[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7EGTTAK7JK6MTN1R4FWRTYS":3,"skill-same:01M7EGTTAK7JK6MTN1R4FWRTYS":52},{"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":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":6,"ownerHtmlUrl":29,"ownerBlogUrl":6,"ownerDescription":30,"updatedAt":31,"install":32,"resources":45},"01M7EGTTAK7JK6MTN1R4FWRTYS","customize",null,"Adapt this C\u002FC++ ASAN vulnerability pipeline to a different vulnerability class, target shape, language, or detection mechanism. Use when the user wants to port, migrate, retarget, customize, or fork the pipeline for something other than C\u002FC++ memory-safety bugs — web apps, smart contracts, deserialization, ML systems, or any other domain.","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\u002Fcustomize\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: customize\ndescription: Adapt this C\u002FC++ ASAN vulnerability pipeline to a different vulnerability class, target shape, language, or detection mechanism. Use when the user wants to port, migrate, retarget, customize, or fork the pipeline for something other than C\u002FC++ memory-safety bugs — web apps, smart contracts, deserialization, ML systems, or any other domain.\n---\n\n# Customize the vuln-pipeline\n\nThis pipeline ships as an opinionated C\u002FC++ + AddressSanitizer demo. Its real shape is more general: **an agent crafts an input, runs a target in a sandbox, a detector fires, a second agent verifies, a third agent analyzes exploitability.** Every noun in that sentence can be swapped. Your job is to interview the user, figure out which nouns they want to swap, and rewrite the relevant files.\n\nThe existing C\u002FC++ code is the worked example. You don't need a playbook for each domain — read what's there, understand what's generic vs. ASAN-specific, and adapt.\n\n## STEP 1 — Read the pipeline (do this BEFORE asking anything)\n\nSkim these files so your questions are grounded:\n\n- `README.md` — pipeline overview (recon → find → grade → judge → report)\n- `harness\u002Fcli.py` — orchestration; shows how stages wire together and what lands on disk\n- `harness\u002Ffind.py`, `harness\u002Fgrade.py`, `harness\u002Freport.py` — the three container-agent loops; mostly generic plumbing\n- `harness\u002Fprompts\u002Ffind_prompt.py`, `harness\u002Fprompts\u002Fgrade_prompt.py` — **the C\u002FC++-specific parts**; bug taxonomy, quality tiers, grading rubric\n- `harness\u002Fprompts\u002Freport_prompt.py`, `harness\u002Fprompts\u002Freport_grader_prompt.py` — **also C\u002FC++-specific**; exploitability sections (primitive, heap layout, escalation path) and the rubric that scores them\n- `harness\u002Fprompts\u002Fjudge_prompt.py` — triage prompt; keys on ASAN excerpts and memory-safety crash classes\n- `harness\u002Fprompts\u002Fsystem_prompt.py` — authorization block; hard-codes \"C\u002FC++ target\" and \"sanitizer output\"\n- `harness\u002Fasan.py` — stack-trace parser for dedup\u002Fjudge signatures; ASAN-specific regex\n- `harness\u002Fartifacts.py` — `CrashArtifact`, `GraderVerdict`, `JudgeVerdict`, `ReportVerdict` data contracts\n- `harness\u002Fconfig.py`, `targets\u002Fdrlibs\u002Fconfig.yaml` — target config schema\n- `targets\u002FREADME.md` — how a target directory is structured (Dockerfile + config.yaml + entry wrapper)\n\nYou don't need `agent.py`, `docker_ops.py`, `recon.py`, `judge.py`, or `novelty.py` in detail — they're generic plumbing (judge\u002Fnovelty domain-specificity lives in the prompts and the asan parser, not the flow). One caution: part of `harness\u002F` doubles as shared infrastructure — `dnr_harness\u002F` (the detection-and-response pipeline) imports `agent.py`, `sandbox.py`, `docker_ops.py`, `agent_image.py`'s `build`\u002F`ensure_base`, `auth.py`, `cli.py`'s `resolve_target_dir`\u002F`terminate_subprocesses`, and `prompts\u002Fsystem_prompt.py`'s `load_engagement_context`. A domain port should touch only the domain-specific files (the prompts, `asan.py`, `artifacts.py`, target Dockerfiles); treat any edit to the shared files as a change to both pipelines.