{ "schemaVersion": "1.0", "item": { "slug": "coding-lead", "name": "Coding Lead", "source": "tencent", "type": "skill", "category": "开发工具", "sourceUrl": "https://clawhub.ai/beyound87/coding-lead", "canonicalUrl": "https://clawhub.ai/beyound87/coding-lead", "targetPlatform": "OpenClaw" }, "install": { "downloadMode": "redirect", "downloadUrl": "/downloads/coding-lead", "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=coding-lead", "sourcePlatform": "tencent", "targetPlatform": "OpenClaw", "installMethod": "Manual import", "extraction": "Extract archive", "prerequisites": [ "OpenClaw" ], "packageFormat": "ZIP package", "includedAssets": [ "references/complex-tasks.md", "references/prompt-templates.md", "SKILL.md" ], "primaryDoc": "SKILL.md", "quickSetup": [ "Download the package from Yavira.", "Extract the archive and review SKILL.md first.", "Import or place the package into your OpenClaw setup." ], "agentAssist": { "summary": "Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.", "steps": [ "Download the package from Yavira.", "Extract it into a folder your agent can access.", "Paste one of the prompts below and point your agent at the extracted folder." ], "prompts": [ { "label": "New install", "body": "I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Tell me what you changed and call out any manual steps you could not complete." }, { "label": "Upgrade existing", "body": "I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Summarize what changed and any follow-up checks I should run." } ] }, "sourceHealth": { "source": "tencent", "status": "healthy", "reason": "direct_download_ok", "recommendedAction": "download", "checkedAt": "2026-04-30T16:55:25.780Z", "expiresAt": "2026-05-07T16:55:25.780Z", "httpStatus": 200, "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=network", "contentType": "application/zip", "probeMethod": "head", "details": { "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=network", "contentDisposition": "attachment; filename=\"network-1.0.0.zip\"", "redirectLocation": null, "bodySnippet": null }, "scope": "source", "summary": "Source download looks usable.", "detail": "Yavira can redirect you to the upstream package for this source.", "primaryActionLabel": "Download for OpenClaw", "primaryActionHref": "/downloads/coding-lead" }, "validation": { "installChecklist": [ "Use the Yavira download entry.", "Review SKILL.md after the package is downloaded.", "Confirm the extracted package contains the expected setup assets." ], "postInstallChecks": [ "Confirm the extracted package includes the expected docs or setup files.", "Validate the skill or prompts are available in your target agent workspace.", "Capture any manual follow-up steps the agent could not complete." ] }, "downloadPageUrl": "https://openagent3.xyz/downloads/coding-lead", "agentPageUrl": "https://openagent3.xyz/skills/coding-lead/agent", "manifestUrl": "https://openagent3.xyz/skills/coding-lead/agent.json", "briefUrl": "https://openagent3.xyz/skills/coding-lead/agent.md" }, "agentAssist": { "summary": "Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.", "steps": [ "Download the package from Yavira.", "Extract it into a folder your agent can access.", "Paste one of the prompts below and point your agent at the extracted folder." ], "prompts": [ { "label": "New install", "body": "I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Tell me what you changed and call out any manual steps you could not complete." }, { "label": "Upgrade existing", "body": "I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Summarize what changed and any follow-up checks I should run." } ] }, "documentation": { "source": "clawhub", "primaryDoc": "SKILL.md", "sections": [ { "title": "Coding Lead", "body": "This skill supersedes inline coding rules in agent SOUL.md files.