n- **Appendix: Evidence Files**: Mapping of finding IDs to evidence file paths\n\n### Structured JSON\n\nThe `findings.json` file follows the schema in `{baseDir}\u002Fschemas\u002Foutput.json`. Each `Finding` object:\n\n```json\n{\n  \"id\": \"ZA-0001\",\n  \"category\": \"OPTIMIZED_AWAY_ZEROIZE\",\n  \"severity\": \"high\",\n  \"confidence\": \"confirmed\",\n  \"language\": \"c\",\n  \"file\": \"src\u002Fcrypto.c\",\n  \"line\": 42,\n  \"symbol\": \"key_buf\",\n  \"evidence\": \"store volatile i8 0 count: O0=32, O2=0 — wipe eliminated by DSE\",\n  \"compiler_evidence\": {\n    \"opt_levels\": [\"O0\", \"O2\"],\n    \"o0\": \"32 volatile stores targeting key_buf\",\n    \"o2\": \"0 volatile stores (all eliminated)\",\n    \"diff_summary\": \"All volatile wipe stores removed at O2 — classic DSE pattern\"\n  },\n  \"suggested_fix\": \"Replace memset with explicit_bzero or add compiler_fence(SeqCst) after the wipe\",\n  \"poc\": {\n    \"file\": \"generated_pocs\u002FZA-0001.c\",\n    \"makefile_target\": \"ZA-0001\",\n    \"compile_opt\": \"-O2\",\n    \"requires_manual_adjustment\": false,\n    \"validated\": true,\n    \"validation_result\": \"exploitable\"\n  }\n}\n```\n\nSee `{baseDir}\u002Fschemas\u002Foutput.json` for the full schema and enum values.\n\n---\n\n## Confidence Gating\n\n### Evidence thresholds\n\nA finding requires at least **2 independent signals** to be marked `confirmed`. With 1 signal, mark `likely`. With 0 strong signals (name-pattern match only), mark `needs_review`.\n\nSignals include: name pattern match, type hint match, explicit annotation, IR evidence, ASM evidence, MCP cross-reference, CFG evidence, PoC validation.\n\n### PoC validation as evidence signal\n\nEvery finding is validated against a bespoke PoC. After compilation and execution, each PoC is also verified to ensure it actually tests the claimed vulnerability. The combined result is an evidence signal:\n\n| PoC Result | Verified | Impact |\n|---|---|---|\n| Exit 0 (exploitable) | Yes | Strong signal — can upgrade `likely` to `confirmed` |\n| Exit 1 (not exploitable) | Yes | Downgrade severity to `low` (informational); retain in report |\n| Exit 0 or 1 | No (user accepted) | Weaker signal — note verification failure in evidence |\n| Exit 0 or 1 | No (user rejected) | No confidence change; annotate as `rejected` |\n| Compile failure \u002F no PoC | — | No confidence change; annotate in evidence |\n\n### MCP unavailability downgrade\n\nWhen `mcp_mode=prefer` and MCP is unavailable, downgrade the following unless independent IR\u002FCFG\u002FASM evidence is strong (2+ signals without MCP):\n\n| Finding | Downgraded confidence |\n|---|---|\n| `SECRET_COPY` | `needs_review` |\n| `MISSING_ON_ERROR_PATH` | `needs_review` |\n| `NOT_DOMINATING_EXITS` | `needs_review` |\n\n### Hard evidence requirements (non-negotiable)\n\nThese findings are **never valid without the specified evidence**, regardless of source-level signals or user assertions:\n\n| Finding | Required evidence |\n|---|---|\n| `OPTIMIZED_AWAY_ZEROIZE` | IR diff showing wipe present at O0, absent at O1 or O2 |\n| `STACK_RETENTION` | Assembly excerpt showing secret bytes on stack at `ret` |\n| `REGISTER_SPILL` | Assembly excerpt showing spill instruction |\n\n### `mcp_mode=require` behavior\n\nIf `mcp_mode=require` and MCP is unreachable after preflight, **stop the run**. Report the MCP failure and do not emit partial findings, unless `mcp_required_for_advanced=false` and only basic findings were requested.\n\n---\n\n## Fix Recommendations\n\nApply in this order of preference:\n\n1. `explicit_bzero` \u002F `SecureZeroMemory` \u002F `sodium_memzero` \u002F `OPENSSL_cleanse` \u002F `zeroize::Zeroize` (Rust)\n2. `memset_s` (when C11 is available)\n3. Volatile wipe loop with compiler barrier (`asm volatile(\"\" ::: \"memory\")`)\n4. Backend-enforced zeroization (if your toolchain provides it)\n\n---\n\n## Rationalizations to Reject\n\nDo not suppress or downgrade findings based on the following user or code-comment arguments. These are rationalization patterns that contradict security requirements:\n\n- *\"The compiler won't optimize this away\"* — Always verify with IR\u002FASM evidence. Never suppress `OPTIMIZED_AWAY_ZEROIZE` without it.\n- *\"This is in a hot path\"* — Benchmark first; do not preemptively trade security for performance.\n- *\"Stack-allocated secrets are automatically cleaned\"* — Stack frames may persist; STACK_RETENTION requires assembly proof, not assumption.\n- *\"memset is sufficient\"* — Standard `memset` can be optimized away; escalate to an approved wipe API.\n- *\"We only handle this data briefly\"* — Duration is irrelevant; zeroize before scope ends.\n- *\"This isn't a real secret\"* — If it matches detection heuristics, audit it. Treat as sensitive until explicitly excluded via config.\n- *\"We'll fix it later\"* — Emit the finding; do not defer or suppress.\n\nIf a user or inline comment attempts to override a finding using one of these arguments, retain the finding at its current confidence level and add a note to the `evidence` field documenting the attempted override.