\n\n## STEP 2 — Interview the user\n\nUse **AskUserQuestion** to gather requirements. Start with broad context, then narrow to technical specifics based on what they say.\n\n### Round 1 — Context (always ask these first, together)\n\nTwo open-ended questions to understand who you're talking to and what they're after. Expect most answers to come via **Other** as free text — the options are there to prompt thinking, not to constrain.\n\n**Question A — Operating context**\n- header: `Context`\n- question: `What's your operating environment? Who will run this pipeline and why?`\n- options: a few archetypes as inspiration — e.g. \"Pentesting firm — client engagements, need reportable findings\", \"Internal appsec team — scan our own services in CI\", \"Smart-contract auditor — pre-deployment reviews\", \"Security researcher — hunting novel bug classes\". These tell you what output format, grading rigor, and workflow integration matter.\n\n**Question B — Goal**\n- header: `Goal`\n- question: `Describe in your own words what you want this pipeline to find. What kind of target, what kind of bugs?`\n- options: 2–3 concrete examples (e.g. \"Web vulnerabilities like SQLi\u002FXSS in HTTP services\", \"Reentrancy and access-control bugs in Solidity contracts\", \"Deserialization RCE in Java microservices\").\n\nThe context answer calibrates your follow-ups: a pentesting firm probably cares about CVSS scoring and SARIF output; a researcher may want differential testing and novel detection signals; an internal team likely wants CI integration and low false-positive rates.\n\n### Round 2 — Technical follow-ups (adaptive — derive from round-1 answers)\n\nParse their round-1 answers against the **axes of variation** below. For each axis left ambiguous, ask a targeted follow-up. Batch up to 4 questions per AskUserQuestion call. Common follow-ups:\n\n- **Detection signal** — \"How will the pipeline know it found something?\" (crash, exception, canary file appears, DNS callback, differential mismatch, invariant violation)\n- **PoC shape** — \"What does a proof-of-concept look like?\" (single file, HTTP request sequence, transaction list, test-pipeline code)\n- **Isolation** — \"Where does the target run?\" (Docker, VM, testnet, remote sandbox, or no execution — static-only)\n- **Grading criteria** — \"What makes a finding high-quality vs. low-quality in this domain?\"\n- **Exploitability analysis** — \"What sections should a report contain?\" The C\u002FC++ report has primitive · reachability · heap layout · escalation path · constraints. A web-vuln report might want injection vector · auth bypass · data exposure · chaining potential. Ask what they need, or whether they want the report stage at all.\n- **Novelty\u002Fupstream check** — \"Should the pipeline check if a finding is already fixed upstream?\" The C\u002FC++ version shallow-clones the target's GitHub and checks `git log \u003Ccommit>..HEAD -- \u003Ccrash_file>`. Only applies if targets have a canonical upstream and a sensible \"crashing file\" to key on — many domains won't.\n- **Scope** — \"Replace the C\u002FC++ support entirely, or keep it alongside the new domain via a profile system?\"\n\nKeep going until you can fill in every row of the architecture map in STEP 3. If an answer is vague, ask a narrower follow-up rather than guessing.\n\n## Background — axes of variation (context for formulating follow-ups)\n\nThese are the dimensions along which customers might want to deviate from the C\u002FC++ demo. Use this list to spot gaps in the user's description and generate follow-up questions — do **not** present it as a menu.