[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7EQJ8YPCCVF6RERTJJ16KDE":3,"skill-same:01M7EQJ8YPCCVF6RERTJJ16KDE":64},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":26,"commitSha":27,"skillDirectory":28,"forksCount":29,"repositoryHtmlUrl":30,"repositoryHomepageUrl":6,"ownerHtmlUrl":31,"ownerBlogUrl":6,"ownerDescription":32,"updatedAt":33,"install":34,"resources":47},"01M7EQJ8YPCCVF6RERTJJ16KDE","post-patch-validation",null,"Validates security patches with reproducible baseline-versus-patched evidence, including original exploits, root-cause variants, behavior preservation, regressions, and newly introduced security failures. Use after a patch exists and before accepting, merging, or reporting it as fixed; also use when an AI-generated patch, remediation commit, pull request, or proposed upstream fix needs adversarial post-patch validation across any language.","lza6\u002FArmor-piercing-system-prompts","lza6","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F128880206?v=4","01M7EB3K7WPX5T28EJ2NVHE6EG","Chinese-tingfeng","01M7EB3K97E1KQ16QWKJMFNGX8",11,"MIT","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F04-安全渗透\u002Fpost-patch-validation\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: post-patch-validation\ndescription: >\n  Validates security patches with reproducible baseline-versus-patched evidence, including\n  original exploits, root-cause variants, behavior preservation, regressions, and newly\n  introduced security failures. Use after a patch exists and before accepting, merging, or\n  reporting it as fixed; also use when an AI-generated patch, remediation commit, pull request,\n  or proposed upstream fix needs adversarial post-patch validation across any language.\nallowed-tools: Read Write Edit Grep Glob Bash Workflow\n---\n\n# Post-Patch Validation\n\nValidate a security patch against the reported bug and the surrounding code it affects. Give the\npatch author reproducible failures to fix and identify what still needs testing. Apply the same\nchecks to human and agent patches. The diff, author, upstream implementation, and original proof\nof concept alone cannot establish correctness.\n\n## When to Use\n\n- A security fix, remediation commit, patch file, or pull request already exists.\n- An AI-generated patch needs validation before human review or merge.\n- A fix may cover one exploit path while missing variants of the same root cause.\n- A security fix may alter legitimate behavior or introduce a new vulnerability.\n- A patch author needs concrete failures and coverage gaps before another revision.\n\n## When NOT to Use\n\n- No patch exists yet; use vulnerability discovery or fix implementation first.\n- The task is to review an audit finding against a report without executing patch evidence.\n- The task is only to convert a finding into a permanent project test.\n- The target is remote or production. This skill executes local code and tests only.\n- The user has not authorized execution of the repository's code or test suite.\n\n## Quick Start\n\n1. Pin the vulnerable base and patched input. Prefer immutable commits. For uncommitted work,\n   create a binary patch file first; do not validate in the user's working tree.\n2. Scaffold a pinned plan:\n\n   ```bash\n   uv run {baseDir}\u002Fscripts\u002Fpost_patch_validation.py scaffold \\\n     --repo . \\\n     --base-ref \u003Cvulnerable-ref> \\\n     --patched-ref \u003Cpatched-ref> \\\n     --finding-id \u003Cstable-id> \\\n     --finding-summary \"\u003Croot cause and impact>\" \\\n     --evidence-level runtime \\\n     --output post-patch-validation\u002Fplan.json\n   ```\n\n   Use `--patch-file \u003Cpath>` instead of `--patched-ref` for a patch artifact. Choose the highest\n   honest evidence level: `source` for source\u002Fpatch invariants only, `build` when target code is\n   compiled or analyzed but the reported behavior is not executed, or `runtime` when the checks\n   execute the reported behavior and its safety assertions.\n3. Inspect the finding, diff, callers, sibling paths, cleanup\u002Ferror paths, and existing tests.\n   Populate `checks` in the generated plan. Run `print-schema` for the structural schema:\n\n   ```bash\n   uv run {baseDir}\u002Fscripts\u002Fpost_patch_validation.py print-schema\n   ```\n\n4. Run `validate-plan` for the complete validation, including coverage, command restrictions,\n   and pinned inputs, before executing code:\n\n   ```bash\n   uv run {baseDir}\u002Fscripts\u002Fpost_patch_validation.py validate-plan \\\n     --plan post-patch-validation\u002Fplan.json\n   ```\n\n5. Execute the evidence plan:\n\n   ```bash\n   uv run {baseDir}\u002Fscripts\u002Fpost_patch_validation.py run \\\n     --plan post-patch-validation\u002Fplan.json \\\n     --output post-patch-validation\u002Fresults\n   ```\n\n6. Report `result.json`, `report.md`, the evidence level, and the complete `assessment`.