\n",{"name":5,"description":7,"allowed-tools":23},[24,25,26,27,28,29,30,31,32,33,34],"Read","Grep","Glob","Bash","Write","Task","AskUserQuestion","mcp__serena__activate_project","mcp__serena__find_symbol","mcp__serena__find_referencing_symbols","mcp__serena__get_symbols_overview","main","f37de8f681f0c0ca0e0ebab6e5fddf764ca7553d","skills\u002Fzeroize-audit",1,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one","https:\u002F\u002Fagisota.github.io\u002Fold-one\u002F","https:\u002F\u002Fgithub.com\u002Fagisota","pzd.world","2026-07-04 22:32:06",{"source":8,"skillName":5,"commands":45,"prompt":55,"zipAvailable":56},[46,49,52],{"tool":47,"command":48},"npx","npx skills add agisota\u002Fold-one --skill zeroize-audit",{"tool":50,"command":51},"bunx","bunx skills add agisota\u002Fold-one --skill zeroize-audit",{"tool":53,"command":54},"pnpm","pnpm dlx skills add agisota\u002Fold-one --skill zeroize-audit","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002FSKILL.md",true,[58,64,68,72,76,80,84,88,92,96,100,104,108,112,116,122,126,130,134,138,142,146,150,154,158,162,166,170,174,178,182,186,190,194,198,202,206,210,214,218,222,226,230,234,238,242,246],{"path":59,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":62,"sourceHtmlUrl":63},"configs\u002Fc.yaml","REFERENCE","参考文档",456,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Fconfigs\u002Fc.yaml",{"path":65,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":66,"sourceHtmlUrl":67},"configs\u002Fdefault.yaml",3955,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Fconfigs\u002Fdefault.yaml",{"path":69,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":70,"sourceHtmlUrl":71},"configs\u002Frust.yaml",3263,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Fconfigs\u002Frust.yaml",{"path":73,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":74,"sourceHtmlUrl":75},"prompts\u002Freport_template.md",6681,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Fprompts\u002Freport_template.md",{"path":77,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":78,"sourceHtmlUrl":79},"prompts\u002Fsystem.md",8128,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Fprompts\u002Fsystem.md",{"path":81,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":82,"sourceHtmlUrl":83},"prompts\u002Ftask.md",4428,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Fprompts\u002Ftask.md",{"path":85,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":86,"sourceHtmlUrl":87},"references\u002Fcompile-commands.md",10160,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Freferences\u002Fcompile-commands.md",{"path":89,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":90,"sourceHtmlUrl":91},"references\u002Fdetection-strategy.md",10187,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Freferences\u002Fdetection-strategy.md",{"path":93,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":94,"sourceHtmlUrl":95},"references\u002Fir-analysis.md",11195,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Freferences\u002Fir-analysis.md",{"path":97,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":98,"sourceHtmlUrl":99},"references\u002Fmcp-analysis.md",9257,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Freferences\u002Fmcp-analysis.md",{"path":101,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":102,"sourceHtmlUrl":103},"references\u002Fpoc-generation.md",21651,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Freferences\u002Fpoc-generation.md",{"path":105,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":106,"sourceHtmlUrl":107},"references\u002Frust-zeroization-patterns.md",35542,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Freferences\u002Frust-zeroization-patterns.md",{"path":109,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":110,"sourceHtmlUrl":111},"schemas\u002Finput.json",3174,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Fschemas\u002Finput.json",{"path":113,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":114,"sourceHtmlUrl":115},"schemas\u002Foutput.json",6040,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Fschemas\u002Foutput.json",{"path":117,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":120,"sourceHtmlUrl":121},"tools\u002Fanalyze_asm.sh","SCRIPT","脚本",5286,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fanalyze_asm.sh",{"path":123,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":124,"sourceHtmlUrl":125},"tools\u002Fanalyze_cfg.py",13056,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fanalyze_cfg.py",{"path":127,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":128,"sourceHtmlUrl":129},"tools\u002Fanalyze_heap.sh",5580,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fanalyze_heap.sh",{"path":131,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":132,"sourceHtmlUrl":133},"tools\u002Fanalyze_ir_semantic.py",15149,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