\n\n**Vulnerability class:** memory safety · web\u002FAPI (SQLi, XSS, SSRF, XXE, path traversal, IDOR) · deserialization RCE · logic\u002Frace (TOCTOU, privilege escalation) · crypto (weak RNG, timing, nonce reuse) · DoS (ReDoS, hash flooding) · smart contracts (reentrancy, access control, front-running) · ML\u002FAI (prompt injection, jailbreaks, data extraction) · protocol parsing\n\n**Target shape:** CLI binary + file · HTTP service · library via test harness · network daemon · smart contract · browser extension · mobile app\n\n**Detection mechanism:** crash\u002Fabort · uncaught exception · sanitizer hooks (Jazzer\u002FAtheris) · outcome-based (canary file, DNS callback, shell spawn) · differential testing · invariant violation · taint tracking\n\n**Input modality:** single file · HTTP request chain · multi-file archive · stdin stream · args + env + config combo · transaction sequence\n\n**Isolation boundary:** Docker container · full VM · remote sandbox · local testnet · none (static analysis)\n\n**Dedup signature:** (crash_type, top_frame) · (vuln_type, endpoint, param) · (function, state_transition) · (component, precondition)\n\n**Report structure:** primitive\u002Fheap\u002Fescalation (memory safety) · vector\u002Fauth\u002Fexposure (web) · invariant\u002Fpath\u002Fimpact (contracts) · or drop the report stage entirely if find+grade is the deliverable\n\n**Output format:** result.json + poc.bin · SARIF · Nuclei template · prose report\n\n**Patch verification signal:** ASAN-clean exit · uncaught-exception-free · sanitizer hook silent (Jazzer\u002FAtheris) · canary file untouched · invariant assertion holds · differential output matches reference. This is what `_t1_passes()` in `patch_grade.py` encodes — \"the bug is gone\" for the new domain.\n\n## Background — architecture map (what changes vs. what stays)\n\n| File | C\u002FC++-specific? | What it does |\n|---|---|---|\n| `harness\u002Fprompts\u002Ffind_prompt.py` | **Yes — rewrite** | Bug taxonomy, quality tiers, ASAN output format, exit-code examples |\n| `harness\u002Fprompts\u002Fgrade_prompt.py` | **Yes — rewrite** | 5-criterion rubric assumes ASAN traces and Unix signal exit codes |\n| `harness\u002Fprompts\u002Freport_prompt.py` | **Yes — rewrite** | Exploitability sections: primitive, heap layout, escalation path — memory-safety-specific |\n| `harness\u002Fprompts\u002Freport_grader_prompt.py` | **Yes — rewrite** | Scores the above sections; rubric is tied to the section set |\n| `harness\u002Fprompts\u002Fjudge_prompt.py` | **Yes — rewrite** | Triage keys on ASAN excerpts and crash-class taxonomy |\n| `harness\u002Fprompts\u002Fpatch_prompt.py` | **Yes — rewrite** | Asks for `git diff -- '*.c' '*.h'`, assumes ASAN trace, `memcpy`-style root-cause guidance |\n| `harness\u002Fprompts\u002Fsystem_prompt.py` | **Yes — rewrite** | Authorization block says \"C\u002FC++ target\", \"sanitizer output\" |\n| `harness\u002Fasan.py` | **Yes — rewrite** | Regex for `#N 0xHEX in func \u002Fpath:line` frames; feeds dedup, judge, novelty |\n| `targets\u002FREADME.md` + Dockerfile template | **Yes — rewrite** | `gcc -fsanitize=address`, `entry.c` wrapper pattern |\n| `harness\u002Fpatch_grade.py` | Light edit | `_t1_passes()` checks `AddressSanitizer:` substring; rest of the verification ladder is generic |\n| `harness\u002Freport.py` | Light edit | `_SECTIONS` tuple and token lists need to match the new report structure; flow is generic |\n| `harness\u002Fnovelty.py` | Light edit | `crash_file_from_frame()` is ASAN-specific; git-log logic is generic. Drop entirely if no upstream. |\n| `harness\u002Fconfig.py` | Light edit | May need new fields (`profile`, `run_command` instead of `binary_path`); `attack_surface` likely stays |\n| `harness\u002Fartifacts.py` | Light edit | `crash_type`\u002F`exit_code` semantics may shift; `ReportVerdict.section_scores` keys must match new sections |\n| `harness\u002Fdedup.py` | Light edit | Signature function needs the new parser; grouping logic is generic |\n| `harness\u002Fprompts\u002Frecon_prompt.py` | Light edit | Mostly language-agnostic; scrub C idioms |\n| `harness\u002Fcli.py` | **Unchanged** | Orchestration is domain-neutral |\n| `harness\u002Fagent.py` | **Unchanged** | Agent runner is generic |\n| `harness\u002Fdocker_ops.py` | **Unchanged** | Container plumbing is generic (may need changes if isolation ≠ Docker) |\n| `harness\u002Ffind.py`, `grade.py`, `recon.py`, `judge.py`, `patch.py` | **Unchanged** | Flow is generic; only injected prompts change |\n\n## STEP 3 — Present a plan and get confirmation\n\nBefore editing anything, summarize back to the user:\n\n1. **What you understood** — restate their goal in one sentence\n2. **What will change** — list each file you'll edit with a one-line rationale\n3. **What stays** — reassure them the orchestration core is untouched; call\n   out explicitly if the plan does reach into the shared files listed above\n   (they also serve `dnr_harness\u002F`) and why\n4. **Open questions** — anything you're still unsure about\n\nWait for explicit approval. If they adjust the plan, incorporate and re-confirm.