[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7CY8S0R2G8K67C6W37WS484":3,"skill-same:01M7CY8S0R2G8K67C6W37WS484":47},{"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":29,"ownerHtmlUrl":30,"ownerBlogUrl":29,"ownerDescription":31,"updatedAt":32,"install":33,"resources":46},"01M7CY8S0R2G8K67C6W37WS484","reverse-engineering",null,"Work out how shipped software behaves from the artifact itself — native binaries, Electron and web bundles, APKs, firmware, databases, packet captures — using the rea tool server where it reaches further and the shell everywhere else. Use to explain how a feature works, recover versions and dependencies, compare two builds, or rebuild a behavior from what shipped.","Aeven-AI\u002FAeven","Aeven-AI","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F216843211?v=4","01M7CXZ1Z7B67AKZV0D9QNPCN3","Aeven AI","01M7CXZ1ZY55DZ484NN64XE9TW",1,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FAeven-AI\u002FAeven\u002Fblob\u002Fmain\u002Fpackages\u002Fskill\u002Fskill-catalog\u002Fauthored\u002Freverse-engineering\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: reverse-engineering\ndescription: Work out how shipped software behaves from the artifact itself — native binaries, Electron and web bundles, APKs, firmware, databases, packet captures — using the rea tool server where it reaches further and the shell everywhere else. Use to explain how a feature works, recover versions and dependencies, compare two builds, or rebuild a behavior from what shipped.\n---\n\n# Reverse-engineer shipped software\n\nTwo instruments, one route: the `rea` server answers what a command cannot — pseudocode, callers and callees, formats it parses directly, an evidence record that outlives the session — and the shell answers everything else without starting a process. Load the server's tools with one `tool_search` for `rea` before the first target; they arrive as `mcp__rea__*` and nothing else here needs registering.\n\nIdentify the artifact with `file -b` and the magic bytes, take the cheapest inventory, then read only the part that answers the question.\n\n## Route by artifact\n\n| Target | rea tool | Shell path when the server is absent or a provider engine is missing |\n| --- | --- | --- |\n| Native executable, library, analysis database | `open_binary`, then `binary_overview` | `lipo -info`, `otool -L`, `readelf -hlSW`, `nm`, `strings` |\n| One function: pseudocode, callers, callees | `analyze_function` after `open_binary` | `otool -tV T > dis.txt`, `objdump -d -M intel`, grep `dis.txt` |\n| Electron or extracted JavaScript tree, ASAR | `analyze_javascript_application` | `npx -y @electron\u002Fasar extract`, then `rg -uu` |\n| APK classes, methods, incoming references | `inspect_android_package` | `jadx -d X --show-bad-code`, `apktool d --no-src` |\n| Managed PE or CLI assembly | `inspect_managed_artifact` | `ilspycmd -p -o X T.dll` |\n| Firmware image | `inspect_firmware_regions`, `extract_firmware` | `binwalk -e`, `unblob -e` |\n| Go build info, modules, build flags | `inspect_go_binary` | `go version -m T`, or `strings -a T \\| rg -A200 'Go buildinf'` |\n| Rust cargo-auditable dependencies | `inspect_rust_binary` | inflate `.dep-v0` from `__DATA` |\n| SQLite snapshot, schema, rows | `inspect_sqlite_database` | `sqlite3 'file:C?mode=ro' '.schema'` |\n| HAR, SAZ, recorded web traffic | `inspect_web_network_capture` | `jq -r '.log.entries[].request \\| .method+\" \"+.url'` |\n| Packet captures and field discovery | `inspect_packet_capture`, `inspect_pcap_capture`, `inspect_packet_fields` | `tshark -r C -q -z conv,tcp` |\n| WASM artifact, EVM bytecode | `inspect_wasm_artifact`, `inspect_evm_interface` | `wasm-objdump -x`, `wasm2wat` |\n| ELF