\n   Return each finding to the patch author with its check ID, assertion, and saved logs. Identify\n   each validation gap separately, including gaps that coexist with supported findings. After\n   the author revises the patch, pin the new inputs and save a fresh validation run. Preserve the\n   prior evidence. Passing supplied checks still requires human review before acceptance.\n\n## Evidence Contract\n\nThe runner rejects incomplete plans. Supply at least one check of every kind:\n\n| Kind | Required observation |\n|---|---|\n| `control` | Benign harness succeeds on both base and patch |\n| `exploit` | Original safety assertion fails on base and succeeds on patch |\n| `variant` | A distinct root-cause variant fails on base and succeeds on patch |\n| `behavior` | Unaffected behavior succeeds with byte-identical selected output |\n| `regression` | Targeted non-security regression check succeeds on both revisions |\n| `security` | Adjacent\u002Fnew-vulnerability check succeeds on base and patch |\n| `suite` | Existing project suite, sanitizer, or deterministic fuzz campaign succeeds on patch |\n\nCommands are argv arrays, never shell strings. Put complex setup in a checked-in or plan artifact\nscript and invoke it with `{plan_dir}`. The runner fixes locale\u002Ftimezone\u002Fhash-seed inputs, executes\nchecks in lexical ID order, records raw stdout\u002Fstderr, and never edits the original worktree.\nEach check's `timeout_seconds` defaults to 300 and accepts integers from 1 through 3600.\nExceeding the timeout leaves a validation gap. A timeout alone does not establish a regression.\nEvery plan also contains a sorted `submodules` array (`[]` when none). Scaffolding infers affected\nGitlinks from the changed-file inventory. The runner initializes those pinned commits from the\nsource repository's existing Git module objects, never from `.gitmodules` network URLs; initialize\nor fetch them in the source repository before validation.\n\n### Exploit and variant checks must prove they ran\n\nA nonzero exit does not mean the vulnerability reproduced. An import error, a failed build, a\nmissing dependency, and a failed safety assertion all exit nonzero and are indistinguishable to the\nrunner. Every `exploit` and `variant` check must print and flush `PPV_REACHED` immediately\nbefore it evaluates its assertion, on both revisions:\n\n```json\n\"argv\": [\"python3\", \"-c\", \"import app; value = app.render('\u003C'); print('PPV_REACHED', flush=True); assert value == '&lt;'\"]\n```\n\nThe token is also in the environment as `PPV_REACHED_MARKER`. It must land on **stdout, as a line\nof its own**. Stderr is not scanned, because a Python `SyntaxError` traceback echoes the offending\nsource and would otherwise satisfy the check for a harness that executed nothing. A run without it\nleaves a `marker_missing` gap. Independent findings from other checks remain in the result.\nFlush explicitly: a harness whose payload segfaults or calls `_exit` loses buffered output and forfeits its own evidence.\n\nThese checks also run **side-blind**. `{side}` is not expanded for them, `PPV_SIDE` is absent from\ntheir environment, the checkout directory is randomly named, and the plan validator rejects any\n`exploit` or `variant` check whose `argv` or `env` mentions either. An assertion that can see which\nrevision it is on can assert on *that* instead of on the code, which is the cheapest possible way\nto fake a reproduction followed by a fix.\n\n### Environment\n\nChecks run under a fixed minimal environment: `PATH`, `HOME`, and a handful of temp\u002Fuser keys,\nplus `LANG`\u002F`LC_ALL=C`, `TZ=UTC`, `PYTHONHASHSEED=0`, `NO_COLOR`, `TERM=dumb`. Everything else in\nthe caller's environment is dropped. Toolchains that need more get it explicitly:\n\n```bash\nuv run {baseDir}\u002Fscripts\u002Fpost_patch_validation.py run \\\n  --plan post-patch-validation\u002Fplan.json \\\n  --output post-patch-validation\u002Fresults \\\n  --allow-env JAVA_HOME --allow-env CARGO_HOME\n```\n\nForwarded names and values are recorded in `result.json`. A requested variable that is unset is an\nerror, not an empty string. Two classes are refused outright: names that read as credentials\n(`*SECRET*`, `*TOKEN*`, `*API_KEY*`, …), because the value would be written into the result; and\nnames that change what executes (`LD_PRELOAD`, `BASH_ENV`, `NODE_OPTIONS`, `GIT_SSH_COMMAND`, …),\nbecause those variables can change which code executes.