[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6KMM0RAVR43Y31NC75Z5NA3":3,"skill-same:01M6KMM0RAVR43Y31NC75Z5NA3":58},{"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":30,"linkOnly":31,"markdownContent":32,"markdownBody":33,"frontMatter":34,"refName":36,"commitSha":37,"skillDirectory":38,"forksCount":39,"repositoryHtmlUrl":40,"repositoryHomepageUrl":6,"ownerHtmlUrl":41,"ownerBlogUrl":42,"ownerDescription":6,"updatedAt":43,"install":44,"resources":57},"01M6KMM0RAVR43Y31NC75Z5NA3","tools-spin-model",null,"Create a new Spin\u002FPromela formal verification model for a system property","AI-native-Systems-Research\u002Fai-native-storage-certus","AI-native-Systems-Research","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F270755458?v=4","01M6KKQWSJX0D4X33E0KKSR4WM","AI-native Systems Research","01M6KKQWXTKR6MYVQMPKVJ5WD5",3,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FAI-native-Systems-Research\u002Fai-native-storage-certus\u002Fblob\u002Funstable\u002Fcomponents\u002Fblock-device-kernel\u002F.claude\u002Fskills\u002Ftools-spin-model\u002FSKILL.md",[],[19,20,21,22,23,24,25,26,27,28,29],"Tools","model","tools spin model","create","verification","system","formal verification","number","for","prove","new","2026-11-10 10:12:30",false,"---\nname: tools-spin-model\ndescription: Create a new Spin\u002FPromela formal verification model for a system property\nargument-hint: \"[property-number or description of property to prove]\"\n---\n\nCreate a new Spin\u002FPromela specification in `modelling\u002Fspin\u002F\u003Cname>\u002F` to verify a concurrency property of the Certus system.\n\n## Input\n\nIf no argument is provided, first ask the user to define the **scope**, then present the property menu.\n\n### Step A: Scope Selection\n\nAsk the user to choose the modeling scope:\n\n- **Single component** — Model the internal concurrency of one component in isolation (e.g., just the dispatcher, just the extent-manager, just the block-device actor). Threads and channels within that component are modeled explicitly; interactions with other components are abstracted as nondeterministic environment stubs.\n- **Whole system** — Model the interaction between multiple components as wired together in certus-server (e.g., client → dispatcher → memory-tier → block-device → extent-manager). Includes cross-component communication channels, the gRPC thread pool, and background workers.\n\nThe scope determines:\n- Which source files to read (single component = one crate's `src\u002F`; whole system = multiple crates + `apps\u002Fcertus-server\u002F`)\n- How to abstract boundaries (single = environment stubs with nondeterministic responses; whole = explicit `proctype` per component role)\n- Model complexity (single = fewer processes, deeper per-component detail; whole = more processes, coarser per-component abstraction)\n\n### Step B: Property Selection\n\nPresent the full property menu and ask the user to choose. The user may also describe a custom property.\n\n**System-level properties** (scope: whole system recommended):\n\n1. **Reference-count balance** — Every read_ref increment has a matching decrement on all paths. No leaked refs (memory leak) or double-decrements (use-after-free).\n2. **Write-before-evict** — Models the dispatcher's populate → write-through → eviction lifecycle to verify that memory-tier eviction never produces dangling SSD references.\n3. **Cold-path promotion atomicity** — At most one thread succeeds in promoting a given key from SSD to memory-tier. No double-allocation or lost updates.\n4. **Populate-lookup linearizability** — After populate(key) returns Ok, a concurrent lookup(key) never observes KeyNotFound unless an explicit remove or eviction intervened.\n5. **Shutdown ordering** — No NVMe command is issued after its block device shuts down. Background writer drains before drives are torn down.