layout, PE resources, core dump | `inspect_binary_layout`, `inspect_pe_resources`, `inspect_recorded_crash` | `readelf -a`, `objdump -x`, `lldb --core` |\n| Packaged Electron app, fuses, open runtime | `inspect_electron_fuses`, `list_electron_targets` | `npx -y @electron\u002Ffuses read --app A.app` |\n| Browser page, or actions to drive | `list_browser_targets`, `capture_browser_scenario` | `curl -sL URL` and the browser beside the session |\n\nHermes (`c6 1f bc 03`) and bytenode `.jsc` have no server route: `hermes-dec` or `hbc-decompiler` for the bundle, strings and runtime hooks for the rest. A stripped binary whose call target needs a decompiler database is exactly the case the server exists for.\n\n## Ground rules\n\n- Work on a copy (`cp -c` is instant on APFS) in `$TMPDIR\u002Freverse\u002F\u003Cname>`, reused for the task.\n- Search extracted trees with `rg -uu`: `rg` skips hidden and ignored paths, and bundlers hide code in `.webpack\u002F` and `.vite\u002F`.\n- Never `cat` a large or minified file. Slice it: `rg -o '.{0,200}NEEDLE.{0,200}' F | head`. Write a disassembly to a file once, then grep that file.\n- Batch independent probes into one shell call, separated by `;` and `echo '--- name'` markers. Quote them: zsh expands a bare `==x`. Cap every output with `head`.\n- Mark each claim **seen** (cite the file and offset, the symbol and string, or the tool's evidence id) or **inferred**. A fact you did not find is **unknown**, not absent, and static reading is never observed execution.\n- Read each result's stated limitations and unknowns instead of assuming it covered more; a missing provider engine narrows that one answer, not the task.\n- Independent artifacts fan out one target per subagent, three or more tracked in `todo_write`. Turns cost more than bytes: aim for two to four calls, then answer.\n\n## One call each\n\nNative binary, when the server is not the route:\n\n```sh\nT=.\u002Fbin; D=$TMPDIR\u002Freverse\u002F$(basename $T); mkdir -p $D; file $T; lipo -info $T; otool -L $T\necho '--- EXPORTS'; nm -gU $T | head -150\necho '--- IMPORTS'; nm -u $T | rg -v '^__Z|napi_' | tr '\\n' ' '; echo\necho '--- STRINGS'; strings -a -n 6 $T | rg -v '^[_$.]' | head -400\notool -tV $T > $D\u002Fdis.txt\necho '--- FUNCS'; rg '^[_a-zA-Z].*:$' $D\u002Fdis.txt | rg -v '^_?_?Z?N?St3' | head -150\n```\n\nElectron app, and the IPC surface that a feature answer usually needs:\n\n```sh\nA=.\u002Fapp.asar; X=$TMPDIR\u002Freverse\u002Fasar; rm -rf $X\nnpx -y @electron\u002Fasar extract $A $X && jq '{name,version,main}' $X\u002Fpackage.json\nrg -n -o '.{0,120}(ipcMain\\.(handle|handleOnce|on|once)|ipcRenderer\\.(invoke|send|sendSync|on)|contextBridge\\.exposeIn(Main|Isolated)World|webPreferences|preload:|nodeIntegration|contextIsolation|sandbox|\\.node\\b).{0,160}' -uu $X --glob '!node_modules'\n```\n\n## Read the code, not the file\n\n- Let one call answer the whole question where it can: `otool -L`\u002F`readelf -dW` for linked libraries, `inspect_go_binary` for a toolchain, `nm -u` for the capability map.\n- Go from the boundary inward: native call, caller class, IPC registration, the renderer that invokes it.\n- One function at a time: `analyze_function` for its pseudocode and callers, or `awk '\u002F^_FUNC:$\u002F{p=1;next} p&&\u002F^[_a-zA-Z].*:$\u002F{exit} p' dis.txt | rg 'bl|call|cmp|literal'`. A C++ name at a `bl` target names the protocol, so the call sequence is usually enough.\n- Two versions compare by inventory first (`nm`, the module list, `inspect_go_binary`). A rebuild means running the smallest reimplementation against the original on the same inputs.