\nThe runner's fixed variables and every `PPV_*` name are also reserved and cannot be forwarded.\n\nPlaceholders expanded in `argv` and per-check `env` values: `{checkout}` (the revision under test),\n`{plan_dir}` (an isolated copy of the plan artifacts for that one invocation), `{scratch}` (a\nfresh opaque directory for that one check invocation), and\n`{side}` (`base` or `patched`, and not available to `exploit`\u002F`variant` checks). The same values\narrive as `PPV_CHECKOUT`, `PPV_PLAN_DIR`, `PPV_SCRATCH`, `PPV_SIDE`, and `PPV_CASE_ID`. Write only\nunder `{scratch}`; the evidence directory path is not passed to checks. Base and patched invocations\ndo not share runner-managed scratch, plan, or worktree roots. After each invocation exits, its\nscratch tree is archived under the deterministic `results\u002Fscratch\u002F\u003Ccheck-id-and-side>` path, its\nprivate plan copy is discarded, and every readable argv element that resolves to a file is hashed\nin `argv_files`. Files inside the isolated plan or checkout roots are additionally retained under\n`results\u002Fhelpers\u002F\u003Csha256>` up to 16 MiB; the record explains why any other file was not archived.\nUse a dedicated directory for `plan.json`: its sibling files and directories are copied into each\ninvocation's `{plan_dir}`. Keep helper code under that directory's `checks\u002F` directory or checked\ninto the target repository so its bytes are reviewable. The machine plan containing commit pins,\nthe current output directory, and detected prior result trees are excluded; symlinks are rejected.\nThe clean snapshot remains only in runner memory, and\nexploit\u002Fvariant sides execute in random order while evidence filenames remain deterministic.\nStdout\u002Fstderr use anonymous or randomly named capture descriptors and are copied to the named\nevidence files only after the child exits, so fd inspection cannot disclose the side label.\n\nThis isolation is not a host sandbox: checks run with the caller's privileges and a malicious\nhelper could use arbitrary external state or deliberately infer the revision from source or Git\nmetadata. Inspect the content-addressed helper artifacts, and use an OS\u002Fcontainer sandbox when the\ncheck code itself is untrusted.\n\nActive validation worktrees are Git-locked with random owner tokens backed by kernel file locks, so\nanother concurrent validator cannot prune them and PID reuse cannot impersonate an owner. If the\nrunner is forcibly killed, the next run unlocks stale validator-owned registrations. For manual\nrecovery, inspect `git worktree list`, then use `git worktree unlock \u003Cpath>` and\n`git worktree remove --force \u003Cpath>` (or `git worktree prune` after the path is gone).\n\nRead [evidence-model.md](references\u002Fevidence-model.md) when designing coverage, selecting\nvariants, or interpreting findings and validation gaps. Do not read it for routine CLI execution.\n\n## Coverage Rules\n\n- Derive variants from the root cause, not cosmetic mutations of the original payload.\n- Enumerate sibling call sites, alternate callbacks\u002Foutputs, error paths, teardown, ownership,\n  serialization, and boundary values touched by the fix.\n- Make each exploit or variant test assert the safe behavior. It must fail on the vulnerable\n  base; a test that passes on both revisions proves nothing about remediation. Read the base-side\n  stderr and confirm the failure is the assertion you wrote, not a harness that never got there.\n- Keep exploit and variant assertions limited to the security invariant. Test liveness, exact\n  error types\u002Fmessages, timing, and compatibility separately as `behavior` or `regression` checks;\n  otherwise an unrelated contract change can masquerade as proof that the vulnerability remains.\n- Keep the `control` harness benign and make it exercise the changed component. It\n  establishes that the harness works on both revisions. Failed controls leave gaps and prevent\n  attributing other failures to the patch. The raw observations remain available for review.\n- Use `behavior` only for behavior that should remain unchanged. Exact output comparison is\n  deliberate; move unstable values behind a deterministic test harness instead of normalizing\n  them away in prose.\n- Make `security` checks pass on the vulnerable base before treating a patched failure as newly\n  introduced. A failed baseline leaves attribution unresolved.\n- Do not edit the patch during validation. Return failures to the patch author and start a new,\n  freshly pinned run.