\n\n## STEP 4 — Execute\n\nEdit the files per the approved plan. Work through them in dependency order: prompts and parser first (they're standalone), then config\u002Fartifacts, then the target template, then README. Commit incrementally if the user wants checkpoints.\n\n## STEP 5 — Validate\n\n1. Add a canary target under `targets\u002F\u003Cdomain>-canary\u002F` with 2–3 planted bugs of the new class\n2. Run: `bin\u002Fvp-sandboxed run \u003Cdomain>-canary --model \u003Cmodel-id> --runs 3 --parallel --stream --max-turns 50` (use Claude Opus unless the user specifies a different model). Run `.\u002Fscripts\u002Fsetup_sandbox.sh` once first if the sandbox isn't already set up.\n3. Confirm all planted bugs are found and graded PASS\n4. Confirm judge triage worked: `cat results\u002F\u003Cdomain>-canary\u002F\u003Cts>\u002Freports\u002Fjudge_log.jsonl` — expect one NEW per distinct bug, DUP_SKIP for repeats\n5. Confirm reports landed: `ls results\u002F\u003Cdomain>-canary\u002F\u003Cts>\u002Freports\u002Fbug_*\u002Freport.json` and spot-check section scores\n6. Run `vuln-pipeline dedup results\u002F\u003Cdomain>-canary\u002F` and confirm signatures group correctly\n","# Customize the vuln-pipeline\n\nThis pipeline ships as an opinionated C\u002FC++ + AddressSanitizer demo. Its real shape is more general: **an agent crafts an input, runs a target in a sandbox, a detector fires, a second agent verifies, a third agent analyzes exploitability.** Every noun in that sentence can be swapped. Your job is to interview the user, figure out which nouns they want to swap, and rewrite the relevant files.\n\nThe existing C\u002FC++ code is the worked example. You don't need a playbook for each domain — read what's there, understand what's generic vs. ASAN-specific, and adapt.\n\n## STEP 1 — Read the pipeline (do this BEFORE asking anything)\n\nSkim these files so your questions are grounded:\n\n- `README.md` — pipeline overview (recon → find → grade → judge → report)\n- `harness\u002Fcli.py` — orchestration; shows how stages wire together and what lands on disk\n- `harness\u002Ffind.py`, `harness\u002Fgrade.py`, `harness\u002Freport.py` — the three container-agent loops; mostly generic plumbing\n- `harness\u002Fprompts\u002Ffind_prompt.py`, `harness\u002Fprompts\u002Fgrade_prompt.py` — **the C\u002FC++-specific parts**; bug taxonomy, quality tiers, grading rubric\n- `harness\u002Fprompts\u002Freport_prompt.py`, `harness\u002Fprompts\u002Freport_grader_prompt.py` — **also C\u002FC++-specific**; exploitability sections (primitive, heap layout, escalation path) and the rubric that scores them\n- `harness\u002Fprompts\u002Fjudge_prompt.py` — triage prompt; keys on ASAN excerpts and memory-safety crash classes\n- `harness\u002Fprompts\u002Fsystem_prompt.py` — authorization block; hard-codes \"C\u002FC++ target\" and \"sanitizer output\"\n- `harness\u002Fasan.py` — stack-trace parser for dedup\u002Fjudge signatures; ASAN-specific regex\n- `harness\u002Fartifacts.py` — `CrashArtifact`, `GraderVerdict`, `JudgeVerdict`, `ReportVerdict` data contracts\n- `harness\u002Fconfig.py`, `targets\u002Fdrlibs\u002Fconfig.yaml` — target config schema\n- `targets\u002FREADME.md` — how a target directory is structured (Dockerfile + config.yaml + entry wrapper)\n\nYou don't need `agent.py`, `docker_ops.py`, `recon.py`, `judge.py`, or `novelty.py` in detail — they're generic plumbing (judge\u002Fnovelty domain-specificity lives in the prompts and the asan parser, not the flow). One caution: part of `harness\u002F` doubles as shared infrastructure — `dnr_harness\u002F` (the detection-and-response pipeline) imports `agent.py`, `sandbox.py`, `docker_ops.py`, `agent_image.py`'s `build`\u002F`ensure_base`, `auth.py`, `cli.py`'s `resolve_target_dir`\u002F`terminate_subprocesses`, and `prompts\u002Fsystem_prompt.py`'s `load_engagement_context`. A domain port should touch only the domain-specific files (the prompts, `asan.py`, `artifacts.py`, target Dockerfiles); treat any edit to the shared files as a change to both pipelines.