\n6. **No lost extents** — Every reserve_extent is followed by either publish() or drop (abort). No extent is permanently allocated but uncommitted.\n\n**Component-level properties** (scope: single component recommended):\n\n7. **Pipeline ring slot safety** — A ring buffer slot is never resubmitted for NVMe read while its CUDA DMA transfer is still in flight. (Component: dispatcher\u002Fpipeline)\n8. **Starvation freedom** — No client is starved indefinitely waiting to enqueue a write job under backpressure. (Component: dispatcher\u002Fbackground)\n9. **Eviction fairness** — An entry touched within the last N operations is never evicted while untouched entries exist. (Component: dispatcher\u002Feviction)\n10. **Actor mailbox ordering** — Messages sent to a block-device actor are processed in FIFO order; no command reordering across the channel boundary. (Component: block-device-spdk-nvme)\n11. **Extent bitmap consistency** — Concurrent reserve\u002Fpublish\u002Fremove never leave the bitmap in a state where an extent is both allocated and free. (Component: extent-manager)\n\nIf the user provides a number (1-11), use that property. If they provide free text, treat it as a custom property description and ask for a short kebab-case name for the subdirectory.\n\n## Steps\n\n1. **Determine the property and directory name.**\n\n   Map selections to directory names:\n   - 1 → `ref-count-balance`\n   - 2 → `write-before-evict`\n   - 3 → `promotion-atomicity`\n   - 4 → `populate-lookup-linearizability`\n   - 5 → `shutdown-ordering`\n   - 6 → `no-lost-extents`\n   - 7 → `pipeline-ring-safety`\n   - 8 → `starvation-freedom`\n   - 9 → `eviction-fairness`\n   - 10 → `actor-mailbox-ordering`\n   - 11 → `extent-bitmap-consistency`\n   - Custom → ask user for a kebab-case name\n\n2. **Read the relevant source code** to understand the real synchronization protocol.\n\n   Key source files by property and scope:\n\n   **Whole-system scope** (read certus-server wiring + component internals):\n   - 1,4: `components\u002Fdispatcher\u002Fsrc\u002Flib.rs` (populate, lookup, evict_for_space, batch_lookup)\n   - 2: `components\u002Fdispatcher\u002Fsrc\u002Flib.rs` + `components\u002Fdispatcher\u002Fsrc\u002Fbackground.rs`\n   - 3: `components\u002Fdispatcher\u002Fsrc\u002Flib.rs` (batch_lookup cold-path promotion, lines ~1045-1215)\n   - 5: `components\u002Fdispatcher\u002Fsrc\u002Flib.rs` (shutdown), `components\u002Fdispatcher\u002Fsrc\u002Fbackground.rs` (BackgroundWriter::shutdown), `apps\u002Fcertus-server\u002Fsrc\u002Fmain.rs` (shutdown sequence)\n   - 6: `components\u002Fdispatcher\u002Fsrc\u002Flib.rs` (prepare_store, commit_store, cancel_store, process_write_job)\n\n   **Single-component scope** (read only the target component):\n   - 7: `components\u002Fdispatcher\u002Fsrc\u002Fpipeline.rs` (PipelineRing, pipelined_ssd_to_gpu_zero_copy)\n   - 8: `components\u002Fdispatcher\u002Fsrc\u002Fbackground.rs` (channel + worker_loop)\n   - 9: `components\u002Fdispatcher\u002Fsrc\u002Flib.rs` (evict_for_space, mt.evict_lru, mt.oldest_keys)\n   - 10: `components\u002Fblock-device-spdk-nvme\u002Fsrc\u002F` (actor thread, command channel, completion channel)\n   - 11: `components\u002Fextent-manager\u002Fsrc\u002F` (bitmap allocation, reserve_extent, publish, remove_extent)\n\n   Also reference the existing model at `modelling\u002Fspin\u002Fwrite-before-evict\u002Fwrite_before_evict.pml` for style and conventions.\n\n3. **Create the subdirectory** `modelling\u002Fspin\u002F\u003Cname>\u002F`.\n\n4. **Write the Promela specification** `modelling\u002Fspin\u002F\u003Cname>\u002F\u003Cname_underscored>.pml`.