\n\n## Reading the result\n\n`result.json` schema 2.0 contains an `assessment` with `status`, `findings`, `gaps`, and\n`human_review_required`. Status is `complete` when all required checks produced usable evidence,\neven if some checks found failures. Status is `incomplete` when any gap remains. Findings name the\ncheck ID, check kind, and failed expectation. Gaps name the check ID and missing evidence, with a\nnull check ID for run-level problems such as cleanup failures.\n\nSuite checks run on the patched revision first. A completed failure triggers the same check on\nbaseline. If baseline passes, report the failure after the patch. If both fail, preserve both logs\nand report that attribution is unresolved. Do not assume matching exit codes mean the same failure.\n\nThe runner exits 0 for complete checks with no findings, 1 for complete checks with findings,\n10 for incomplete validation, and 64 for invalid inputs. Read the artifact even after exit 10:\nit can contain supported findings alongside gaps. Source or build evidence cannot establish\nruntime behavior. State the evidence level alongside any passing result.\n\n## Claude Dynamic Workflow\n\nClaude Code exposes the bundled workflow as `\u002Fpost-patch-validation:validate-patch`.\nTo pass structured inputs through the `Workflow` tool, use the name without a leading slash:\n\n```javascript\nWorkflow({\n  name: 'post-patch-validation:validate-patch',\n  args: {\n    finding: '\u003Cfinding text or local path>',\n    baseRef: '\u003Cvulnerable-ref>',\n    patchRef: '\u003Cpatched-ref>',\n    workdir: 'post-patch-validation',\n  },\n})\n```\n\nUse `patchFile` instead of `patchRef` when appropriate. The workflow uses fixed coverage\nlenses to propose checks and a fixed executor to run this skill. Agents may author test\nartifacts. The Python runner records findings and gaps, and reviewers report coverage or evidence\nconcerns separately. `NEEDS_REPAIR` returns supported failures even when other checks left gaps.\n`BLOCKED` means evidence is incomplete without a supported failure. Passing checks advance to\n`READY_FOR_HUMAN_REVIEW` only after both evidence reviews approve, otherwise `REVIEW_REQUIRED`.\nThe workflow cannot ask questions after launch, so pass every input up front.\n\n## Rationalizations to Reject\n\n| Rationalization | Required response |\n|---|---|\n| \"The original PoC no longer works\" | Test at least one independent root-cause variant |\n| \"The exploit failed on base, so it reproduced\" | Confirm the marker and that the failure is the assertion, not a broken harness |\n| \"The full suite passes\" | Prove baseline reproduction and targeted behavior explicitly |\n| \"This matches the upstream\u002Fcanonical patch\" | Treat provenance as context, not evidence |\n| \"The diff is tiny\" | Exercise callers, failure paths, and teardown affected by the change |\n| \"All supplied checks passed\" | Preserve artifacts and require human review |\n| \"A flaky rerun passed\" | Keep the first pinned result; fix nondeterminism before retrying |\n| \"There is no obvious variant\" | Inspect sibling sites and boundaries; otherwise report the missing coverage |\n","# Post-Patch Validation\n\nValidate a security patch against the reported bug and the surrounding code it affects. Give the\npatch author reproducible failures to fix and identify what still needs testing. Apply the same\nchecks to human and agent patches. The diff, author, upstream implementation, and original proof\nof concept alone cannot establish correctness.\n\n## When to Use\n\n- A security fix, remediation commit, patch file, or pull request already exists.\n- An AI-generated patch needs validation before human review or merge.\n- A fix may cover one exploit path while missing variants of the same root cause.\n- A security fix may alter legitimate behavior or introduce a new vulnerability.\n- A patch author needs concrete failures and coverage gaps before another revision.\n\n## When NOT to Use\n\n- No patch exists yet; use vulnerability discovery or fix implementation first.\n- The task is to review an audit finding against a report without executing patch evidence.\n- The task is only to convert a finding into a permanent project test.\n- The target is remote or production. This skill executes local code and tests only.\n- The user has not authorized execution of the repository's code or test suite.\n\n## Quick Start\n\n1. Pin the vulnerable base and patched input. Prefer immutable commits. For uncommitted work,\n   create a binary patch file first; do not validate in the user's working tree.