\n\n## When the artifact is not enough\n\nStatic reading shows what the code can do, never what it did. For runtime behavior, observe: the app's log, `fs_usage`, `lsof -p PID`, a proxy, or `capture_browser_scenario` with the actions to perform.\n\nFor a screenshot, a native dialog, or a window on the person's own screen, use computer use: load the `mcp__cua-driver-mcp__*` tools with tool search, check permissions with `prompt: false` first, and look at the screen before and after each action. Ask the person before anything that deletes data, sends a message, or changes an account.\n","# Reverse-engineer shipped software\n\nTwo instruments, one route: the `rea` server answers what a command cannot — pseudocode, callers and callees, formats it parses directly, an evidence record that outlives the session — and the shell answers everything else without starting a process. Load the server's tools with one `tool_search` for `rea` before the first target; they arrive as `mcp__rea__*` and nothing else here needs registering.\n\nIdentify the artifact with `file -b` and the magic bytes, take the cheapest inventory, then read only the part that answers the question.\n\n## Route by artifact\n\n| Target | rea tool | Shell path when the server is absent or a provider engine is missing |\n| --- | --- | --- |\n| Native executable, library, analysis database | `open_binary`, then `binary_overview` | `lipo -info`, `otool -L`, `readelf -hlSW`, `nm`, `strings` |\n| One function: pseudocode, callers, callees | `analyze_function` after `open_binary` | `otool -tV T > dis.txt`, `objdump -d -M intel`, grep `dis.txt` |\n| Electron or extracted JavaScript tree, ASAR | `analyze_javascript_application` | `npx -y @electron\u002Fasar extract`, then `rg -uu` |\n| APK classes, methods, incoming references | `inspect_android_package` | `jadx -d X --show-bad-code`, `apktool d --no-src` |\n| Managed PE or CLI assembly | `inspect_managed_artifact` | `ilspycmd -p -o X T.dll` |\n| Firmware image | `inspect_firmware_regions`, `extract_firmware` | `binwalk -e`, `unblob -e` |\n| Go build info, modules, build flags | `inspect_go_binary` | `go version -m T`, or `strings -a T \\| rg -A200 'Go buildinf'` |\n| Rust cargo-auditable dependencies | `inspect_rust_binary` | inflate `.dep-v0` from `__DATA` |\n| SQLite snapshot, schema, rows | `inspect_sqlite_database` | `sqlite3 'file:C?mode=ro' '.schema'` |\n| HAR, SAZ, recorded web traffic | `inspect_web_network_capture` | `jq -r '.log.entries[].request \\| .method+\" \"+.url'` |\n| Packet captures and field discovery | `inspect_packet_capture`, `inspect_pcap_capture`, `inspect_packet_fields` | `tshark -r C -q -z conv,tcp` |\n| WASM artifact, EVM bytecode | `inspect_wasm_artifact`, `inspect_evm_interface` | `wasm-objdump -x`, `wasm2wat` |\n| ELF layout, PE resources, core dump | `inspect_binary_layout`, `inspect_pe_resources`, `inspect_recorded_crash` | `readelf -a`, `objdump -x`, `lldb --core` |\n| Packaged Electron app, fuses, open runtime | `inspect_electron_fuses`, `list_electron_targets` | `npx -y @electron\u002Ffuses read --app A.app` |\n| Browser page, or actions to drive | `list_browser_targets`, `capture_browser_scenario` | `curl -sL URL` and the browser beside the session |\n\nHermes (`c6 1f bc 03`) and bytenode `.jsc` have no server route: `hermes-dec` or `hbc-decompiler` for the bundle, strings and runtime hooks for the rest. A stripped binary whose call target needs a decompiler database is exactly the case the server exists for.\n\n## Ground rules\n\n- Work on a copy (`cp -c` is instant on APFS) in `$TMPDIR\u002Freverse\u002F\u003Cname>`, reused for the task.\n- Search extracted trees with `rg -uu`: `rg` skips hidden and ignored paths, and bundlers hide code in `.webpack\u002F` and `.vite\u002F`.