\n\n   Follow these conventions (from the existing write-before-evict model):\n   - Header comment block explaining what is being verified, the scope, and how to run\n   - `#define` parameters section (N_CLIENTS, N_KEYS, etc.) tuned small for tractable state space\n   - Per-key or per-resource state as `mtype` enums and arrays\n   - `proctype` for each thread role in the system\n   - `inline` helpers for shared logic (like eviction)\n   - Inline `assert()` statements for safety properties\n   - `init` process that starts all proctypes, waits for completion, signals shutdown, runs final invariant checks\n   - Aim for 100% state coverage (no unreached states)\n   - Use `atomic{}` for operations that share a mutex in the real code\n   - Use `d_step{}` for deterministic sequences (scans)\n   - Use generation counters to handle key reuse across lifecycles\n   - Model channels as bounded `chan` (even if real code uses unbounded)\n\n   **Scope-specific conventions:**\n\n   For **single component** models:\n   - Abstract external interfaces as nondeterministic stubs (e.g., `if :: success :: failure fi`)\n   - Focus on internal thread interactions and shared state\n   - Name the header comment \"Component-level model: \u003Ccomponent-name>\"\n   - Document which interfaces are stubbed and what assumptions are made\n\n   For **whole system** models:\n   - One `proctype` per major thread role across components\n   - Use channels to model inter-component communication (gRPC, receptacles)\n   - Name the header comment \"System-level model: \u003Cproperty-name>\"\n   - Document the component wiring in the header\n\n5. **Write the Makefile** `modelling\u002Fspin\u002F\u003Cname>\u002FMakefile`:\n\n   ```makefile\n   MODEL = \u003Cname_underscored>.pml\n   PAN   = pan\n   CC    = cc\n   CFLAGS = -O2\n   DEPTH = 200000\n\n   .PHONY: all safety liveness clean\n\n   all: safety\n\n   safety: $(PAN)\n   \t.\u002F$(PAN) -m$(DEPTH)\n\n   liveness: $(PAN)-live\n   \t.\u002F$(PAN)-live -a -m$(DEPTH)\n\n   $(PAN): pan.c\n   \t$(CC) $(CFLAGS) -DSAFETY -o $@ $\u003C\n\n   $(PAN)-live: pan.c\n   \t$(CC) $(CFLAGS) -o $@ $\u003C\n\n   pan.c: $(MODEL)\n   \tspin -a $\u003C\n\n   clean:\n   \trm -f $(PAN) $(PAN)-live pan.* *.trail\n   ```\n\n6. **Write the README.md** `modelling\u002Fspin\u002F\u003Cname>\u002FREADME.md`:\n\n   Structure (follow the pattern in `modelling\u002Fspin\u002Fwrite-before-evict\u002FREADME.md`):\n   - Title: `# \u003CProperty Name>`\n   - **Scope**: Single component (`\u003Ccomponent-name>`) or Whole system\n   - One-paragraph description of what the model verifies\n   - **Properties Verified** table (ID, Property description, Type: Safety\u002FLiveness)\n   - **System Abstraction** table (Real component → Promela process)\n   - **Assumptions \u002F Stubs** section (for single-component scope: list what is abstracted away and how)\n   - **Running** section with shell commands\n   - **Tuning the Model** section explaining parameters\n   - **Correspondence to Source Code** table (Model location → Source file → Line range)\n\n7. **Run the verification** by executing `make` in the new directory.\n\n   - If Spin is not installed, inform the user and skip this step.\n   - If verification fails with an assertion violation, analyze the trail (`spin -t -p \u003Cfile>.pml`), fix the model, and re-run until it passes.\n   - Report: number of states explored, depth reached, errors found, coverage (unreached states).\n\n8. **Report results** to the user:\n   - Scope (single component or whole system)\n   - Property being verified\n   - Pass\u002Ffail status\n   - State space size\n   - Coverage (target: 0 unreached states in all proctypes)\n   - Path to the new model directory\n\n## Notes\n\n- Spin must be installed at `\u002Fusr\u002Flocal\u002Fbin\u002Fspin` (see `modelling\u002Fspin\u002Fwrite-before-evict\u002FREADME.md` for install instructions).\n- Keep parameters small (2-3 clients, 2-4 keys, pool cap \u003C key count) to maintain tractable state spaces (\u003C10M states).