\n2. Scaffold a pinned plan:\n\n   ```bash\n   uv run {baseDir}\u002Fscripts\u002Fpost_patch_validation.py scaffold \\\n     --repo . \\\n     --base-ref \u003Cvulnerable-ref> \\\n     --patched-ref \u003Cpatched-ref> \\\n     --finding-id \u003Cstable-id> \\\n     --finding-summary \"\u003Croot cause and impact>\" \\\n     --evidence-level runtime \\\n     --output post-patch-validation\u002Fplan.json\n   ```\n\n   Use `--patch-file \u003Cpath>` instead of `--patched-ref` for a patch artifact. Choose the highest\n   honest evidence level: `source` for source\u002Fpatch invariants only, `build` when target code is\n   compiled or analyzed but the reported behavior is not executed, or `runtime` when the checks\n   execute the reported behavior and its safety assertions.\n3. Inspect the finding, diff, callers, sibling paths, cleanup\u002Ferror paths, and existing tests.\n   Populate `checks` in the generated plan. Run `print-schema` for the structural schema:\n\n   ```bash\n   uv run {baseDir}\u002Fscripts\u002Fpost_patch_validation.py print-schema\n   ```\n\n4. Run `validate-plan` for the complete validation, including coverage, command restrictions,\n   and pinned inputs, before executing code:\n\n   ```bash\n   uv run {baseDir}\u002Fscripts\u002Fpost_patch_validation.py validate-plan \\\n     --plan post-patch-validation\u002Fplan.json\n   ```\n\n5. Execute the evidence plan:\n\n   ```bash\n   uv run {baseDir}\u002Fscripts\u002Fpost_patch_validation.py run \\\n     --plan post-patch-validation\u002Fplan.json \\\n     --output post-patch-validation\u002Fresults\n   ```\n\n6. Report `result.json`, `report.md`, the evidence level, and the complete `assessment`.\n   Return each finding to the patch author with its check ID, assertion, and saved logs. Identify\n   each validation gap separately, including gaps that coexist with supported findings. After\n   the author revises the patch, pin the new inputs and save a fresh validation run. Preserve the\n   prior evidence. Passing supplied checks still requires human review before acceptance.\n\n## Evidence Contract\n\nThe runner rejects incomplete plans. Supply at least one check of every kind:\n\n| Kind | Required observation |\n|---|---|\n| `control` | Benign harness succeeds on both base and patch |\n| `exploit` | Original safety assertion fails on base and succeeds on patch |\n| `variant` | A distinct root-cause variant fails on base and succeeds on patch |\n| `behavior` | Unaffected behavior succeeds with byte-identical selected output |\n| `regression` | Targeted non-security regression check succeeds on both revisions |\n| `security` | Adjacent\u002Fnew-vulnerability check succeeds on base and patch |\n| `suite` | Existing project suite, sanitizer, or deterministic fuzz campaign succeeds on patch |\n\nCommands are argv arrays, never shell strings. Put complex setup in a checked-in or plan artifact\nscript and invoke it with `{plan_dir}`. The runner fixes locale\u002Ftimezone\u002Fhash-seed inputs, executes\nchecks in lexical ID order, records raw stdout\u002Fstderr, and never edits the original worktree.\nEach check's `timeout_seconds` defaults to 300 and accepts integers from 1 through 3600.\nExceeding the timeout leaves a validation gap. A timeout alone does not establish a regression.\nEvery plan also contains a sorted `submodules` array (`[]` when none). Scaffolding infers affected\nGitlinks from the changed-file inventory. The runner initializes those pinned commits from the\nsource repository's existing Git module objects, never from `.gitmodules` network URLs; initialize\nor fetch them in the source repository before validation.\n\n### Exploit and variant checks must prove they ran\n\nA nonzero exit does not mean the vulnerability reproduced. An import error, a failed build, a\nmissing dependency, and a failed safety assertion all exit nonzero and are indistinguishable to the\nrunner. Every `exploit` and `variant` check must print and flush `PPV_REACHED` immediately\nbefore it evaluates its assertion, on both revisions:\n\n```json\n\"argv\": [\"python3\", \"-c\", \"import app; value = app.render('\u003C'); print('PPV_REACHED', flush=True); assert value == '&lt;'\"]\n```\n\nThe token is also in the environment as `PPV_REACHED_MARKER`. It must land on **stdout, as a line\nof its own**. Stderr is not scanned, because a Python `SyntaxError` traceback echoes the offending\nsource and would otherwise satisfy the check for a harness that executed nothing. A run without it\nleaves a `marker_missing` gap. Independent findings from other checks remain in the result.\nFlush explicitly: a harness whose payload segfaults or calls `_exit` loses buffered output and forfeits its own evidence.