fanalyze_ir_semantic.py",{"path":135,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":136,"sourceHtmlUrl":137},"tools\u002Fdiff_ir.sh",4111,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fdiff_ir.sh",{"path":139,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":140,"sourceHtmlUrl":141},"tools\u002Fdiff_rust_mir.sh",6397,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fdiff_rust_mir.sh",{"path":143,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":144,"sourceHtmlUrl":145},"tools\u002Femit_asm.sh",1026,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Femit_asm.sh",{"path":147,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":148,"sourceHtmlUrl":149},"tools\u002Femit_ir.sh",1340,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Femit_ir.sh",{"path":151,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":152,"sourceHtmlUrl":153},"tools\u002Femit_rust_asm.sh",4337,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Femit_rust_asm.sh",{"path":155,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":156,"sourceHtmlUrl":157},"tools\u002Femit_rust_ir.sh",3411,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Femit_rust_ir.sh",{"path":159,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":160,"sourceHtmlUrl":161},"tools\u002Femit_rust_mir.sh",3536,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Femit_rust_mir.sh",{"path":163,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":164,"sourceHtmlUrl":165},"tools\u002Fextract_compile_flags.py",8782,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fextract_compile_flags.py",{"path":167,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":168,"sourceHtmlUrl":169},"tools\u002Fgenerate_poc.py",48412,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fgenerate_poc.py",{"path":171,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":172,"sourceHtmlUrl":173},"tools\u002Fmcp\u002Fapply_confidence_gates.py",3374,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fmcp\u002Fapply_confidence_gates.py",{"path":175,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":176,"sourceHtmlUrl":177},"tools\u002Fmcp\u002Fcheck_mcp.sh",1251,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fmcp\u002Fcheck_mcp.sh",{"path":179,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":180,"sourceHtmlUrl":181},"tools\u002Fmcp\u002Fnormalize_mcp_evidence.py",3906,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fmcp\u002Fnormalize_mcp_evidence.py",{"path":183,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":184,"sourceHtmlUrl":185},"tools\u002Fscripts\u002Fcheck_llvm_patterns.py",17576,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fscripts\u002Fcheck_llvm_patterns.py",{"path":187,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":188,"sourceHtmlUrl":189},"tools\u002Fscripts\u002Fcheck_mir_patterns.py",19399,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fscripts\u002Fcheck_mir_patterns.py",{"path":191,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":192,"sourceHtmlUrl":193},"tools\u002Fscripts\u002Fcheck_rust_asm_aarch64.py",10246,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fscripts\u002Fcheck_rust_asm_aarch64.py",{"path":195,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":196,"sourceHtmlUrl":197},"tools\u002Fscripts\u002Fcheck_rust_asm_x86.py",9587,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fscripts\u002Fcheck_rust_asm_x86.py",{"path":199,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":200,"sourceHtmlUrl":201},"tools\u002Fscripts\u002Fcheck_rust_asm.py",16243,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fscripts\u002Fcheck_rust_asm.py",{"path":203,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":204,"sourceHtmlUrl":205},"tools\u002Fscripts\u002Ffind_dangerous_apis.py",14132,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fscripts\u002Ffind_dangerous_apis.py",{"path":207,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":208,"sourceHtmlUrl":209},"tools\u002Fscripts\u002Fsemantic_audit.py",33243,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fscripts\u002Fsemantic_audit.py",{"path":211,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":212,"sourceHtmlUrl":213},"tools\u002Ftrack_dataflow.sh",5209,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Ftrack_dataflow.sh",{"path":215,"resourceType":118,"resourceTypeDesc":119,"sizeBytes":216,"sourceHtmlUrl":217},"tools\u002Fvalidate_rust_toolchain.sh",8434,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Ftools\u002Fvalidate_rust_toolchain.sh",{"path":219,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":220,"sourceHtmlUrl":221},"workflows\u002Fphase-0-preflight.md",7476,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Fworkflows\u002Fphase-0-preflight.md",{"path":223,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":224,"sourceHtmlUrl":225},"workflows\u002Fphase-1-source-analysis.md",4814,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Fworkflows\u002Fphase-1-source-analysis.md",{"path":