\n- The model should be self-contained — no external dependencies beyond Spin and a C compiler.\n- Use the generation-counter technique from the write-before-evict model to handle key reuse without conflating lifecycles.\n- Single-component models are faster to verify (fewer processes) and better for finding internal races.\n- Whole-system models catch cross-component protocol violations but have larger state spaces.\n","Create a new Spin\u002FPromela specification in `modelling\u002Fspin\u002F\u003Cname>\u002F` to verify a concurrency property of the Certus system.\n\n## Input\n\nIf no argument is provided, first ask the user to define the **scope**, then present the property menu.\n\n### Step A: Scope Selection\n\nAsk the user to choose the modeling scope:\n\n- **Single component** — Model the internal concurrency of one component in isolation (e.g., just the dispatcher, just the extent-manager, just the block-device actor). Threads and channels within that component are modeled explicitly; interactions with other components are abstracted as nondeterministic environment stubs.\n- **Whole system** — Model the interaction between multiple components as wired together in certus-server (e.g., client → dispatcher → memory-tier → block-device → extent-manager). Includes cross-component communication channels, the gRPC thread pool, and background workers.\n\nThe scope determines:\n- Which source files to read (single component = one crate's `src\u002F`; whole system = multiple crates + `apps\u002Fcertus-server\u002F`)\n- How to abstract boundaries (single = environment stubs with nondeterministic responses; whole = explicit `proctype` per component role)\n- Model complexity (single = fewer processes, deeper per-component detail; whole = more processes, coarser per-component abstraction)\n\n### Step B: Property Selection\n\nPresent the full property menu and ask the user to choose. The user may also describe a custom property.\n\n**System-level properties** (scope: whole system recommended):\n\n1. **Reference-count balance** — Every read_ref increment has a matching decrement on all paths. No leaked refs (memory leak) or double-decrements (use-after-free).\n2. **Write-before-evict** — Models the dispatcher's populate → write-through → eviction lifecycle to verify that memory-tier eviction never produces dangling SSD references.\n3. **Cold-path promotion atomicity** — At most one thread succeeds in promoting a given key from SSD to memory-tier. No double-allocation or lost updates.\n4. **Populate-lookup linearizability** — After populate(key) returns Ok, a concurrent lookup(key) never observes KeyNotFound unless an explicit remove or eviction intervened.\n5. **Shutdown ordering** — No NVMe command is issued after its block device shuts down. Background writer drains before drives are torn down.\n6. **No lost extents** — Every reserve_extent is followed by either publish() or drop (abort). No extent is permanently allocated but uncommitted.\n\n**Component-level properties** (scope: single component recommended):\n\n7. **Pipeline ring slot safety** — A ring buffer slot is never resubmitted for NVMe read while its CUDA DMA transfer is still in flight. (Component: dispatcher\u002Fpipeline)\n8. **Starvation freedom** — No client is starved indefinitely waiting to enqueue a write job under backpressure. (Component: dispatcher\u002Fbackground)\n9. **Eviction fairness** — An entry touched within the last N operations is never evicted while untouched entries exist. (Component: dispatcher\u002Feviction)\n10. **Actor mailbox ordering** — Messages sent to a block-device actor are processed in FIFO order; no command reordering across the channel boundary. (Component: block-device-spdk-nvme)\n11. **Extent bitmap consistency** — Concurrent reserve\u002Fpublish\u002Fremove never leave the bitmap in a state where an extent is both allocated and free. (Component: extent-manager)\n\nIf the user provides a number (1-11), use that property. If they provide free text, treat it as a custom property description and ask for a short kebab-case name for the subdirectory.