\n\nThese checks also run **side-blind**. `{side}` is not expanded for them, `PPV_SIDE` is absent from\ntheir environment, the checkout directory is randomly named, and the plan validator rejects any\n`exploit` or `variant` check whose `argv` or `env` mentions either. An assertion that can see which\nrevision it is on can assert on *that* instead of on the code, which is the cheapest possible way\nto fake a reproduction followed by a fix.\n\n### Environment\n\nChecks run under a fixed minimal environment: `PATH`, `HOME`, and a handful of temp\u002Fuser keys,\nplus `LANG`\u002F`LC_ALL=C`, `TZ=UTC`, `PYTHONHASHSEED=0`, `NO_COLOR`, `TERM=dumb`. Everything else in\nthe caller's environment is dropped. Toolchains that need more get it explicitly:\n\n```bash\nuv run {baseDir}\u002Fscripts\u002Fpost_patch_validation.py run \\\n  --plan post-patch-validation\u002Fplan.json \\\n  --output post-patch-validation\u002Fresults \\\n  --allow-env JAVA_HOME --allow-env CARGO_HOME\n```\n\nForwarded names and values are recorded in `result.json`. A requested variable that is unset is an\nerror, not an empty string. Two classes are refused outright: names that read as credentials\n(`*SECRET*`, `*TOKEN*`, `*API_KEY*`, …), because the value would be written into the result; and\nnames that change what executes (`LD_PRELOAD`, `BASH_ENV`, `NODE_OPTIONS`, `GIT_SSH_COMMAND`, …),\nbecause those variables can change which code executes.\nThe runner's fixed variables and every `PPV_*` name are also reserved and cannot be forwarded.\n\nPlaceholders expanded in `argv` and per-check `env` values: `{checkout}` (the revision under test),\n`{plan_dir}` (an isolated copy of the plan artifacts for that one invocation), `{scratch}` (a\nfresh opaque directory for that one check invocation), and\n`{side}` (`base` or `patched`, and not available to `exploit`\u002F`variant` checks). The same values\narrive as `PPV_CHECKOUT`, `PPV_PLAN_DIR`, `PPV_SCRATCH`, `PPV_SIDE`, and `PPV_CASE_ID`. Write only\nunder `{scratch}`; the evidence directory path is not passed to checks. Base and patched invocations\ndo not share runner-managed scratch, plan, or worktree roots. After each invocation exits, its\nscratch tree is archived under the deterministic `results\u002Fscratch\u002F\u003Ccheck-id-and-side>` path, its\nprivate plan copy is discarded, and every readable argv element that resolves to a file is hashed\nin `argv_files`. Files inside the isolated plan or checkout roots are additionally retained under\n`results\u002Fhelpers\u002F\u003Csha256>` up to 16 MiB; the record explains why any other file was not archived.\nUse a dedicated directory for `plan.json`: its sibling files and directories are copied into each\ninvocation's `{plan_dir}`. Keep helper code under that directory's `checks\u002F` directory or checked\ninto the target repository so its bytes are reviewable. The machine plan containing commit pins,\nthe current output directory, and detected prior result trees are excluded; symlinks are rejected.\nThe clean snapshot remains only in runner memory, and\nexploit\u002Fvariant sides execute in random order while evidence filenames remain deterministic.\nStdout\u002Fstderr use anonymous or randomly named capture descriptors and are copied to the named\nevidence files only after the child exits, so fd inspection cannot disclose the side label.\n\nThis isolation is not a host sandbox: checks run with the caller's privileges and a malicious\nhelper could use arbitrary external state or deliberately infer the revision from source or Git\nmetadata. Inspect the content-addressed helper artifacts, and use an OS\u002Fcontainer sandbox when the\ncheck code itself is untrusted.\n\nActive validation worktrees are Git-locked with random owner tokens backed by kernel file locks, so\nanother concurrent validator cannot prune them and PID reuse cannot impersonate an owner. If the\nrunner is forcibly killed, the next run unlocks stale validator-owned registrations. For manual\nrecovery, inspect `git worktree list`, then use `git worktree unlock \u003Cpath>` and\n`git worktree remove --force \u003Cpath>` (or `git worktree prune` after the path is gone).\n\nRead [evidence-model.md](references\u002Fevidence-model.md) when designing coverage, selecting\nvariants, or interpreting findings and validation gaps. Do not read it for routine CLI execution.\n\n## Coverage Rules\n\n- Derive variants from the root cause, not cosmetic mutations of the original payload.\n- Enumerate sibling call sites, alternate callbacks\u002Foutputs, error paths, teardown, ownership,\n  serialization, and boundary values touched by the fix.