\n- Never `cat` a large or minified file. Slice it: `rg -o '.{0,200}NEEDLE.{0,200}' F | head`. Write a disassembly to a file once, then grep that file.\n- Batch independent probes into one shell call, separated by `;` and `echo '--- name'` markers. Quote them: zsh expands a bare `==x`. Cap every output with `head`.\n- Mark each claim **seen** (cite the file and offset, the symbol and string, or the tool's evidence id) or **inferred**. A fact you did not find is **unknown**, not absent, and static reading is never observed execution.\n- Read each result's stated limitations and unknowns instead of assuming it covered more; a missing provider engine narrows that one answer, not the task.\n- Independent artifacts fan out one target per subagent, three or more tracked in `todo_write`. Turns cost more than bytes: aim for two to four calls, then answer.\n\n## One call each\n\nNative binary, when the server is not the route:\n\n```sh\nT=.\u002Fbin; D=$TMPDIR\u002Freverse\u002F$(basename $T); mkdir -p $D; file $T; lipo -info $T; otool -L $T\necho '--- EXPORTS'; nm -gU $T | head -150\necho '--- IMPORTS'; nm -u $T | rg -v '^__Z|napi_' | tr '\\n' ' '; echo\necho '--- STRINGS'; strings -a -n 6 $T | rg -v '^[_$.]' | head -400\notool -tV $T > $D\u002Fdis.txt\necho '--- FUNCS'; rg '^[_a-zA-Z].*:$' $D\u002Fdis.txt | rg -v '^_?_?Z?N?St3' | head -150\n```\n\nElectron app, and the IPC surface that a feature answer usually needs:\n\n```sh\nA=.\u002Fapp.asar; X=$TMPDIR\u002Freverse\u002Fasar; rm -rf $X\nnpx -y @electron\u002Fasar extract $A $X && jq '{name,version,main}' $X\u002Fpackage.json\nrg -n -o '.{0,120}(ipcMain\\.(handle|handleOnce|on|once)|ipcRenderer\\.(invoke|send|sendSync|on)|contextBridge\\.exposeIn(Main|Isolated)World|webPreferences|preload:|nodeIntegration|contextIsolation|sandbox|\\.node\\b).{0,160}' -uu $X --glob '!node_modules'\n```\n\n## Read the code, not the file\n\n- Let one call answer the whole question where it can: `otool -L`\u002F`readelf -dW` for linked libraries, `inspect_go_binary` for a toolchain, `nm -u` for the capability map.\n- Go from the boundary inward: native call, caller class, IPC registration, the renderer that invokes it.\n- One function at a time: `analyze_function` for its pseudocode and callers, or `awk '\u002F^_FUNC:$\u002F{p=1;next} p&&\u002F^[_a-zA-Z].*:$\u002F{exit} p' dis.txt | rg 'bl|call|cmp|literal'`. A C++ name at a `bl` target names the protocol, so the call sequence is usually enough.\n- Two versions compare by inventory first (`nm`, the module list, `inspect_go_binary`). A rebuild means running the smallest reimplementation against the original on the same inputs.\n\n## When the artifact is not enough\n\nStatic reading shows what the code can do, never what it did. For runtime behavior, observe: the app's log, `fs_usage`, `lsof -p PID`, a proxy, or `capture_browser_scenario` with the actions to perform.\n\nFor a screenshot, a native dialog, or a window on the person's own screen, use computer use: load the `mcp__cua-driver-mcp__*` tools with tool search, check permissions with `prompt: false` first, and look at the screen before and after each action. Ask the person before anything that deletes data, sends a message, or changes an account.