\n\n## STEP 2 — Interview the user\n\nUse **AskUserQuestion** to gather requirements. Start with broad context, then narrow to technical specifics based on what they say.\n\n### Round 1 — Context (always ask these first, together)\n\nTwo open-ended questions to understand who you're talking to and what they're after. Expect most answers to come via **Other** as free text — the options are there to prompt thinking, not to constrain.\n\n**Question A — Operating context**\n- header: `Context`\n- question: `What's your operating environment? Who will run this pipeline and why?`\n- options: a few archetypes as inspiration — e.g. \"Pentesting firm — client engagements, need reportable findings\", \"Internal appsec team — scan our own services in CI\", \"Smart-contract auditor — pre-deployment reviews\", \"Security researcher — hunting novel bug classes\". These tell you what output format, grading rigor, and workflow integration matter.\n\n**Question B — Goal**\n- header: `Goal`\n- question: `Describe in your own words what you want this pipeline to find. What kind of target, what kind of bugs?`\n- options: 2–3 concrete examples (e.g. \"Web vulnerabilities like SQLi\u002FXSS in HTTP services\", \"Reentrancy and access-control bugs in Solidity contracts\", \"Deserialization RCE in Java microservices\").\n\nThe context answer calibrates your follow-ups: a pentesting firm probably cares about CVSS scoring and SARIF output; a researcher may want differential testing and novel detection signals; an internal team likely wants CI integration and low false-positive rates.\n\n### Round 2 — Technical follow-ups (adaptive — derive from round-1 answers)\n\nParse their round-1 answers against the **axes of variation** below. For each axis left ambiguous, ask a targeted follow-up. Batch up to 4 questions per AskUserQuestion call. Common follow-ups:\n\n- **Detection signal** — \"How will the pipeline know it found something?\" (crash, exception, canary file appears, DNS callback, differential mismatch, invariant violation)\n- **PoC shape** — \"What does a proof-of-concept look like?\" (single file, HTTP request sequence, transaction list, test-pipeline code)\n- **Isolation** — \"Where does the target run?\" (Docker, VM, testnet, remote sandbox, or no execution — static-only)\n- **Grading criteria** — \"What makes a finding high-quality vs. low-quality in this domain?\"\n- **Exploitability analysis** — \"What sections should a report contain?\" The C\u002FC++ report has primitive · reachability · heap layout · escalation path · constraints. A web-vuln report might want injection vector · auth bypass · data exposure · chaining potential. Ask what they need, or whether they want the report stage at all.\n- **Novelty\u002Fupstream check** — \"Should the pipeline check if a finding is already fixed upstream?\" The C\u002FC++ version shallow-clones the target's GitHub and checks `git log \u003Ccommit>..HEAD -- \u003Ccrash_file>`. Only applies if targets have a canonical upstream and a sensible \"crashing file\" to key on — many domains won't.\n- **Scope** — \"Replace the C\u002FC++ support entirely, or keep it alongside the new domain via a profile system?\"\n\nKeep going until you can fill in every row of the architecture map in STEP 3. If an answer is vague, ask a narrower follow-up rather than guessing.\n\n## Background — axes of variation (context for formulating follow-ups)\n\nThese are the dimensions along which customers might want to deviate from the C\u002FC++ demo. Use this list to spot gaps in the user's description and generate follow-up questions — do **not** present it as a menu.