\n- Make each exploit or variant test assert the safe behavior. It must fail on the vulnerable\n  base; a test that passes on both revisions proves nothing about remediation. Read the base-side\n  stderr and confirm the failure is the assertion you wrote, not a harness that never got there.\n- Keep exploit and variant assertions limited to the security invariant. Test liveness, exact\n  error types\u002Fmessages, timing, and compatibility separately as `behavior` or `regression` checks;\n  otherwise an unrelated contract change can masquerade as proof that the vulnerability remains.\n- Keep the `control` harness benign and make it exercise the changed component. It\n  establishes that the harness works on both revisions. Failed controls leave gaps and prevent\n  attributing other failures to the patch. The raw observations remain available for review.\n- Use `behavior` only for behavior that should remain unchanged. Exact output comparison is\n  deliberate; move unstable values behind a deterministic test harness instead of normalizing\n  them away in prose.\n- Make `security` checks pass on the vulnerable base before treating a patched failure as newly\n  introduced. A failed baseline leaves attribution unresolved.\n- Do not edit the patch during validation. Return failures to the patch author and start a new,\n  freshly pinned run.\n\n## Reading the result\n\n`result.json` schema 2.0 contains an `assessment` with `status`, `findings`, `gaps`, and\n`human_review_required`. Status is `complete` when all required checks produced usable evidence,\neven if some checks found failures. Status is `incomplete` when any gap remains. Findings name the\ncheck ID, check kind, and failed expectation. Gaps name the check ID and missing evidence, with a\nnull check ID for run-level problems such as cleanup failures.\n\nSuite checks run on the patched revision first. A completed failure triggers the same check on\nbaseline. If baseline passes, report the failure after the patch. If both fail, preserve both logs\nand report that attribution is unresolved. Do not assume matching exit codes mean the same failure.\n\nThe runner exits 0 for complete checks with no findings, 1 for complete checks with findings,\n10 for incomplete validation, and 64 for invalid inputs. Read the artifact even after exit 10:\nit can contain supported findings alongside gaps. Source or build evidence cannot establish\nruntime behavior. State the evidence level alongside any passing result.\n\n## Claude Dynamic Workflow\n\nClaude Code exposes the bundled workflow as `\u002Fpost-patch-validation:validate-patch`.\nTo pass structured inputs through the `Workflow` tool, use the name without a leading slash:\n\n```javascript\nWorkflow({\n  name: 'post-patch-validation:validate-patch',\n  args: {\n    finding: '\u003Cfinding text or local path>',\n    baseRef: '\u003Cvulnerable-ref>',\n    patchRef: '\u003Cpatched-ref>',\n    workdir: 'post-patch-validation',\n  },\n})\n```\n\nUse `patchFile` instead of `patchRef` when appropriate. The workflow uses fixed coverage\nlenses to propose checks and a fixed executor to run this skill. Agents may author test\nartifacts. The Python runner records findings and gaps, and reviewers report coverage or evidence\nconcerns separately. `NEEDS_REPAIR` returns supported failures even when other checks left gaps.\n`BLOCKED` means evidence is incomplete without a supported failure. Passing checks advance to\n`READY_FOR_HUMAN_REVIEW` only after both evidence reviews approve, otherwise `REVIEW_REQUIRED`.\nThe workflow cannot ask questions after launch, so pass every input up front.\n\n## Rationalizations to Reject\n\n| Rationalization | Required response |\n|---|---|\n| \"The original PoC no longer works\" | Test at least one independent root-cause variant |\n| \"The exploit failed on base, so it reproduced\" | Confirm the marker and that the failure is the assertion, not a broken harness |\n| \"The full suite passes\" | Prove baseline reproduction and targeted behavior explicitly |\n| \"This matches the upstream\u002Fcanonical patch\" | Treat provenance as context, not evidence |\n| \"The diff is tiny\" | Exercise callers, failure paths, and teardown affected by the change |\n| \"All supplied checks passed\" | Preserve artifacts and require human review |\n| \"A flaky rerun passed\" | Keep the first pinned result; fix nondeterminism before retrying |\n| \"There is no obvious variant\" | Inspect sibling sites and boundaries; otherwise report the missing coverage |\n",{"name":5,"description":24,"allowed-tools":25},"Validates security patches with reproducible baseline-versus-patched evidence, including original exploits, root-cause variants, behavior preservation, regressions, and newly introduced security failures. Use after a patch exists and before accepting, merging, or reporting it as fixed; also use when an AI-generated patch, remediation commit, pull request, or proposed upstream fix needs adversarial post-patch validation across any language.