\n\nRoute by complexity. ACP fails → auto-fallback to direct execution." }, { "title": "Task Classification", "body": "LevelCriteriaActionSimpleSingle file, <60 linesDirect: read/write/edit/execMedium2-5 files, clear scopeACP → fallback directComplexArchitecture, multi-modulePlan → ACP → fallback chunked direct\n\nWhen in doubt, go one level up." }, { "title": "Tech Stack (New Projects)", "body": "LayerPreferredFallbackBackendPHP (Laravel/ThinkPHP)PythonFrontendVue.jsReactMobileFlutterUniApp-XCSSTailwind-DBMySQLPostgreSQL\n\nExisting projects: follow current stack. New: propose first, wait for confirmation." }, { "title": "Tool Detection & Fallback", "body": "All tools are optional. Detect once per session:\n\nToolAvailable?Fallbacksmart-agent-memorynode ~/.openclaw/skills/smart-agent-memory/scripts/memory-cli.js stats ok?memory_search + manual .md writesqmdqmd --version ok?grep (Linux/macOS) / Select-String (Windows) / findACPsessions_spawn succeeds?Direct read/write/edit/exec\n\nNotation: [memory] [qmd] [acp] = use if available, fallback if not." }, { "title": "ACP Agent Routing", "body": "ACP access: sessions_spawn(runtime=\"acp\") may not be available (platform limitation). Use acpx CLI as the reliable path:\n\n# Detect paths dynamically (run once per session)\n# Windows:\n# ACPX = %APPDATA%\\npm\\node_modules\\openclaw\\extensions\\acpx\\node_modules\\.bin\\acpx.cmd\n# AGENT = node %APPDATA%/npm/node_modules/@zed-industries/claude-agent-acp/dist/index.js\n# macOS/Linux:\n# ACPX = $(npm root -g)/openclaw/extensions/acpx/node_modules/.bin/acpx\n# AGENT = node $(npm root -g)/@zed-industries/claude-agent-acp/dist/index.js\n\n# One-shot execution via acpx (works from any session type)\nacpx --verbose --agent \"\" exec \"\" 2>&1\n\nIf sessions_spawn(runtime=\"acp\") is available in your session, use it. Otherwise fall back to acpx CLI via exec.\n\nCheck acp.allowedAgents in openclaw.json for available agents:\n\nTask TypeBest AgentWhyComplex backend, multi-file refactor, deep reasoningclaudeSuperior cross-file reasoning, long contextQuick iteration, autonomous exploration, sandbox taskscodexFast, autonomous, good at iterative fixesCode reviewDifferent agent than the one that wrote itAvoid same-bias blind spots" }, { "title": "How to route", "body": "# Default (uses acp.defaultAgent, typically \"claude\"):\nsessions_spawn(runtime: \"acp\", task: , cwd: )\n\n# Explicit agent:\nsessions_spawn(runtime: \"acp\", agentId: \"claude\", task: , cwd: )\nsessions_spawn(runtime: \"acp\", agentId: \"codex\", task: , cwd: )" }, { "title": "Fallback chain", "body": "Try preferred agent → 2. Try alternate agent → 3. Direct execution\n\nIf one agent fails/unavailable, try the other before falling back to direct execution." }, { "title": "Parallel with different agents", "body": "For complex tasks with independent sub-tasks (max 2 parallel):\n\nSession 1: claude → backend refactor (needs deep reasoning)\nSession 2: codex → frontend fixes (needs fast iteration)" }, { "title": "Layer 1: Project-level (Claude Code owns)", "body": "Projects may have their own CLAUDE.md, .cursorrules, docs/ — these are Claude Code's responsibility. It reads them automatically. Do NOT paste project-level rules into ACP prompts." }, { "title": "Layer 2: Team-level (OpenClaw owns)", "body": "shared/knowledge/tech-standards.md — cross-project standards (security, change control, tech stack preferences). Only relevant for direct execution (simple tasks without ACP)." }, { "title": "When spawning ACP", "body": "Don't embed coding standards in the prompt — Claude Code has its own CLAUDE.md\nDo include: task description, acceptance criteria, relevant context (file paths, decisions)\nDo include task-specific constraints if any (e.g., \"don't change the API contract\")" }, { "title": "When executing directly (no ACP)", "body": "Load standards once per session, first match wins:\n\nshared/knowledge/tech-standards.md (team-level, if exists)\nBuilt-in defaults (below, if nothing exists)" }, { "title": "Built-in Defaults (fallback for direct execution)", "body": "KISS + SOLID + DRY, research before modifying\nMethods <200 lines, follow existing architecture\nNo hardcoded secrets, minimal change scope, clear commits\nDB changes via SQL scripts, new tech requires confirmation" }, { "title": "Simple Tasks", "body": "Read target file(s) (standards already loaded per above)\n[memory] Recall related decisions\nExecute with read/write/edit/exec\n[memory] Record what changed and why" }, { "title": "Step 1: Build Context File", "body": "Write to /.openclaw/context-.md (ACP reads from disk, not from prompt):\n\n# [qmd] or grep: find relevant code\n# [memory] recall + lessons: find past