\n\n**Vulnerability class:** memory safety · web\u002FAPI (SQLi, XSS, SSRF, XXE, path traversal, IDOR) · deserialization RCE · logic\u002Frace (TOCTOU, privilege escalation) · crypto (weak RNG, timing, nonce reuse) · DoS (ReDoS, hash flooding) · smart contracts (reentrancy, access control, front-running) · ML\u002FAI (prompt injection, jailbreaks, data extraction) · protocol parsing\n\n**Target shape:** CLI binary + file · HTTP service · library via test harness · network daemon · smart contract · browser extension · mobile app\n\n**Detection mechanism:** crash\u002Fabort · uncaught exception · sanitizer hooks (Jazzer\u002FAtheris) · outcome-based (canary file, DNS callback, shell spawn) · differential testing · invariant violation · taint tracking\n\n**Input modality:** single file · HTTP request chain · multi-file archive · stdin stream · args + env + config combo · transaction sequence\n\n**Isolation boundary:** Docker container · full VM · remote sandbox · local testnet · none (static analysis)\n\n**Dedup signature:** (crash_type, top_frame) · (vuln_type, endpoint, param) · (function, state_transition) · (component, precondition)\n\n**Report structure:** primitive\u002Fheap\u002Fescalation (memory safety) · vector\u002Fauth\u002Fexposure (web) · invariant\u002Fpath\u002Fimpact (contracts) · or drop the report stage entirely if find+grade is the deliverable\n\n**Output format:** result.json + poc.bin · SARIF · Nuclei template · prose report\n\n**Patch verification signal:** ASAN-clean exit · uncaught-exception-free · sanitizer hook silent (Jazzer\u002FAtheris) · canary file untouched · invariant assertion holds · differential output matches reference. This is what `_t1_passes()` in `patch_grade.py` encodes — \"the bug is gone\" for the new domain.\n\n## Background — architecture map (what changes vs. what stays)\n\n| File | C\u002FC++-specific? | What it does |\n|---|---|---|\n| `harness\u002Fprompts\u002Ffind_prompt.py` | **Yes — rewrite** | Bug taxonomy, quality tiers, ASAN output format, exit-code examples |\n| `harness\u002Fprompts\u002Fgrade_prompt.py` | **Yes — rewrite** | 5-criterion rubric assumes ASAN traces and Unix signal exit codes |\n| `harness\u002Fprompts\u002Freport_prompt.py` | **Yes — rewrite** | Exploitability sections: primitive, heap layout, escalation path — memory-safety-specific |\n| `harness\u002Fprompts\u002Freport_grader_prompt.py` | **Yes — rewrite** | Scores the above sections; rubric is tied to the section set |\n| `harness\u002Fprompts\u002Fjudge_prompt.py` | **Yes — rewrite** | Triage keys on ASAN excerpts and crash-class taxonomy |\n| `harness\u002Fprompts\u002Fpatch_prompt.py` | **Yes — rewrite** | Asks for `git diff -- '*.c' '*.h'`, assumes ASAN trace, `memcpy`-style root-cause guidance |\n| `harness\u002Fprompts\u002Fsystem_prompt.py` | **Yes — rewrite** | Authorization block says \"C\u002FC++ target\", \"sanitizer output\" |\n| `harness\u002Fasan.py` | **Yes — rewrite** | Regex for `#N 0xHEX in func \u002Fpath:line` frames; feeds dedup, judge, novelty |\n| `targets\u002FREADME.md` + Dockerfile template | **Yes — rewrite** | `gcc -fsanitize=address`, `entry.c` wrapper pattern |\n| `harness\u002Fpatch_grade.py` | Light edit | `_t1_passes()` checks `AddressSanitizer:` substring; rest of the verification ladder is generic |\n| `harness\u002Freport.py` | Light edit | `_SECTIONS` tuple and token lists need to match the new report structure; flow is generic |\n| `harness\u002Fnovelty.py` | Light edit | `crash_file_from_frame()` is ASAN-specific; git-log logic is generic. Drop entirely if no upstream. |\n| `harness\u002Fconfig.py` | Light edit | May need new fields (`profile`, `run_command` instead of `binary_path`); `attack_surface` likely stays |\n| `harness\u002Fartifacts.py` | Light edit | `crash_type`\u002F`exit_code` semantics may shift; `ReportVerdict.section_scores` keys must match new sections |\n| `harness\u002Fdedup.py` | Light edit | Signature function needs the new parser; grouping logic is generic |\n| `harness\u002Fprompts\u002Frecon_prompt.py` | Light edit | Mostly language-agnostic; scrub C idioms |\n| `harness\u002Fcli.py` | **Unchanged** | Orchestration is domain-neutral |\n| `harness\u002Fagent.py` | **Unchanged** | Agent runner is generic |\n| `harness\u002Fdocker_ops.py` | **Unchanged** | Container plumbing is generic (may need changes if isolation ≠ Docker) |\n| `harness\u002Ffind.py`, `grade.py`, `recon.py`, `judge.py`, `patch.py` | **Unchanged** | Flow is generic; only injected prompts change |\n\n## STEP 3 — Present a plan and get confirmation\n\nBefore editing anything, summarize back to the user:\n\n1. **What you understood** — restate their goal in one sentence\n2. **What will change** — list each file you'll edit with a one-line rationale\n3. **What stays** — reassure them the orchestration core is untouched; call\n   out explicitly if the plan does reach into the shared files listed above\n   (they also serve `dnr_harness\u002F`) and why\n4. **Open questions** — anything you're still unsure about\n\nWait for explicit approval. If they adjust the plan, incorporate and re-confirm.\n\n## STEP 4 — Execute\n\nEdit the files per the approved plan. Work through them in dependency order: prompts and parser first (they're standalone), then config\u002Fartifacts, then the target template, then README. Commit incrementally if the user wants checkpoints.\n\n## STEP 5 — Validate\n\n1. Add a canary target under `targets\u002F\u003Cdomain>-canary\u002F` with 2–3 planted bugs of the new class\n2. Run: `bin\u002Fvp-sandboxed run \u003Cdomain>-canary --model \u003Cmodel-id> --runs 3 --parallel --stream --max-turns 50` (use Claude Opus unless the user specifies a different model). Run `.\u002Fscripts\u002Fsetup_sandbox.sh` once first if the sandbox isn't already set up.\n3. Confirm all planted bugs are found and graded PASS\n4. Confirm judge triage worked: `cat results\u002F\u003Cdomain>-canary\u002F\u003Cts>\u002Freports\u002Fjudge_log.jsonl` — expect one NEW per distinct bug, DUP_SKIP for repeats\n5. Confirm reports landed: `ls results\u002F\u003Cdomain>-canary\u002F\u003Cts>\u002Freports\u002Fbug_*\u002Freport.json` and spot-check section scores\n6. Run `vuln-pipeline dedup results\u002F\u003Cdomain>-canary\u002F` and confirm signatures group correctly\n",{"name":5,"description":7},"main","d69096e239f2a89aabc2cd078c47664a9d3b1e11","04-安全渗透\u002Fcustomize",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":33,"prompt":43,"zipAvailable":44},[34,37,40],{"tool":35,"command":36},"npx","npx skills add lza6\u002FArmor-piercing-system-prompts --skill customize",{"tool":38,"command":39},"bunx","bunx skills add lza6\u002FArmor-piercing-system-prompts --skill customize",{"tool":41,"command":42},"pnpm","pnpm dlx skills add lza6\u002FArmor-piercing-system-prompts --skill customize","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F04-安全渗透\u002Fcustomize\u002FSKILL.md",true,[46],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"README.md","REFERENCE","参考文档",1715,"https:\u002F\u002Fgithub.com\u002Flza6\u002FArmor-piercing-system-prompts\u002Fblob\u002Fmain\u002F04-安全渗透\u002Fcustomize\u002FREADME.md",[53,61,66,73,80,87],{"skillNo":54,"name":55,"nameZh":6,"description":56,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":57,"categories":58,"keywords":59,"publishedAt":60},"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":62,"name":55,"nameZh":6,"description":56,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":63,"categories":64,"keywords":65,"publishedAt":60},"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":67,"name":68,"nameZh":6,"description":69,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":70,"categories":71,"keywords":72,"publishedAt":60},"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":74,"name":75,"nameZh":6,"description":76,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":77,"categories":78,"keywords":79,"publishedAt":60},"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":81,"name":82,"nameZh":6,"description":83,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":84,"categories":85,"keywords":86,"publishedAt":60},"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":88,"name":89,"nameZh":6,"description":90,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":91,"categories":92,"keywords":93,"publishedAt":60},"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",[],[]]