\n","Read Write Edit Grep Glob Bash Workflow","main","d69096e239f2a89aabc2cd078c47664a9d3b1e11","04-安全渗透\u002Fpost-patch-validation",3,"https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts","https:\u002F\u002Fgithub.com\u002Flza6","个人vx：Tf00798\r\n个人✈：@TFGY999","2026-10-08 14:29:22",{"source":8,"skillName":5,"commands":35,"prompt":45,"zipAvailable":46},[36,39,42],{"tool":37,"command":38},"npx","npx skills add lza6\u002FArmor-piercing-system-prompts --skill post-patch-validation",{"tool":40,"command":41},"bunx","bunx skills add lza6\u002FArmor-piercing-system-prompts --skill post-patch-validation",{"tool":43,"command":44},"pnpm","pnpm dlx skills add lza6\u002FArmor-piercing-system-prompts --skill post-patch-validation","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F04-安全渗透\u002Fpost-patch-validation\u002FSKILL.md",true,[48,54,58],{"path":49,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":52,"sourceHtmlUrl":53},"agents\u002Fopenai.yaml","REFERENCE","参考文档",244,"https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F04-安全渗透\u002Fpost-patch-validation\u002Fagents\u002Fopenai.yaml",{"path":55,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":56,"sourceHtmlUrl":57},"references\u002Fevidence-model.md",12305,"https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F04-安全渗透\u002Fpost-patch-validation\u002Freferences\u002Fevidence-model.md",{"path":59,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":62,"sourceHtmlUrl":63},"scripts\u002Fpyproject.toml","SCRIPT","脚本",116,"https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F04-安全渗透\u002Fpost-patch-validation\u002Fscripts\u002Fpyproject.toml",[65,73,78,85,92,99],{"skillNo":66,"name":67,"nameZh":6,"description":68,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":72},"01M7FFG485KY83TD6HSC4YPCMC","obsidian-clipper-template-creator","Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobsidian-clipper-template-creator\u002FSKILL.md",[],[],"2026-11-16 10:05:34",{"skillNo":74,"name":67,"nameZh":6,"description":68,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":75,"categories":76,"keywords":77,"publishedAt":72},"01M7FFG0T6AZ0STEJPSB3CX84K","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobsidian-clipper-template-creator-antigravity-awesome-skills-main\u002FSKILL.md",[],[],{"skillNo":79,"name":80,"nameZh":6,"description":81,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":82,"categories":83,"keywords":84,"publishedAt":72},"01M7FFFY0QEM14SKWPDYGMXM2P","python-api","Programmatic Object-Detection-Metrics API guidance for BoundingBox, BoundingBoxes, Evaluator, enums, coordinate conversion, AP\u002FmAP semantics, and source-style import diagnostics.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobject-detection-metrics\u002Fsub-skills\u002Fpython-api\u002FSKILL.md",[],[],{"skillNo":86,"name":87,"nameZh":6,"description":88,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":89,"categories":90,"keywords":91,"publishedAt":72},"01M7FFFVZQZ2GJ0C0TGP7JHD20","file-evaluation","Folder-based PASCAL VOC AP\u002FmAP evaluation from ground-truth and detection text files.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobject-detection-metrics\u002Fsub-skills\u002Ffile-evaluation\u002FSKILL.md",[],[],{"skillNo":93,"name":94,"nameZh":6,"description":95,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":96,"categories":97,"keywords":98,"publishedAt":72},"01M7FFFSFNYSE77SJ6PE1CK2DA","object-detection-metrics","Guides legacy Object-Detection-Metrics PASCAL VOC AP\u002FmAP evaluation from detection text folders and direct Python API classes.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fobject-detection-metrics\u002FSKILL.md",[],[],{"skillNo":100,"name":101,"nameZh":6,"description":102,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":103,"categories":104,"keywords":105,"publishedAt":72},"01M7FFFP0GCK1RVXBCPQ8APBMT","nx-workspace-patterns","Configure and optimize Nx monorepo workspaces. Use when setting up Nx, configuring project boundaries, optimizing build caching, or implementing affected commands.","https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F07-后端与架构\u002Fnx-workspace-patterns\u002FSKILL.md",[],[]]