decisions\n# Standards already loaded (see \"Coding Standards Loading\" above)\n# Write context file with 3-5 key rules from loaded standards — do NOT paste full file\n\nMinimal context file structure:\n\n# Task Context: \n## Project — path, stack, architecture style\n## Relevant Code — file paths + brief descriptions from qmd/grep\n## History — past decisions/lessons from memory (if any)\n## Constraints — task-specific rules only (NOT general coding standards — Claude Code has CLAUDE.md)\n\nFull template with examples → see references/prompt-templates.md" }, { "title": "Step 2: Lean Prompt", "body": "Project: | Stack: \nContext file: .openclaw/context-.md (read it first if it exists)\n\n## Task\n\n\n## Acceptance Criteria\n- [ ] \n- [ ] Tests pass, no unrelated changes, clean code\n\nBefore finishing: run linter + tests, include results.\nWhen done: openclaw system event --text \"Done: \" --mode now" }, { "title": "Step 3: Spawn", "body": "# Option A: sessions_spawn (if available in your session)\nsessions_spawn(runtime: \"acp\", task: , cwd: , mode: \"run\")\n\n# Option B: acpx CLI (always works, detect AGENT path per above)\nexec: acpx --agent \"\" exec \"\"\n# Set cwd to project dir in exec command" }, { "title": "Step 4: Fallback Detection", "body": "ConditionActionSpawn failed / timeout→ Direct executionEmpty output / no file changes→ Direct executionPartial completion→ Agent fixes remaining\n\nFallback: [memory] log failure → agent executes directly → report to user.\n\nNever silently fail. Always complete or report why not." }, { "title": "Step 5: Verify & Record", "body": "Check acceptance criteria + run tests\n[memory] Record: what changed, decisions, lessons\nClean up context file" }, { "title": "Complex Tasks", "body": "Read references/complex-tasks.md only for Complex-level tasks — roles, QA isolation, parallel strategies, RESEARCH→PLAN→EXECUTE→REVIEW flow." }, { "title": "Context Reuse (Token Savings)", "body": "Context file on disk instead of prompt embedding → ~90% token savings per spawn\nParallel: one context file, multiple ACP sessions read it\nSerial: use mode: \"session\" + sessions_send for follow-ups\n[qmd]: precision search → only relevant snippets in context file\nNo standards in ACP prompts: Claude Code reads its own CLAUDE.md/.cursorrules — don't duplicate\nACP prompt stays lean: task + acceptance criteria + context file path. No generic rules\nDirect execution: load team standards once per session, not per task" }, { "title": "Memory Integration", "body": "[memory] Before: recall related work + lessons for context file.\n[memory] After: record what changed, decisions made, lessons learned.\nCross-session: agent remembers across sessions; Claude Code doesn't. This is the core advantage." }, { "title": "Multi-Project Parallel", "body": "Each project gets its own context file in its own .openclaw/ dir\nSpawn with different cwd per project — zero cross-contamination\nTag memory entries per project: --tags code,\nMax 2 parallel ACP sessions — keep token/resource use predictable\nACP runs in background while agent works on simple tasks directly\n\nSee references/prompt-templates.md for multi-project examples." }, { "title": "Smart Retry (max 3)", "body": "Analyze failure → 2. Adjust prompt → 3. Retry improved → 4. Max 3 then fallback/report.\nEach retry must be meaningfully different." }, { "title": "Progress Updates", "body": "Start → 1 short message. Error → immediate report. Completion → summary. Fallback → explain." }, { "title": "Safety", "body": "Never spawn in ~/.openclaw/ — coding agents may damage config\nAlways set cwd to project directory\nReview before commit — especially complex tasks\nKill runaway sessions — timeout or nonsensical output" }, { "title": "See Also", "body": "references/complex-tasks.md — roles, QA, parallel (Complex only)\nreferences/prompt-templates.md — context file template, prompt examples" } ], "body": "Coding Lead\n\nThis skill supersedes inline coding rules in agent SOUL.md files.\n\nRoute by complexity. ACP fails → auto-fallback to direct execution.\n\nTask Classification\nLevel\tCriteria\tAction\nSimple\tSingle file, <60 lines\tDirect: read/write/edit/exec\nMedium\t2-5 files, clear scope\tACP → fallback direct\nComplex\tArchitecture, multi-module\tPlan → ACP → fallback chunked direct\n\nWhen in doubt, go one level up.