\n\n## Steps\n\n1. **Determine the property and directory name.**\n\n   Map selections to directory names:\n   - 1 → `ref-count-balance`\n   - 2 → `write-before-evict`\n   - 3 → `promotion-atomicity`\n   - 4 → `populate-lookup-linearizability`\n   - 5 → `shutdown-ordering`\n   - 6 → `no-lost-extents`\n   - 7 → `pipeline-ring-safety`\n   - 8 → `starvation-freedom`\n   - 9 → `eviction-fairness`\n   - 10 → `actor-mailbox-ordering`\n   - 11 → `extent-bitmap-consistency`\n   - Custom → ask user for a kebab-case name\n\n2. **Read the relevant source code** to understand the real synchronization protocol.\n\n   Key source files by property and scope:\n\n   **Whole-system scope** (read certus-server wiring + component internals):\n   - 1,4: `components\u002Fdispatcher\u002Fsrc\u002Flib.rs` (populate, lookup, evict_for_space, batch_lookup)\n   - 2: `components\u002Fdispatcher\u002Fsrc\u002Flib.rs` + `components\u002Fdispatcher\u002Fsrc\u002Fbackground.rs`\n   - 3: `components\u002Fdispatcher\u002Fsrc\u002Flib.rs` (batch_lookup cold-path promotion, lines ~1045-1215)\n   - 5: `components\u002Fdispatcher\u002Fsrc\u002Flib.rs` (shutdown), `components\u002Fdispatcher\u002Fsrc\u002Fbackground.rs` (BackgroundWriter::shutdown), `apps\u002Fcertus-server\u002Fsrc\u002Fmain.rs` (shutdown sequence)\n   - 6: `components\u002Fdispatcher\u002Fsrc\u002Flib.rs` (prepare_store, commit_store, cancel_store, process_write_job)\n\n   **Single-component scope** (read only the target component):\n   - 7: `components\u002Fdispatcher\u002Fsrc\u002Fpipeline.rs` (PipelineRing, pipelined_ssd_to_gpu_zero_copy)\n   - 8: `components\u002Fdispatcher\u002Fsrc\u002Fbackground.rs` (channel + worker_loop)\n   - 9: `components\u002Fdispatcher\u002Fsrc\u002Flib.rs` (evict_for_space, mt.evict_lru, mt.oldest_keys)\n   - 10: `components\u002Fblock-device-spdk-nvme\u002Fsrc\u002F` (actor thread, command channel, completion channel)\n   - 11: `components\u002Fextent-manager\u002Fsrc\u002F` (bitmap allocation, reserve_extent, publish, remove_extent)\n\n   Also reference the existing model at `modelling\u002Fspin\u002Fwrite-before-evict\u002Fwrite_before_evict.pml` for style and conventions.\n\n3. **Create the subdirectory** `modelling\u002Fspin\u002F\u003Cname>\u002F`.\n\n4. **Write the Promela specification** `modelling\u002Fspin\u002F\u003Cname>\u002F\u003Cname_underscored>.pml`.\n\n   Follow these conventions (from the existing write-before-evict model):\n   - Header comment block explaining what is being verified, the scope, and how to run\n   - `#define` parameters section (N_CLIENTS, N_KEYS, etc.) tuned small for tractable state space\n   - Per-key or per-resource state as `mtype` enums and arrays\n   - `proctype` for each thread role in the system\n   - `inline` helpers for shared logic (like eviction)\n   - Inline `assert()` statements for safety properties\n   - `init` process that starts all proctypes, waits for completion, signals shutdown, runs final invariant checks\n   - Aim for 100% state coverage (no unreached states)\n   - Use `atomic{}` for operations that share a mutex in the real code\n   - Use `d_step{}` for deterministic sequences (scans)\n   - Use generation counters to handle key reuse across lifecycles\n   - Model channels as bounded `chan` (even if real code uses unbounded)\n\n   **Scope-specific conventions:**\n\n   For **single component** models:\n   - Abstract external interfaces as nondeterministic stubs (e.g., `if :: success :: failure fi`)\n   - Focus on internal thread interactions and shared state\n   - Name the header comment \"Component-level model: \u003Ccomponent-name>\"\n   - Document which interfaces