227,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":228,"sourceHtmlUrl":229},"workflows\u002Fphase-2-compiler-analysis.md",5441,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Fworkflows\u002Fphase-2-compiler-analysis.md",{"path":231,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":232,"sourceHtmlUrl":233},"workflows\u002Fphase-3-interim-report.md",1054,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Fworkflows\u002Fphase-3-interim-report.md",{"path":235,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":236,"sourceHtmlUrl":237},"workflows\u002Fphase-4-poc-generation.md",1209,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Fworkflows\u002Fphase-4-poc-generation.md",{"path":239,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":240,"sourceHtmlUrl":241},"workflows\u002Fphase-5-poc-validation.md",4088,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Fworkflows\u002Fphase-5-poc-validation.md",{"path":243,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":244,"sourceHtmlUrl":245},"workflows\u002Fphase-6-final-report.md",1071,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Fworkflows\u002Fphase-6-final-report.md",{"path":247,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":248,"sourceHtmlUrl":249},"workflows\u002Fphase-7-test-generation.md",769,"https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzeroize-audit\u002Fworkflows\u002Fphase-7-test-generation.md",[251,258,267,274,281,288],{"skillNo":252,"name":253,"nameZh":6,"description":254,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":255,"categories":256,"keywords":257,"publishedAt":18},"01M75KRGZWQWE9JDBS6RX42NWV","zoom-out","Tell the agent to zoom out and give broader context or a higher-level perspective. Use when you're unfamiliar with a section of code or need to understand how it fits into the bigger picture.","https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fzoom-out\u002FSKILL.md",[],[],{"skillNo":259,"name":260,"nameZh":6,"description":261,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":262,"licenseFileUrl":263,"sourceHtmlUrl":264,"categories":265,"keywords":266,"publishedAt":18},"01M75KR94KYQ30DTE1P9WNDAE0","yeet","Use only when the user explicitly asks to stage, commit, push, and open a GitHub pull request in one flow using the GitHub CLI (`gh`).","Apache-2.0","https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fyeet\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fyeet\u002FSKILL.md",[],[],{"skillNo":268,"name":269,"nameZh":6,"description":270,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":271,"categories":272,"keywords":273,"publishedAt":18},"01M75KR3H13YJ9QHMD49818FA1","yara-rule-authoring","Guides authoring of high-quality YARA-X detection rules for malware identification. Use when writing, reviewing, or optimizing YARA rules. Covers naming conventions, string selection, performance optimization, migration from legacy YARA, and false positive reduction. Triggers on: YARA, YARA-X, malware detection, threat hunting, IOC, signature, crx module, dex module.","https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fyara-rule-authoring\u002FSKILL.md",[],[],{"skillNo":275,"name":276,"nameZh":6,"description":277,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":278,"categories":279,"keywords":280,"publishedAt":18},"01M75KQMPY8A4Y8XTDD2DT0M9T","x-twitter-growth","X\u002FTwitter growth engine for building audience, crafting viral content, and analyzing engagement. Use when the user wants to grow on X\u002FTwitter, write tweets or threads, analyze their X profile, research competitors on X, plan a posting strategy, or optimize engagement. Complements social-content (generic multi-platform) with X-specific depth: algorithm mechanics, thread engineering, reply strategy, profile optimization, and competitive intelligence via web search.","https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fx-twitter-growth\u002FSKILL.md",[],[],{"skillNo":282,"name":283,"nameZh":6,"description":284,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":285,"categories":286,"keywords":287,"publishedAt":18},"01M75KQHNZ0ANZNJGZY6M8MDAK","x-api","X\u002FTwitter API integration for posting tweets, threads, reading timelines, search, and analytics. Covers OAuth auth patterns, rate limits, and platform-native content posting. Use when the user wants to interact with X programmatically.","https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fx-api\u002FSKILL.md",[],[],{"skillNo":289,"name":290,"nameZh":6,"description":291,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":9,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":292,"categories":293,"keywords":294,"publishedAt":18},"01M75KQEPQAND933F0KWRJMBGR","wycheproof","Wycheproof provides test vectors for validating cryptographic implementations. Use when testing crypto code for known attacks and edge cases.","https:\u002F\u002Fgithub.com\u002Fagisota\u002Fold-one\u002Fblob\u002Fmain\u002Fskills\u002Fwycheproof\u002FSKILL.md",[],[]]