\n\nTech Stack (New Projects)\nLayer\tPreferred\tFallback\nBackend\tPHP (Laravel/ThinkPHP)\tPython\nFrontend\tVue.js\tReact\nMobile\tFlutter\tUniApp-X\nCSS\tTailwind\t-\nDB\tMySQL\tPostgreSQL\n\nExisting projects: follow current stack. New: propose first, wait for confirmation.\n\nTool Detection & Fallback\n\nAll tools are optional. Detect once per session:\n\nTool\tAvailable?\tFallback\nsmart-agent-memory\tnode ~/.openclaw/skills/smart-agent-memory/scripts/memory-cli.js stats ok?\tmemory_search + manual .md writes\nqmd\tqmd --version ok?\tgrep (Linux/macOS) / Select-String (Windows) / find\nACP\tsessions_spawn succeeds?\tDirect read/write/edit/exec\n\nNotation: [memory] [qmd] [acp] = use if available, fallback if not.\n\nACP Agent Routing\n\nACP access: sessions_spawn(runtime=\"acp\") may not be available (platform limitation). Use acpx CLI as the reliable path:\n\n# Detect paths dynamically (run once per session)\n# Windows:\n# ACPX = %APPDATA%\\npm\\node_modules\\openclaw\\extensions\\acpx\\node_modules\\.bin\\acpx.cmd\n# AGENT = node %APPDATA%/npm/node_modules/@zed-industries/claude-agent-acp/dist/index.js\n# macOS/Linux:\n# ACPX = $(npm root -g)/openclaw/extensions/acpx/node_modules/.bin/acpx\n# AGENT = node $(npm root -g)/@zed-industries/claude-agent-acp/dist/index.js\n\n# One-shot execution via acpx (works from any session type)\nacpx --verbose --agent \"\" exec \"\" 2>&1\n\n\nIf sessions_spawn(runtime=\"acp\") is available in your session, use it. Otherwise fall back to acpx CLI via exec.\n\nCheck acp.allowedAgents in openclaw.json for available agents:\n\nTask Type\tBest Agent\tWhy\nComplex backend, multi-file refactor, deep reasoning\tclaude\tSuperior cross-file reasoning, long context\nQuick iteration, autonomous exploration, sandbox tasks\tcodex\tFast, autonomous, good at iterative fixes\nCode review\tDifferent agent than the one that wrote it\tAvoid same-bias blind spots\nHow to route\n# Default (uses acp.defaultAgent, typically \"claude\"):\nsessions_spawn(runtime: \"acp\", task: , cwd: )\n\n# Explicit agent:\nsessions_spawn(runtime: \"acp\", agentId: \"claude\", task: , cwd: )\nsessions_spawn(runtime: \"acp\", agentId: \"codex\", task: , cwd: )\n\nFallback chain\nTry preferred agent → 2. Try alternate agent → 3. Direct execution\n\nIf one agent fails/unavailable, try the other before falling back to direct execution.\n\nParallel with different agents\n\nFor complex tasks with independent sub-tasks (max 2 parallel):\n\nSession 1: claude → backend refactor (needs deep reasoning)\nSession 2: codex → frontend fixes (needs fast iteration)\n\nCoding Standards — Two Layers, No Overlap\nLayer 1: Project-level (Claude Code owns)\n\nProjects may have their own CLAUDE.md, .cursorrules, docs/ — these are Claude Code's responsibility. It reads them automatically. Do NOT paste project-level rules into ACP prompts.\n\nLayer 2: Team-level (OpenClaw owns)\n\nshared/knowledge/tech-standards.md — cross-project standards (security, change control, tech stack preferences). Only relevant for direct execution (simple tasks without ACP).\n\nWhen spawning ACP\nDon't embed coding standards in the prompt — Claude Code has its own CLAUDE.md\nDo include: task description, acceptance criteria, relevant context (file paths, decisions)\nDo include task-specific constraints if any (e.g., \"don't change the API contract\")\nWhen executing directly (no ACP)\n\nLoad standards once per session, first match wins:\n\nshared/knowledge/tech-standards.md (team-level, if exists)\nBuilt-in defaults (below, if nothing exists)\nBuilt-in Defaults (fallback for direct execution)\nKISS + SOLID + DRY, research before modifying\nMethods <200 lines, follow existing architecture\nNo hardcoded secrets, minimal change scope, clear commits\nDB changes via SQL scripts, new tech requires confirmation\nSimple Tasks\nRead target file(s) (standards already loaded per above)\n[memory] Recall related decisions\nExecute with read/write/edit/exec\n[memory] Record what changed and why\nMedium/Complex Tasks\nStep 1: Build Context File\n\nWrite to /.openclaw/context-.md (ACP reads from disk, not from prompt):\n\n# [qmd] or grep: find relevant code\n# [memory] recall + lessons: find past decisions\n# Standards already loaded (see \"Coding Standards Loading\" above)\n# Write context file with 3-5 key rules from loaded standards — do NOT paste full file\n\n\nMinimal context file structure:\n\n# Task Context: \n## Project — path, stack, architecture style\n## Relevant Code — file paths + brief descriptions from qmd/grep\n## History — past decisions/lessons from memory (if any)\n## Constraints — task-specific rules only (NOT general coding standards — Claude Code has CLAUDE.md)\n\n\nFull template with examples → see references/prompt-templates.md\n\nStep 2: Lean Prompt\nProject: | Stack: \nContext file: .openclaw/context-.md (read it first if it exists)\n\n## Task\n\n\n## Acceptance Criteria\n- [ ] \n- [ ] Tests pass, no unrelated changes, clean code\n\nBefore finishing: run linter + tests, include results.\nWhen done: openclaw system event --text \"Done: \" --mode now\n\nStep 3: Spawn\n# Option A: sessions_spawn (if available in your session)\nsessions_spawn(runtime: \"acp\", task: , cwd: , mode: \"run\")\n\n# Option B: acpx CLI (always works, detect AGENT path per above)\nexec: acpx --agent \"\" exec \"\"\n# Set cwd to project dir in exec command\n\nStep 4: Fallback Detection\nCondition\tAction\nSpawn failed / timeout\t→ Direct execution\nEmpty output / no file changes\t→ Direct execution\nPartial completion\t→ Agent fixes remaining\n\nFallback: [memory] log failure → agent executes directly → report to user.\n\nNever silently fail. Always complete or report why not.\n\nStep 5: Verify & Record\nCheck acceptance criteria + run tests\n[memory] Record: what changed, decisions, lessons\nClean up context file\nComplex Tasks\n\nRead references/complex-tasks.md only for Complex-level tasks — roles, QA isolation, parallel strategies, RESEARCH→PLAN→EXECUTE→REVIEW flow.\n\nContext Reuse (Token Savings)\nContext file on disk instead of prompt embedding → ~90% token savings per spawn\nParallel: one context file, multiple ACP sessions read it\nSerial: use mode: \"session\" + sessions_send for follow-ups\n[qmd]: precision search → only relevant snippets in context file\nNo standards in ACP prompts: Claude Code reads its own CLAUDE.md/.cursorrules — don't duplicate\nACP prompt stays lean: task + acceptance criteria + context file path. No generic rules\nDirect execution: load team standards once per session, not per task\nMemory Integration\n\n[memory] Before: recall related work + lessons for context file. [memory] After: record what changed, decisions made, lessons learned. Cross-session: agent remembers across sessions; Claude Code doesn't. This is the core advantage.\n\nMulti-Project Parallel\nEach project gets its own context file in its own .openclaw/ dir\nSpawn with different cwd per project — zero cross-contamination\nTag memory entries per project: --tags code,\nMax 2 parallel ACP sessions — keep token/resource use predictable\nACP runs in background while agent works on simple tasks directly\n\nSee references/prompt-templates.md for multi-project examples.\n\nSmart Retry (max 3)\nAnalyze failure → 2. Adjust prompt → 3. Retry improved → 4. Max 3 then fallback/report. Each retry must be meaningfully different.\nProgress Updates\n\nStart → 1 short message. Error → immediate report. Completion → summary. Fallback → explain.\n\nSafety\nNever spawn in ~/.openclaw/ — coding agents may damage config\nAlways set cwd to project directory\nReview before commit — especially complex tasks\nKill runaway sessions — timeout or nonsensical output\nSee Also\nreferences/complex-tasks.md — roles, QA, parallel (Complex only)\nreferences/prompt-templates.md — context file template, prompt examples" }, "trust": { "sourceLabel": "tencent", "provenanceUrl": "https://clawhub.ai/beyound87/coding-lead", "publisherUrl": "https://clawhub.ai/beyound87/coding-lead", "owner": "beyound87", "version": "1.0.30", "license": null, "verificationStatus": "Indexed source record" }, "links": { "detailUrl": "https://openagent3.xyz/skills/coding-lead", "downloadUrl": "https://openagent3.xyz/downloads/coding-lead", "agentUrl": "https://openagent3.xyz/skills/coding-lead/agent", "manifestUrl": "https://openagent3.xyz/skills/coding-lead/agent.json", "briefUrl": "https://openagent3.xyz/skills/coding-lead/agent.md" } }