are stubbed and what assumptions are made\n\n   For **whole system** models:\n   - One `proctype` per major thread role across components\n   - Use channels to model inter-component communication (gRPC, receptacles)\n   - Name the header comment \"System-level model: \u003Cproperty-name>\"\n   - Document the component wiring in the header\n\n5. **Write the Makefile** `modelling\u002Fspin\u002F\u003Cname>\u002FMakefile`:\n\n   ```makefile\n   MODEL = \u003Cname_underscored>.pml\n   PAN   = pan\n   CC    = cc\n   CFLAGS = -O2\n   DEPTH = 200000\n\n   .PHONY: all safety liveness clean\n\n   all: safety\n\n   safety: $(PAN)\n   \t.\u002F$(PAN) -m$(DEPTH)\n\n   liveness: $(PAN)-live\n   \t.\u002F$(PAN)-live -a -m$(DEPTH)\n\n   $(PAN): pan.c\n   \t$(CC) $(CFLAGS) -DSAFETY -o $@ $\u003C\n\n   $(PAN)-live: pan.c\n   \t$(CC) $(CFLAGS) -o $@ $\u003C\n\n   pan.c: $(MODEL)\n   \tspin -a $\u003C\n\n   clean:\n   \trm -f $(PAN) $(PAN)-live pan.* *.trail\n   ```\n\n6. **Write the README.md** `modelling\u002Fspin\u002F\u003Cname>\u002FREADME.md`:\n\n   Structure (follow the pattern in `modelling\u002Fspin\u002Fwrite-before-evict\u002FREADME.md`):\n   - Title: `# \u003CProperty Name>`\n   - **Scope**: Single component (`\u003Ccomponent-name>`) or Whole system\n   - One-paragraph description of what the model verifies\n   - **Properties Verified** table (ID, Property description, Type: Safety\u002FLiveness)\n   - **System Abstraction** table (Real component → Promela process)\n   - **Assumptions \u002F Stubs** section (for single-component scope: list what is abstracted away and how)\n   - **Running** section with shell commands\n   - **Tuning the Model** section explaining parameters\n   - **Correspondence to Source Code** table (Model location → Source file → Line range)\n\n7. **Run the verification** by executing `make` in the new directory.\n\n   - If Spin is not installed, inform the user and skip this step.\n   - If verification fails with an assertion violation, analyze the trail (`spin -t -p \u003Cfile>.pml`), fix the model, and re-run until it passes.\n   - Report: number of states explored, depth reached, errors found, coverage (unreached states).\n\n8. **Report results** to the user:\n   - Scope (single component or whole system)\n   - Property being verified\n   - Pass\u002Ffail status\n   - State space size\n   - Coverage (target: 0 unreached states in all proctypes)\n   - Path to the new model directory\n\n## Notes\n\n- Spin must be installed at `\u002Fusr\u002Flocal\u002Fbin\u002Fspin` (see `modelling\u002Fspin\u002Fwrite-before-evict\u002FREADME.md` for install instructions).\n- Keep parameters small (2-3 clients, 2-4 keys, pool cap \u003C key count) to maintain tractable state spaces (\u003C10M states).\n- The model should be self-contained — no external dependencies beyond Spin and a C compiler.\n- Use the generation-counter technique from the write-before-evict model to handle key reuse without conflating lifecycles.\n- Single-component models are faster to verify (fewer processes) and better for finding internal races.\n- Whole-system models catch cross-component protocol violations but have larger state spaces.\n",{"name":5,"description":7,"argument-hint":35},"[property-number or description of property to prove]","unstable","dc564aa356769b7754391c9fa4fdaa0f179f476c","components\u002Fblock-device-kernel\u002F.claude\u002Fskills\u002Ftools-spin-model",4,"https:\u002F\u002Fgithub.com\u002FAI-native-Systems-Research\u002Fai-native-storage-certus","https:\u002F\u002Fgithub.com\u002FAI-native-Systems-Research","https:\u002F\u002Fai-native-systems-research.github.io\u002Fai-native-systems-research\u002F","2026-11-12 06:08:32",{"source":8,"skillName":5,"commands":45,"prompt":55,"zipAvailable":56},[46,49,52],{"tool":47,"command":48},"npx","npx skills add AI-native-Systems-Research\u002Fai-native-storage-certus --skill tools-spin-model",{"tool":50,"command":51},"bunx","bunx skills add AI-native-Systems-Research\u002Fai-native-storage-certus --skill tools-spin-model",{"tool":53,"command":54},"pnpm","pnpm dlx skills add AI-native-Systems-Research\u002Fai-native-storage-certus --skill tools-spin-model","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FAI-native-Systems-Research\u002Fai-native-storage-certus\u002Fblob\u002Funstable\u002Fcomponents\u002Fblock-device-kernel\u002F.claude\u002Fskills\u002Ftools-spin-model\u002FSKILL.md",true,[],[59,76,81,92,109,121],{"skillNo":60,"name":61,"nameZh":6,"description":62,"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":30},"01M6KP29TWSNGADF3AJ47BR7TY","tools-spin-sync","Synchronize Spin\u002FPromela models with current source code and re-verify","https:\u002F\u002Fgithub.com\u002FAI-native-Systems-Research\u002Fai-native-storage-certus\u002Fblob\u002Funstable\u002Fcomponents\u002Fzyre\u002F.claude\u002Fskills\u002Ftools-spin-sync\u002FSKILL.md",[],[66,19,67,68,69,20,70,71,72,73,74,75],"tools spin sync","sync","code","verify","models","name","with","re","source","source code",{"skillNo":77,"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":78,"categories":79,"keywords":80,"publishedAt":30},"01M6KP277TDY3ZAJEBP5P6MDT8","https:\u002F\u002Fgithub.com\u002FAI-native-Systems-Research\u002Fai-native-storage-certus\u002Fblob\u002Funstable\u002Fcomponents\u002Fzyre\u002F.claude\u002Fskills\u002Ftools-spin-model\u002FSKILL.md",[],[21,20,19,24,25,26,27,28,29,22,23],{"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":30},"01M6KP250W406CEDSF02BRXT9D","tools-creusot-install","Install the Creusot Rust verification tool","https:\u002F\u002Fgithub.com\u002FAI-native-Systems-Research\u002Fai-native-storage-certus\u002Fblob\u002Funstable\u002Fcomponents\u002Fzyre\u002F.claude\u002Fskills\u002Ftools-creusot-install\u002FSKILL.md",[],[88,89,19,90,23,91],"tools creusot install","install","rust","tool",{"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":30},"01M6KP22H28FQ1X0V7VJHEFWXD","tools-creusot-init","Prepare an existing Cargo project to use Creusot formal verification","https:\u002F\u002Fgithub.com\u002FAI-native-Systems-Research\u002Fai-native-storage-certus\u002Fblob\u002Funstable\u002Fcomponents\u002Fzyre\u002F.claude\u002Fskills\u002Ftools-creusot-init\u002FSKILL.md",[],[99,19,100,101,102,103,104,23,105,25,106,107,108],"tools creusot init","init","project","directory","components","prepare","use","cargo","manager","Existing",{"skillNo":110,"name":111,"nameZh":6,"description":112,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":113,"categories":114,"keywords":115,"publishedAt":30},"01M6KP1V506J9YM8DKKVDG1NRE","tools-creusot-coq-install","Add Coq 8.20 as an additional prover to an existing Creusot installation, for verification conditions that SMT solvers cannot discharge","https:\u002F\u002Fgithub.com\u002FAI-native-Systems-Research\u002Fai-native-storage-certus\u002Fblob\u002Funstable\u002Fcomponents\u002Fzyre\u002F.claude\u002Fskills\u002Ftools-creusot-coq-install\u002FSKILL.md",[],[116,117,89,19,27,108,118,119,23,120],"tools creusot coq install","Coq","as","installation","add",{"skillNo":122,"name":123,"nameZh":6,"description":124,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":125,"categories":126,"keywords":127,"publishedAt":30},"01M6KP1PQJF3H765TYZAZ7P3EF","tools-creusot-annotate-interfaces","Formally verify a component's properties with Creusot and annotate its interface definitions","https:\u002F\u002Fgithub.com\u002FAI-native-Systems-Research\u002Fai-native-storage-certus\u002Fblob\u002Funstable\u002Fcomponents\u002Fzyre\u002F.claude\u002Fskills\u002Ftools-creusot-annotate-interfaces\u002FSKILL.md",[],[19,128,129,72,71,130,69,131,132],"tools creusot annotate interfaces","annotate","properties","component","interface"]