\n",{"name":5,"description":7},"main","6071877b951fef978c8859ca6ed86f7ce841b4df","packages\u002Fskill\u002Fskill-catalog\u002Fauthored\u002Freverse-engineering",0,"https:\u002F\u002Fgithub.com\u002FAeven-AI\u002FAeven","https:\u002F\u002Faeven.io","https:\u002F\u002Fgithub.com\u002FAeven-AI","Aeven is an all-plugin agent harness: models, tools and plugins in one runtime, working on your files, your shell and your apps.","2026-11-15 04:04:54",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add Aeven-AI\u002FAeven --skill reverse-engineering",{"tool":39,"command":40},"bunx","bunx skills add Aeven-AI\u002FAeven --skill reverse-engineering",{"tool":42,"command":43},"pnpm","pnpm dlx skills add Aeven-AI\u002FAeven --skill reverse-engineering","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FAeven-AI\u002FAeven\u002Fblob\u002Fmain\u002Fpackages\u002Fskill\u002Fskill-catalog\u002Fauthored\u002Freverse-engineering\u002FSKILL.md",true,[],[48,55,60,67,74,81],{"skillNo":49,"name":50,"nameZh":6,"description":51,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":52,"categories":53,"keywords":54,"publishedAt":19},"01M7E2H36CVFPD4AQT7D371BFB","cv","Make a CV or résumé with the user in one of Aeven's forty templates, from bold portrait-cover designs (Black, Slate, Fresh) to classic one-page resumes (Harvard, Tang, Minimal Rule), and deliver it as PDF, Word (DOCX) and Pages. Use for any CV, résumé or curriculum vitae request, when the user picked a template on the Resumes page (\"…with the Black resume template\"), or when they want an existing CV rebuilt in a new design.","https:\u002F\u002Fgithub.com\u002FAeven-AI\u002FAeven\u002Fblob\u002Fmain\u002Fpackages\u002Fpreset\u002Fagent-presets\u002Fpresets\u002Faeven\u002Fskills\u002Fcv\u002FSKILL.md",[],[],{"skillNo":56,"name":50,"nameZh":6,"description":6,"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":19},"01M7E2H1RSHQW0HY1R25R8J3Z0","https:\u002F\u002Fgithub.com\u002FAeven-AI\u002FAeven\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fcv\u002FSKILL.md",[],[],{"skillNo":61,"name":62,"nameZh":6,"description":63,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":64,"categories":65,"keywords":66,"publishedAt":19},"01M7CY96F99ZF6ZQWPR2V7YK9T","user-only-skill","Prove model-disabled skills stay outside the model catalog.","https:\u002F\u002Fgithub.com\u002FAeven-AI\u002FAeven\u002Fblob\u002Fmain\u002Fsnapshots\u002Fsession\u002Fskill-load\u002Fworkspace\u002F.aeven\u002Fskills\u002Fuser-only-skill\u002FSKILL.md",[],[],{"skillNo":68,"name":69,"nameZh":6,"description":70,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":71,"categories":72,"keywords":73,"publishedAt":19},"01M7CY94ZG214EW8KT5WSKCNJY","snapshot-skill","Exercise project skill discovery and loading in snapshot tests.","https:\u002F\u002Fgithub.com\u002FAeven-AI\u002FAeven\u002Fblob\u002Fmain\u002Fsnapshots\u002Fsession\u002Fskill-load\u002Fworkspace\u002F.aeven\u002Fskills\u002Fsnapshot-skill\u002FSKILL.md",[],[],{"skillNo":75,"name":76,"nameZh":6,"description":77,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":78,"categories":79,"keywords":80,"publishedAt":19},"01M7CY93DE4TJQ724SRAE9V714","model-only-skill","Prove user-disabled skills remain available to the model.","https:\u002F\u002Fgithub.com\u002FAeven-AI\u002FAeven\u002Fblob\u002Fmain\u002Fsnapshots\u002Fsession\u002Fskill-load\u002Fworkspace\u002F.aeven\u002Fskills\u002Fmodel-only-skill\u002FSKILL.md",[],[],{"skillNo":82,"name":83,"nameZh":6,"description":84,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":85,"categories":86,"keywords":87,"publishedAt":19},"01M7CY8YRXNNHE64BY5FE289PM","editing-cordis-compositions","Use when creating, changing, or validating a Cordis composition for this harness — writing or editing an agent preset, adding or removing a plugin row, deciding whether something belongs to the host composition or to one session, checking whether a preset you authored actually mounts, or diagnosing a row that mounted but contributed nothing.","https:\u002F\u002Fgithub.com\u002FAeven-AI\u002FAeven\u002Fblob\u002Fmain\u002Fsnapshots\u002Fsession\u002Fskill-load\u002Fworkspace\u002F.aeven\u002Fskills\u002Fediting-cordis-compositions\u002FSKILL.md",[],[]]